OXPHOS remodeling in high-grade prostate cancer involves mtDNA mutations and increased succinate oxidation

Abstract

Rewiring of energy metabolism and adaptation of mitochondria are considered to impact on prostate cancer development and progression. Here, we report on mitochondrial respiration, DNA mutations and gene expression in paired benign/malignant human prostate tissue samples. Results reveal reduced respiratory capacities with NADH-pathway substrates glutamate and malate in malignant tissue and a significant metabolic shift towards higher succinate oxidation, particularly in high-grade tumors. The load of potentially deleterious mitochondrial-DNA mutations is higher in tumors and associated with unfavorable risk factors. High levels of potentially deleterious mutations in mitochondrial Complex I-encoding genes are associated with a 70% reduction in NADH-pathway capacity and compensation by increased succinate-pathway capacity. Structural analyses of these mutations reveal amino acid alterations leading to potentially deleterious effects on Complex I, supporting a causal relationship. A metagene signature extracted from the transcriptome of tumor samples exhibiting a severe mitochondrial phenotype enables identification of tumors with shorter survival times.

Fig. 1. Sample workflow and tissue sample confirmation.

a From each of 50 radical prostatectomy specimens a tumor and a non-malignant benign tissue punch needle biopsy was extracted by an experienced uropathologist from contralateral sites of the specimens (blue/red circles). While a small portion of the extracted tissue cores was fixed and used for histological stains (pink arrow) and confirmation of tissue identity, the rest was used immediately for high-resolution respirometry (HRR, blue arrow), subsequent NGS mtDNA profiling (orange arrow) and mtDNA copy number determination (gray arrow). From 16 cases of this cohort, frozen tissue samples directly adjacent to the biopsy cores (green area) were isolated by macrodissection followed by RNA extraction for gene expression profiling via total RNA-NGS (green arrow). b Representative hematoxylin and eosin staining (H&E) and p63 (non-malignant tissue marker, brown)/AMACR (malignant cell marker, red) double-immunostaining (p63/AMACR) of fixed and paraffin-embedded benign and malignant prostate tissue samples extracted for HRR. One representative of 50 cases is shown. Scale bars indicate 2000 µm (H&E, P63/AMACRA stains) and 1000 µm (P63/AMACR higher magnification), respectively.

Fig. 2. High-resolution respirometry of prostate tissue samples.

a Coupling/pathway control diagram showing the sequential steps in the substrate-uncoupler-inhibitor titration (SUIT) protocol with different coupling states. See Supplementary Tables 1–2 for HRR protocol details. b Representative HRR traces with permeabilized tissue. Red line (left Y-axis): wet mass-specific O₂ flux (oxygen consumption (pmol s^­1 mg^­−1)). Blue line (right Y-axis): O₂ concentration [µM]. Substrate-uncoupler-inhibitor titrations are indicated by arrows. Different coupling/pathway control states are indicated in boxes: LEAK (orange); OXPHOS (green); ET (blue); ROX (black). c Schematic representation of mitochondrial electron transfer from NADH-linked substrates through Complexes CI, CIII, and CIV (N-pathway) and from succinate through CII, CIII, and CIV (S-pathway). G, M and P support the N-pathway through CI into the Q-cycle (Q); succinate provides electrons via CII into the Q junction. Electrons are transferred from CIII via cytochrome c (c) to CIV where O₂ is reduced to H₂O. H⁺ ions are pumped across the mt-inner membrane by CI, CIII, and CIV to generate an electrochemical potential difference across the mt-inner membrane, which drives phosphorylation of ADP to ATP by F_OF₁-ATPase. d Respiratory capacity in benign (blue, N = 50) versus malignant (red, N = 50) tissue samples: OXPHOS-capacity (GM_P, N_P and NS_P) and ET-capacity (NS_E and S_E). e Effects of substrates GM, pyruvate, succinate, oxidative stress, uncoupler FCCP, and CI inhibitor rotenone on O₂ flux in benign (blue, N = 50) and malignant (red, N = 50) tissue samples. f. Normalized respiratory capacities of high-grade tumor (Gleason > 7; dark red; N = 10) and low-grade tumor (Gleason ≤ 7, light red, N = 40) compared to benign samples (blue, N = 50). g Effects of substrates, oxidative stress, uncoupler, and CI inhibitor on O₂ flux in low-grade (light red, N = 40) and high-grade (dark red, N = 10) tissue samples. Data in (d-g) are presented as mean values ± SD. Statistical differences were tested using two-tailed paired-samples test (d–e), one-way ANOVA followed by Tukey’s HSD (f) or Wilcoxon rank-sum test (g), respectively. Correction for multiple testing was performed using the Bonferroni-Holm procedure. Source data are provided as a Source Data file.

Fig. 3. mtDNA heteroplasmies in benign and malignant tissue samples.

a Levels and locations of HPs in benign (blue, N = 50) and malignant (red, N = 50) samples across indicated loci of the mtDNA. The regions shown include the non-coding control region (CR) and the 13 protein coding genes, marked by colored boxes. ND, CO and ATP refer to genes coding for subunits of Complex I (CI; NADH:ubiquinone oxidoreductase), Complex IV (CIV; ferrocytochrome c:oxygen oxidoreductase and ATP synthase, respectively, whereas CYB encodes a subunit of Complex III (CIII; ubiquinol:ferricytochrome c oxidoreductase). b Total cumulative count of private mutations, located in either the non-coding D-loop or coding areas of the mt-genome in benign (blue, N = 50) and malignant (red, N = 50) tissue, respectively. Differences were tested for significance using Fisher’s exact test. c Incidence of synonymous (Syn) vs non-synonymous (Non-syn) HPs located in protein coding sequences (F_OF₁-ATP synthase=CV; tRNA genes=tRNA) found in either benign (blue, N = 50) or malignant (red, N = 50) tissue samples. d Detailed proportions of samples harboring variants with defined HP levels (<10%, 10–20%, 20–50% and >50%) in the benign (blue, N = 50) and malignant (red, N = 50) samples. e Boxplot of HP levels comprising all heteroplasmies in benign and malignant tissue samples. Data are presented as boxplots indicating median, 25–75th percentile (box) and median ± 1.5IQR (whiskers), minimum and maximum values (dots). Differences in mean values were tested for significance using Wilcoxon rank-sum test. f Numbers of HPs found per sample in benign (blue circle, N = 50) and the malignant (red circle, N = 50) samples. g Correlation of gene size in kbp to HP count as detected in the mt-ND genes of malignant samples (N = 50). Linear correlation was established using Pearson’s correlation coefficient (r) while correlation was tested for significance using two-tailed t-test. h MutPred Pathocenicity Scores of all non-synonymous HPs in benign (blue, N = 50) and malignant samples (red, N = 50) identified across the mtDNA. The size of the spots indicates the likely functional effects. i Proportion of benign (blue, N = 50) and malignant (red, N = 50) samples carrying HPs with defined MutPred Pathogenicity scores reflecting their likely functional effect. Source data are provided as a Source Data file.

Fig. 4. mtDNA heteroplasmies and respiratory capacities.

a Relative GM-OXPHOS respiratory capacity with glutamate&malate (GM_P/NS_E) in benign (blue) or malignant (red) samples carrying either no or only synonymous HPs (−) (N_BE = 36 and N_CA = 23) versus samples carrying non-synonymous HPs (+) (N_BE = 14 and N_CA = 27) in the coding regions of the mt-genome. Values present mean ± SD. Differences in mean values were tested for significance using Wilcoxon rank-sum test. b Comparison of relative GM-OXPHOS capacity in samples without mutations (−) (N_BE = 38 and N_CA = 37) versus samples with mutations (+) (N_BE = 12 and N_CA = 13) within the non-coding (control) region of the mt-genome. Values represent mean ± SD. c Impact of location of non-synonymous HPs on relative GM-OXPHOS capacities. Tumor tissue samples were categorized according to no HPs (−) (orange; N = 24), and HPs located in genes encoding proteins of CIII–CV (CO, ATP, CYB; green; N = 10) or in genes encoding proteins of CI (ND1–ND5; blue; N = 20). d–e Relative GM-OXPHOS (GM_P/NS_E, d) and relative S-ET (S_E/NS_E, e) respiratory capacities in malignant tissue samples harboring non-synonymous HPs in CI-coding mt-genes. Tumors were grouped into samples carrying no non-synonymous HPs ((−); N = 23), samples with variant levels of 30–60% (30–60%; N = 6) and samples with variant levels >60% (>60%; N = 4). Differences in mean values were tested for significance using one-way ANOVA followed by Tukey’s HSD test. f N-pathway (blue) and S-pathway (orange) respiratory capacities in malignant samples harboring high-level (>60%) non-synonymous HPs in either CIV-coding genes (N = 4) or CI-coding genes (N = 4). Differences of in mean values were tested for significance using Wilcoxon rank-sum test. Values presented in c-f represent mean values and individual data points. g S-pathway OXPHOS capacity upregulation by partial inhibition of N-pathway oxidative flux in benign (RWPE1, N = 3; EP156T, N = 3) and malignant (PC3, N = 6; LNCaP, N = 4; DuCaP, N = 3, N represents number of biologically independent experiments) prostate cell lines. Relative S-pathway OXPHOS capacity (normalized to total respiratory capacity, NS_E) with different degrees of N-pathway inhibition is shown for the five cell lines. Values represent mean ± SD. Source data are provided as a Source Data file.

Fig. 5. Structural impact of non-synonymous CI gene mutations.

a List of malignant samples carrying non-synonymous variants with HP levels ≥30% (HP% = heteroplasmy levels, Gene=mt-gene, mt-pos=location of variant on mtDNA sequence, Ref= nucleotide in rCRS sequence, Mut=altered nucleotide found in sample, AA-pos=position of encoded amino acid in protein sequence, Ref=amino-acid in wild-type protein, Mut=altered amino acid). b Structure of human respiratory Complex I and location of mutations found in this study. mtDNA-encoded CI-subunits (ND1-6) are shown in colors (ND1 = dark red, ND2 = light blue, ND3 = orange, ND4 = red, ND4L = purple, ND5 = dark blue and ND6 = light blue), mutations are shown as black circles and Fe-S clusters as yellow spheres. To visualize mutation sites see complemental iSee package using any JavaScript enabled web browser ([https://github.com/genepi/mt-c1]). c Structural change caused by the F411S mutation (red circle) in ND4. Wild-type (F) and mutated variant (S) amino acids are depicted and superimposed together with amino-acid residues involved in the hydrophobic interaction network of the α-helix structure (W359, L360, and W416, respectively). d Structural change caused by the T387A mutation located within the loop of the discontinuing helix 12 in the central axis of CI membrane domain (blue circle) in ND5. Wild-type (T) and mutated variant (A) are superimposed and the stabilizing hydrogen bond present in the wild-type protein is depicted by a scattered line. e–f HRR traces of the malignant biopsies carrying the F411S (e) or the T387A mutation (f), respectively. The red line represents wet mass-specific O₂ flux (oxygen consumption (pmol s⁻¹ mg⁻¹]). g–h Respiratory capacities of malignant samples (red) carrying either the F411S mutation (g) or the T387A mutation (h), compared to the corresponding benign tissue (blue). Values represent mean ± SD of the two separate measurements for each tissue sample. See Fig. 2 and Supplementary Tables 1–2 for abbreviations, coupling states and SUIT protocol.

Fig. 6. mtDNA copy number and mt-mass load.

a Boxplot of mt-CN per cell in the benign (blue, N = 50) and malignant (red, N = 50) tissue samples. Data are presented as boxplots indicating median, 25th–75th percent percentile (box) and minimum and maximum values (whiskers). b Correlation of maximal NS-ET capacity (NS_E, y-axis) and mt-CN (x-axis) in benign (BE, blue circles, N = 50) and malignant tissue (CA, red circles, N = 50). Linear correlation was established using Pearson’s correlation coefficient (r) while correlation was tested for significance using two-tailed t-test. c Mean mt-CNs ± SD in malignant samples without ((−), light red, N = 17) or with ((+), dark red, N = 33) mtDNA mutations, respectively. Data are presented as boxplots indicating median, 25th-75th percent percentile (box) and minimum and maximum values (whiskers). Differences in mean values were tested for significance using Wilcoxon rank-sum test. d Comparison of mt-CN ratios calculated as mt-CN_CA/mt-CN_BE in tumor samples in low grade (Gleason score ≤7, light red, N = 40) vs high grade (Gleason score >7, dark red, N = 10) tumors. Differences in mean values were tested for significance using Wilcoxon rank-sum test. e–g Tissue immunostaining of mitochondrial markers for mt-mass (VDAC1, e), CI (NDUFS4, f) and CII (SDHA, g) in paired BE and CA tissues, N = 24. Quick scores are presented and differences in mean values were tested for significance using Wilcoxon rank-sum test. h CI/CII marker ratio, stratified according to mt-CI gene mutation load. No HPs in CI genes, (light blue), HP levels of 30-60% (blue) and >60% (dark blue) in CI genes. Data are presented as boxplot indicating median, 25th–75th percent percentile (box) and minimum and maximum values (whiskers). Differences in mean values were tested for significance using one-way ANOVA followed by Tukey’s HSD test. Source data are provided as a Source Data file.

Fig. 7. Transcriptome profile of mt-related genes.

Differential mRNA expression of mitochondrial genes and enriched metabolic KEGG pathways were analyzed in 16 paired tissue samples representative for the sample cohort. a Respiratory characteristics of tissue samples for which RNA-seq expression analysis was performed. The effects of the substrates glutamate&malate, pyruvate, succinate, of oxidative stress, uncoupler FCCP, and CI inhibitor rontenone on O₂ flux in benign (blue, N = 16) and malignant (red, N = 16) tissue samples are shown. Data are presented as mean values ± SD. Differences were tested for significance using Wilcoxon signed-rank test. b Heatmap and hierarchical clustering of all significantly differentially expressed mt-related genes (FDR < 0.01) based on the MitoCarta 2.0 Gene Catalog in the benign (blue) or malignant (red) tissue samples. c Enriched metabolic KEGG pathways upregulated (red) or downregulated (blue) in cancer tissue based on the InnateDB pathway overrepresentation analysis of all differentially expressed mt-related genes. d Boxplots representing Fragments Per Kilobase Million (FPKM) values for expression of significantly increased metabolic key-enzymes in the malignant (red, N = 16) vs the benign (blue, N = 16) tissue samples. Data are presented as boxplots with median, 25th–75th percent percentile (box), minimum and median ± 1.5 IQR (whiskers) and minimum and maximum values (dots). Differences were tested for significance using multiple t-test followed by Benjamini–Hochberg correction for multiple testing. e Detailed annotation of single upregulated TCA-cycle key-enzymes in the benign (blue arrows) and malignant (red arrows) samples. Blue boxes mark steps mainly involved in NADH-linked electron transfer, orange boxes highlight steps related to succinate-linked electron transfer, and green boxes mark steps involved in ROS detoxification. Substrates used during the respirometry experiments are indicated in bold type while blue or red colored names and arrows mark enzymes and enzymatic reactions that are higher in either benign or malignant samples, respectively. Full names of enzymes are given in the main text. Source data are provided as a Source Data file.

Fig. 8. Association of severe respiratory phenotype and survival.

a Glutamate&malate-linked OXPHOS capacity (GM_P) in the severe (orange) versus mild (blue) respiratory phenotype PCa tissue samples. Data are presented as mean values ± SD. Differences were tested for significance using Wilcoxon rank-sum test. b Relative proportion of synonymous (gray), non-synonymous CIII-IV (green) and non-synonymous CI (blue) mtDNA mutations detected in samples of the mild and severe respiratory phenotype tumors, respectively. c mt-CN ratios (CA/BE) as determined by duplex qPCR in the severe (orange, N = 8) versus mild (blue, N = 8) respiratory phenotype PCa samples. Mean and individual data points are presented and differences were tested for significance using Wilcoxon rank-sum test. d–e Gene expression profiles and survival information (OS, overall survival; DFS, disease-free survival; BCR, time to biochemical recurrence) were retrieved for five prostate cancer cohorts. The metagene set of 11 genes (Table 2) highly correlated to the gene expression profile of the severe respiratory phenotype PCa samples was used to dichotomize the PCa cohorts into samples exhibiting a mild or a severe respiratory phenotype-like gene expression signature and perform Kaplan–Meier and Hazard ratio (HR) analysis of survival probabilities. d Disease-free survival Caplan–Meier curves for the PRAD-TCGA cohort (N = 497). A HR of 0.53 (confidence interval 0.32–0.89, p = 0.001, stratified log-rank test) was calculated for this set. e Survival probabilities (HRs and confidence intervals, CI), for all five independent prostate cancer cohorts (N = 1067). Cases exhibiting a severe respiratory phenotype-like metagene signature showed a significantly shorter survival probability in four of the five cohorts. HRs were tested using stratified log-rank tests. Source data are provided as a Source Data file.