The NADPH oxidase NOX4 regulates redox and metabolic homeostasis preventing HCC progression

Abstract

Background and aims: The NADPH oxidase NOX4 plays a tumor-suppressor function in HCC. Silencing NOX4 confers higher proliferative and migratory capacity to HCC cells and increases their in vivo tumorigenic potential in xenografts in mice. NOX4 gene deletions are frequent in HCC, correlating with higher tumor grade and worse recurrence-free and overall survival rates. However, despite the accumulating evidence of a protective regulatory role in HCC, the cellular processes governed by NOX4 are not yet understood. Accordingly, the aim of this work was to better understand the molecular mechanisms regulated by NOX4 in HCC in order to explain its tumor-suppressor action.

Approach and results: Experimental models: cell-based loss or gain of NOX4 function experiments, in vivo hepatocarcinogenesis induced by diethylnitrosamine in Nox4 -deficient mice, and analyses in human HCC samples. Methods include cellular and molecular biology analyses, proteomics, transcriptomics, and metabolomics, as well as histological and immunohistochemical analyses in tissues. Results identified MYC as being negatively regulated by NOX4. MYC mediated mitochondrial dynamics and a transcriptional program leading to increased oxidative metabolism, enhanced use of both glucose and fatty acids, and an overall higher energetic capacity and ATP level. NOX4 deletion induced a redox imbalance that augmented nuclear factor erythroid 2-related factor 2 (Nrf2) activity and was responsible for MYC up-regulation.

Conclusions: Loss of NOX4 in HCC tumor cells induces metabolic reprogramming in a Nrf2/MYC-dependent manner to promote HCC progression.

Graphical abstract

FIGURE 1

MYC expression and activation inversely correlates with NADPH oxidase (NOX4) levels. (A) Ingenuity Pathway Analysis of proteomic data in PLC/PRF/5 shNOX4 cells compared with shControl cells. Edges and nodes are color‐coded based on the predicted relationship as indicated in the prediction legend. (B) Relative MYC mRNA expression. Data are presented as mean ± SD (n = 10). (C) Left: MYC protein levels. β‐Actin was used as loading control. Right: Densitometric analysis, expressed as relative to β‐Actin. Data are presented as mean ± SD (n = 3). (D) Left: Immunofluorescence of MYC (green) and DAPI (blue) for nuclei staining. Scale bar, 20 μm. Right: Quantification of nuclear fluorescence intensity. Each dot represents one nucleus. (E) Transcriptional MYC activity. Results are shown relative to each control. Data are presented as mean ± SD (n = 12–16). (F) Left: EIF4H protein levels. β‐Actin was used as loading control. Right: Densitometric analysis, expressed as relative to β‐actin. Data are presented as mean ± SD (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001

FIGURE 2

Analysis of Nox4/NOX4 and Myc/MYC expression in mice/human under a hepatocarcinogenic process (see Figures S4 and S5 for complementary information). (A) Ki67 immunohistochemistry in tumors from wild‐type (WT) and Nox4 ^−/− mice at 11 months following diethylnitrosamine (DEN) treatment. Scale bar, 50 μm. Right: Quantification of positive nuclei. Each dot represents one nucleus. (B) Myc mRNA expression in tumor (T) and nontumor (NT) areas from WT and Nox4 ^−/− mice at 11 months following DEN treatment. Data are presented as a ratio. (A,B) Results are expressed as box plots with whiskers (min to max) (n = 9; 3 lysates from 3 different animals). (C) Pearson correlation analysis of the ratio T/NT between NOX4 and MYC gene expression in all patients used in the study. (D) MYC expression in NOX4‐low (Q1) patients versus the others, expressed as a percentage. (E) Kaplan–Meier curve for relapse‐free survival in patients with NOX4‐low/MYC‐high HCC versus the others. (F,G) Percentage of patients with HCC according to microinvasion (F) and histological grade (G) in patients with NOX4‐low/MYC‐high HCC versus other patients. Differences are considered statistically significant when p‐values < 0.05. **p < 0.01

FIGURE 3

Low levels of NOX4 correlate with a more elongated mitochondrial phenotype, higher respiration capacity, and ATP production. (A) Top: Immunocytochemical analysis of mitochondrial ATP synthase subunit beta (ATPb) in green. DAPI (blue) was used for nuclei staining. Scale bar, 10 μm. Bottom: Quantification of mitochondrial size, circularity, branches per network, and branch length. Each dot represents the average of all mitochondria from independent experiments (n = 5). (B) Intracellular ATP levels normalized to cell number. Data are presented as mean ± SD (n = 3). (C) Mitochondrial to nuclear DNA ratio. Data are presented as mean ± SD (n = 3). (D) Mitochondrial dynamics‐related protein levels. Tubulin was used as loading control. A representative experiment is shown. (E,F) Seahorse analysis of oxidative phosphorylation (OXPHOS) in PLC/PRF/5 (E) and SNU449 (F) HCC cells. Left: Continuous oxygen consumption rate (OCR) values ([pmoles O₂/min]/cell number) are shown. Middle: Bar graphs of mitochondrial functions were analyzed as explained in the Supporting Information. Data are presented as the mean ± SEM (n = 9–12). Right: Fold mRNA expression of different genes related to OXPHOS. Top: PLC/PRF/5 shNOX4. Bottom: SNU449 + NOX4, normalized to each control. Data are presented as mean ± SD (n = 3). (G) Electron transport chain‐related protein levels. Tubulin was used as loading control. A representative experiment is shown. *p < 0.05, **p < 0.01, ***p < 0.001. CS, citrate synthase; DLD, dihydrolipoamide dehydrogenase; DLST, dihydrolipoamide S‐succinyltransferase; FH, fumarate hydratase; IDH2, isocitrate dehydrogenase 2; IDH3A, isocitrate dehydrogenase 3A; MDH1, malate dehydrogenase 1; MDH2, malate dehydrogenase 2; PCK2, phosphoenolpyruvate carboxykinase 2; PDHA1, pyruvate dehydrogenase E1 subunit alpha 1; PDHB, pyruvate dehydrogenase E1 subunit beta; SDH, succinate dehydrogenase complex flavoprotein subunit A; SUCLG1, succinate‐CoA ligase GDP‐ADP forming subunit alpha.

FIGURE 4

MYC mediates the changes observed in mitochondrial phenotype, respiration capacity, and transcriptional reprogramming of OXPHOS‐related genes in NOX4 silenced cells. (A–F) PLC/PRF/5 shNOX4 were transiently transfected either with control or MYC‐specific small interfering RNA (siRNA) sequences for 48 h. (A) Left: MYC protein levels. β‐Actin was used as loading control. A representative experiment is shown. Middle: Cell proliferation at 72 h. Right: Cell migration at 8 h. Results are expressed as slope (h⁻¹). Data are presented as mean ± SEM (n = 10–16). (B) Left: Immunocytochemistry of ATPb in green and DAPI (blue). Scale bar, 10 μm. Right: Quantification of mitochondrial size, circularity, branches per network, and branch length. Each dot represents the average of all mitochondria from independent experiments (n = 7). (C) Mitochondrial dynamics–related protein levels. Tubulin was used as loading control. A representative experiment is shown. (D) Seahorse analysis of OXPHOS. Left: Continuous OCR values ([pmoles O₂/min)/cell number) are shown. Right: Bar graphs of mitochondrial functions were analyzed as explained in the Supporting Information. Data are presented as mean ± SEM (n = 9–12). (E) Electron transport chain–related protein levels. Tubulin was used as loading control. A representative experiment is shown. (F) Intracellular ATP levels normalized to cell number. Data are presented as mean ± SD (n = 3). (G) PLC/PRF/5 cells were transiently transfected either with control or MYC‐specific siRNA sequences for 48 h. Left: MYC protein levels. β‐Actin was used as loading control. A representative experiment is shown. Right: OXPHOS‐related gene expression. Results are expressed as fold mRNA expression, in which each MYC siRNA condition was normalized to its respective unspecific sequence. Data are presented as mean ± SD (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001. FH, fumarate hydratase; SDHB, succinate dehydrogenase complex iron sulphur subunit B; SDHC, succinate dehydrogenase complex subunit C; SDHD, succinate dehydrogenase complex subunit D.

FIGURE 5

The axis low NOX4/high MYC increases lipid catabolism. (A) Volcano plot of metabolites related to metabolism of monoacylglycerol, diacylglycerol, and acylcarnitine intermediates. It is represented as fold change relative to each control (n = 6 for each group). Detailed information of depicted metabolites is provided in Tables S4 and S5. (B) mRNA fold expression of lipid metabolism–related genes, relative to each control. Data are presented as mean ± SD (n = 3). (C) Lipid metabolism–related protein levels (from proteomics analysis), shown as ratio to each respective control. Data are presented as mean ± SD (n = 3). (D) PLC/PRF/5 shControl and shNOX4 cells were transiently transfected either with control or MYC‐specific siRNA sequences for 48 h. Expression of lipid metabolism–related genes. Results are expressed as fold mRNA expression, where each MYC siRNA condition was normalized to its respective unspecific sequence. Data are presented as mean ± SD (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001. ACAA1, acetyl‐CoA acyltransferase 1; ACAA2, acetyl‐CoA acyltransferase 2; ACAD9, acyl‐CoA dehydrogenase family, member 9; ACAD10, acyl‐CoA dehydrogenase family, member 10; ACAD11, acyl‐CoA dehydrogenase family, member 11; ACADM, acyl‐CoA dehydrogenase, C‐4 to C‐12 straight chain; ACADSB, acyl‐CoA dehydrogenase, short/branched chain; ACADVL, acyl‐CoA dehydrogenase, very long chain; ACOD, cis‐aconitate decarboxylase; ACOX1, acyl‐CoA oxidase 1, palmitoyl; ACOX2, acyl‐CoA oxidase 2, branched chain; ACOX3, acyl‐CoA oxidase 3, pristanoyl; ACSBG1, acyl‐CoA synthetase bubblegum family member 1; ACSL3, acyl‐CoA synthetase long‐chain family member 3; ACSL5, acyl‐CoA synthetase long‐chain family member 5; BDH2, 3‐hydroxybutyrate dehydrogenase, type 2; CPT1A, carnitine palmitoyltransferase 1 A (liver); CPT1C, carnitine palmitoyltransferase 1C; DGLB, diacylglycerol lipase‐beta; ECHA, trifunctional enzyme subunit alpha, mitochondrial; ECI1, enoyl‐CoA delta isomerase 1, mitochondrial; ECI2, enoyl‐CoA delta isomerase 2; ETFA, electron transfer flavoprotein subunit alpha, mitochondrial; ECH1, delta(3,5)‐delta(2,4)‐dienoyl‐CoA isomerase, mitochondrial; ECHM, enoyl‐CoA hydratase, mitochondrial; FABP1, fatty acid binding protein 1, liver; FASN, fatty acid synthase; HCD2, 3‐hydroxyacyl‐CoA dehydrogenase type‐2; IVD, isovaleryl‐CoA dehydrogenase, mitochondrial; MCEE, methylmalonyl CoA epimerase; MUT, methylmalonyl CoA mutase; PECR, peroxisomal trans‐2‐enoyl‐CoA reductase; PPARA, peroxisome proliferator activated receptor alpha; PPARD, peroxisome proliferator‐activated receptor delta; PPARG, peroxisome proliferator activated receptor gamma; PRKAA1, protein kinase, AMP‐activated, alpha 1 catalytic subunit; PRKAA2, protein kinase, AMP‐activated, alpha 2 catalytic subunit; PRKAG1, protein kinase, AMP‐activated, gamma 1 non‐catalytic subunit; SLC27A2, solute carrier family 27 (fatty acid transporter), member 2.

FIGURE 6

NOX4 levels inversely correlate with glycolysis and alters glucose consumption and lactate production in HCC cells. (A) Seahorse analysis of glycolysis in PLC/PRF/5 cells. Left: Continuous extracellular acidification rate (ECAR) values (mpH/min/μg protein) are shown. Data are presented as mean ± SEM (n = 9–10). Right: Bar graphs of glycolytic functions were analyzed as explained in the Supporting Information. (B) mRNA fold expression of genes related to glucose metabolism in PLC/PRF/5 shNOX4 (left) and SNU449 + NOX4 (right), relative to each respective control. Data are presented as mean ± SD (n = 3). (C) Left: Glucose consumption under basal conditions and normalized to protein content. Data are presented as mean ± SD (n = 3). Right: Glucose‐6‐phosphate metabolite is depicted by a box plot with whiskers (min to max) (n = 6 for each group). (D) Left: Lactate production under basal conditions and normalized to protein content. Data are presented as mean ± SD (n = 3). Right: Relative lactate dehydrogenase A (LDHA) mRNA expression. Data are presented as mean ± SD (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001

FIGURE 7

Nrf2 expression and activation inversely correlate with NOX4 levels. Cross‐talk with the MYC pathway. (A) Top: Nrf2 protein levels. β‐Actin was used as loading control. Bottom: Densitometric analysis, expressed as relative to β‐actin. Data are presented as mean ± SD (n = 3). (B) Left: Immunofluorescence of Nrf2 (green) and DAPI (blue) for nuclei staining. Scale bar, 20 μm. Right: Quantification of nuclear fluorescence intensity. Each dot represents one nucleus from independent experiments (n = 3). (C) Transcriptional antioxidant response element (ARE) activity. Results are shown relative to each control. Data are presented as mean ± SD (n = 4–6). (D) NQO1 and HMOX1 mRNA expression. Data are presented as mean ± SD (n = 3). Right: HMOX1 protein levels. β‐Actin was used as loading control. (E) Hmox1 and Nqo1 mRNA expression in PBS‐treated livers and DEN‐induced tumors from WT and Nox4 ^−/− mice at 11 months following DEN treatment. Data are presented as relative expression and presented as box plots with whiskers (min to max) (n = 9; 3 lysates from 3 different animals). (F) PLC/PRF/5 shNOX4 cells were transiently transfected either with control, MYC, or Nrf2‐specific siRNA sequences for 72 h. Left: MYC and Nrf2 protein levels. β‐Actin was used as loading control. A representative experiment is shown. Middle: MYC mRNA expression expressed as Nrf2 siRNA versus unspecific siRNA. Data are presented as mean ± SD (n = 4). Right: Transcriptional MYC activity. Data are presented as mean ± SD (n = 8). *p < 0.05, **p < 0.01, ***p < 0.001

FIGURE 8

NOX1 and NOX2 induce Nrf2 and MYC transcriptional activity in HCC cells. (A) NOX1, NOX2, and nitric oxide synthase 1 (NOS1) and nitric oxide synthase 2 (NOS2) (left and middle) mRNA expression analyzed by real‐time quantitative PCR. Data are presented as mean ± SD (n = 3–5). (B) Right: Extracellular nitrite levels analyzed using the Griess reaction. Data are presented as mean ± SD (n = 7). (PLC/PRF/5 shNOX4 (C) and (D) SNU449 + Control (D) cells were transiently transfected either with control or NOX1 (C) or NOX2 specific (D) siRNA sequences for 72 h. Left: Transcriptional ARE activity. Right: Transcriptional MYC activity. Results are shown relative to each control. Data are presented as mean ± SD (n = 6). *p < 0.05, **p < 0.01, ***p < 0.001. (E) Graphical summary. Image created using BioRender software (https://www.biorender.com)