Transcription factor NRF2 regulates miR-1 and miR-206 to drive tumorigenesis

Abstract

The mechanisms by which deregulated nuclear factor erythroid-2-related factor 2 (NRF2) and kelch-like ECH-associated protein 1 (KEAP1) signaling promote cellular proliferation and tumorigenesis are poorly understood. Using an integrated genomics and ¹³C-based targeted tracer fate association (TTFA) study, we found that NRF2 regulates miR-1 and miR-206 to direct carbon flux toward the pentose phosphate pathway (PPP) and the tricarboxylic acid (TCA) cycle, reprogramming glucose metabolism. Sustained activation of NRF2 signaling in cancer cells attenuated miR-1 and miR-206 expression, leading to enhanced expression of PPP genes. Conversely, overexpression of miR-1 and miR-206 decreased the expression of metabolic genes and dramatically impaired NADPH production, ribose synthesis, and in vivo tumor growth in mice. Loss of NRF2 decreased the expression of the redox-sensitive histone deacetylase, HDAC4, resulting in increased expression of miR-1 and miR-206, and not only inhibiting PPP expression and activity but functioning as a regulatory feedback loop that repressed HDAC4 expression. In primary tumor samples, the expression of miR-1 and miR-206 was inversely correlated with PPP gene expression, and increased expression of NRF2-dependent genes was associated with poor prognosis. Our results demonstrate that microRNA-dependent (miRNA-dependent) regulation of the PPP via NRF2 and HDAC4 represents a novel link between miRNA regulation, glucose metabolism, and ROS homeostasis in cancer cells.

Figure 1. NRF2 reprograms glucose metabolism to promote tumor growth.

(A) NRF2 activity is essential for the growth of KEAP1-deficient cancer cells. A549 (KEAP1^G333C) cells stably transduced with doxycycline-inducible Nrf2 shRNA were implanted into nude mice. Two weeks after tumor implantation, mice were randomly allocated to 2 groups (n = 13–15 per group) and administered vehicle or doxycycline in the drinking water. Tumor volume was recorded. Data are presented as the mean tumor volume ± SEM. *P < 0.05 relative to the vehicle-treated group. (B) Possible ¹³C (red circles) labeling in the intermediates of the pentose cycle, lactate, and glutamate using [1,2-¹³C₂] glucose as the single tracer. (C) Comparison of ¹³CO₂ production by A549-Luc shRNA and A549-Nrf2 shRNA cultures. *P < 0.0000. (D) Decrease in ¹³C-labeled extracellular lactate in the culture medium from A549-Nrf2 shRNA cells. *P < 0.004. (E) Pentose cycle activity relative to glycolysis measured by m1/m2 ratios in lactate in the media. (F) Determination of C₂-labeled through C₅-labeled glutamate levels indicating substrate (OAA and acetyl CoA) entry into the TCA cycle. *P = 0.006 relative to the cells expressing Luc shRNA. (G) Reduced PDH flux in A549-Nrf2 shRNA cells indicating a significant decrease in the TCA cycle flux. *P < 0.0001 relative to cells expressing Luc shRNA. (H) Comparison of ¹⁴C-labeled RNA ribose fraction in Luc shRNA and Nrf2 shRNA cells. *P < 0.05. (I) Relative ¹³CO₂ production by MEF cells. Keap1^–/– cells demonstrated a significantly higher glucose oxidation rate. *P < 0.00 relative to WT. (J) Relative RNA ribose synthesis (mRNA and rRNA) through direct glucose oxidation and the nonoxidative steps of the pentose cycle in MEF cells. *P < 0.01 relative to WT.

Figure 2. Novel set of NRF2-dependent signature genes involved in central glucose metabolism.

(A) Heat map showing metabolic genes significantly altered in response to NRF2 depletion or activation in A549 (KEAP1^G333C) cells. Gene expression signatures were obtained from microarray analysis of A549 cells. Cells transfected with nontargeting siRNA (NS siRNA) and vehicle-treated samples (mock) were used as controls. (B) Real-time RT-PCR analysis of PPP genes in A549-Luc shRNA and A549-Nrf2 shRNA cells. *P < 0.05 relative to Luc shRNA. (C) Immunoblot showing relative levels of PPP enzymes in A549-Luc shRNA and A549-Nrf2 shRNA cells. (D) Real-time RT-PCR analysis of PPP genes in DU145-Luc shRNA and DU145-Nrf2 shRNA cells. *P < 0.05 relative to Luc shRNA. (E) Real-time RT-PCR analysis of NRF2 and PPP genes in vehicle- or doxycycline-treated inducible Nrf2 shRNA–expressing A549 cells. (F and G) Reduced levels of NADPH and a lower NADPH/NADP ratio in doxycycline-treated inducible Nrf2 shRNA–expressing A549 cells. Gene expression data are presented as the fold reduction in gene expression relative to the vehicle-treated group (set arbitrarily to 1). *P < 0.01 relative to the vehicle-treated group. (H) Lentiviral-mediated overexpression of G6PD or TKT in NRF2-depleted A549 cells dramatically increased the proliferation of A549-Nrf2 shRNA cells. *P < 0.01 relative to the GFP group.

Figure 3. miR-1 and miR-206 miRNA target G6PD, PGD, TKT, and GPD2 in cancer cells and non-neoplastic fibroblast cells.

(A) Sequence alignment showing the relative position of the miRNA binding site in the 3′UTR of each gene. (B) Expression of miR-1 and miR-206 decreases luciferase reporter gene activity in A549 cells when linked to the targeted segment of the 3′UTR of G6PD, PGD, TKT, and GPD2. Nonspecific control miR (NS miR) is a nontargeting control miRNA mimic. (C) Relative mRNA levels of G6PD, TKT, PGD, and GPD2 in A549 cells during expression of miR-1 and miR-206 as compared with those in cells expressing control miRNA. (D) Immunoblot analysis and (E) densitometric quantification showing relative protein levels of G6PD, TKT, PGD, and GPD2 in A549 cells during expression of miR-1 and miR-206 or control miRNA. (F) Relative levels of G6PD, TKT, PGD, and GPD2 in H1437 cells during expression of miR-1 and miR-206 as compared with those in the cells expressing nontargeting NS miRNA. Cotransfection of corresponding AMO along with precursor miRNA significantly blocks the action of miR-1 and miR-206. *P < 0.05 relative to the NS miR group.

Figure 4. Gain of NRF2 function in lung cancer upregulates the expression of PPP genes by suppressing the expression of miR-1 and miR-206.

(A and B) Increased expression of miR-1 in NRF2-deficient A549-Nrf2 shRNA and DU145-Nrf2 shRNA cells compared with that in the Luc shRNA control group. *P < 0.05 relative to Luc shRNA. (C and D) Increased expression of PPP genes in A549-Nrf2 shRNA and DU145-Nrf2 shRNA cells expressing anti–miR-1. Cells stably expressing empty vector were used as the baseline control to calculate the fold change. *P ≤ 0.01 relative to the cells expressing empty vector. (E) Increased expression of miR-1 decreases luciferase reporter gene activity associated with the 3′UTR of G6PD, PGD, TKT, and GPD2 in DU145-Nrf2 shRNA cells. The 3′UTR-containing reporter plasmids and pGLO empty vector were transfected into DU145 parent and Nrf2 shRNA cells, and the extracts were analyzed for luciferase activity. Firefly luciferase activity was normalized to Renilla luciferase activity. *P ≤ 0.01 relative to the parent cell line. (F) Increased expression of miR-206 in Nrf2^–/– MEF cells and reduced expression of miR-206 in Keap1^–/– MEF cells. *P ≤ 0.01 relative to WT cells.

Figure 5. Ectopic expression of miR-1 and miR-206 miRNAs inhibits PPP activity and reduces cell proliferation.

(A–D) Total NADPH and the NADPH/NADP ratio in A549 cells and H1437 cells transfected with nonspecific miRNA miR-1 and miR-206. Relative levels of NADPH and the NADPH/NADP ratio were measured 72 hours after transfection. (E and F) Incorporation of ¹⁴C glucose into RNA ribose in A549 and H1437 cells. Cells were incubated with ¹⁴C glucose for 14 to 16 hours and lysed, and total ¹⁴C glucose incorporation into ribose was quantified. (G and H) Effect of miR-1 and miR-206 on proliferation of A549 and H1437 cells. Seventy-two hours after transfection, the cells were replated and exposed to BrdU for 4 hours, and incorporation was quantified calorimetrically. (I) Average change in tumor volume (n = 6; mean ± SEM) with time in nude mice injected with NS-miR, miR-1, and miR-206–transfected H1437 cells. *P < 0.05 relative to the NS miR group.

Figure 6. A redox-sensitive Class IIa HDAC, HDAC4, regulates miR-1 and miR-206 expression.

(A and B) RT-PCR levels showing reexpression of miR-206 expression and repression of PPP pathway gene expression in lung cancer cells after treatment with 5 μM of 5-AzaC for 5 days followed by combined treatment with 5 μM of 5-AzaC and 0.3 μM of TSA for an additional 16 hours. *P < 0.01 relative to the vehicle-treated group. (C and D) TSA induces miR-1 and miR-206 expression and attenuates PPP gene expression in lung cancer cells. Cells were treated with 1 μm of TSA for 24 hours, and total RNA was subjected to real-time RT-PCR. *P < 0.01 relative to the vehicle-treated group. (E and F) Depletion of endogenous HDAC4 induces miR-1 and miR-206 expression and attenuates PPP gene expression in A549 lung cancer cells. Cells were transduced with lentivirus expressing HDAC4 shRNA or luciferase shRNA. Gene expression was analyzed 5 days after infection. *P < 0.01 relative to the Luc shRNA group. (G) Real-time RT-PCR analysis of HDAC4 expression in A549-Luc shRNA and A549-Nrf2 shRNA cells. *P < 0.01 relative to the Luc shRNA group. (H) Immunoblot showing nuclear and cytoplasmic localization of NRF2 and HDAC4 in A549-Luc shRNA and A549-Nrf2 shRNA cells.

Figure 7. NRF2-dependent signaling predicts survival in human NSCLC.

(A and B) miR-1 and miR-206 levels are reduced in primary lung tumors. Real-time RT-PCR–based analysis of mature miR-1 and miR-206 in primary lung adenocarcinoma tumors and pair-matched nonmalignant lung tissues. (C–E) Expression of miR-1 and miR-206 targets G6PD, TKT, and GPD2 is elevated in the primary lung adenocarcinoma tumors as compared with that in the pair-matched control tissues. (F–L) Kaplan-Meier analysis of the Director’s Challenge Consortium dataset of human lung tumors showing upregulation of (F–H) NRF2 signaling and (I–L) metabolic genes and its association with overall survival. RFC, relative fold change.

Figure 8. An autoregulatory loop involving NRF2, HDAC4, and miR-1/206 regulates glucose metabolism.

(A) A genetic model showing NRF2-dependent regulation of metabolic gene expression. The effect of NRF2 on glucose metabolism is partly mediated via miR-1 and miR-206. A redox-sensitive HDAC, HDAC4, regulates the transcription of miR-1 and miR-206. HDAC4, with cysteines in a reduced state, accumulates in the nucleus and suppresses miR-1 and miR-206 gene expression. Oxidation of cysteines in HDAC4 results in nuclear export and induces the expression of miR-1 and miR-206. (B) A549 cells harboring a KEAP1 mutation show increased glucose oxidation in the pentose cycle, with increased nucleic acid synthesis, high TCA cycle flux, and de novo fatty acid synthesis. Inhibition of NRF2 activity in KEAP1-deficient A549 cells decreases glucose oxidation in the pentose cycle, with low TCA cycle flux and de novo fatty acid synthesis. Solid arrows indicate activation, and hollow allows indicate inhibition. G6P, glucose-6-P; 3PG, 3-phosphoglyceric acid; Pyr, pyruvate; Lac, lactate.