Fumarate Mediates a Chronic Proliferative Signal in Fumarate Hydratase-Inactivated Cancer Cells by Increasing Transcription and Translation of Ferritin Genes

Abstract

Germ line mutations of the gene encoding the tricarboxylic acid (TCA) cycle enzyme fumarate hydratase (FH) cause a hereditary cancer syndrome known as hereditary leiomyomatosis and renal cell cancer (HLRCC). HLRCC-associated tumors harbor biallelic FH inactivation that results in the accumulation of the TCA cycle metabolite fumarate. Although it is known that fumarate accumulation can alter cellular signaling, if and how fumarate confers a growth advantage remain unclear. Here we show that fumarate accumulation confers a chronic proliferative signal by disrupting cellular iron signaling. Specifically, fumarate covalently modifies cysteine residues on iron regulatory protein 2 (IRP2), rendering it unable to repress ferritin mRNA translation. Simultaneously, fumarate increases ferritin gene transcription by activating the NRF2 (nuclear factor [erythroid-derived 2]-like 2) transcription factor. In turn, increased ferritin protein levels promote the expression of the promitotic transcription factor FOXM1 (Forkhead box protein M1). Consistently, clinical HLRCC tissues showed increased expression levels of both FOXM1 and its proliferation-associated target genes. This finding demonstrates how FH inactivation can endow cells with a growth advantage.

Keywords: FH; FOXM1; HLRCC; NRF2; ferritin; fumarate.

FIG 1

Fumarate accumulation confers chronic proliferation. (A) Western blots of NRF2 and its transcription target AKR1B10 after reintroduction of FH into the FH^−/− HLRCC tumor cell line UOK262. β-Actin (ACTB) was used as a loading control. (B) Cell viability measured by formazan production in UOK262 cells (FH^−/−) after reintroduction of FH (UOK262-FH^res). Data are presented as means ± standard deviations of results from a representative experiment. Curves were statistically significantly different, as determined by two-way analysis of variance, for the FH genotype (P < 0.001) but not for time (P > 0.05). (C) Cell viability measured by formazan production in UOK262 cells (FH^−/−) after reintroduction of FH (UOK262-FH^res) and 80 μM MMF treatment. Data are presented as means ± standard deviations of results from a representative experiment. Curves were statistically significantly different, as determined by two-way analysis of variance, for MMF treatment (P < 0.001) but not for time (P > 0.05).

FIG 2

Fumarate accumulation increases ferritin levels. (A) Ferritin transcriptional and translational regulation. NRF2 can transcribe the FTL and FTH1 genes. IRP2 can bind FTL and FTH1 mRNA iron response element hairpins in the 5′ untranslated regions to repress translation. (B) Western blots of HK2 cells show relative increases in FTL and FTH1 protein levels upon siRNA-mediated FH knockdown. ACTB was used as a loading control. (C) Western blots of HEK293 cells show relative increases in FTL and FTH1 protein levels upon CRISPR-Cas9-mediated FH knockout. Increased NRF2 levels indicate a stable knockout. ACTB was used as a loading control. (D) Western blots of FH-rescued (FH^res) and control (FH^−/−) UOK262 cells show increased FTL and FTH1 levels under conditions of nonfunctional FH. ACTB was used as a loading control. (E) Addition of 80 μM MMF or 40 μM DMF to HEK293 cells increases FTL and FTH1 protein levels. ACTB was used as a loading control.

FIG 3

IRP2 is a succination target. (A) Multiple-sequence alignment of mouse Aco2, human IRP2 (ACO3), and IRP1 (ACO1). Succination sites reported for Fh^−/− mouse Aco2 are conserved in human IRP1 and IRP2. These sites are highlighted in red. (B) Adduct sites on Flag-tagged wild-type IRP2, as determined by LC-MS/MS. Black bars represent peptides detected by mass spectrometry analysis. (C) Adduct sites on Flag-tagged IRP2-A523R, as determined by LC-MS/MS. Black bars represent peptides detected by mass spectrometry analysis.

FIG 4

Fumarate inhibits IRP-mediated translational repression. (A and B) IRE reporter assays using the Dual-Glo luciferase assay (Promega) indicate appropriate responsiveness to iron levels for both pECE-FTL-IRE-LUX (A) and pECE-FTH1-IRE-LUX (B) cotransfected with pECE-RL into HEK293 cells. Cells were treated with ferric citrate (Fe; 200 micromolar) or an iron chelator (DFO; 200 micromolar) to modulate iron levels. Data are presented as means ± standard deviations. * indicates a P value of <0.05 relative to vehicle (Veh) treatment, as determined by Student's t test. RLU, relative light units. (C) Immunoblots of HEK293 cells treated with DFO show increased IRP2 and decreased FTL, FTH1, and FBXL5 protein levels. Treatment with Fe resulted in opposite effects. ACTB was used as a loading control. (D to G) IRE reporter assays using the Dual-Glo luciferase assay (Promega) with dimethyl fumarate and monomethyl fumarate treatments. HEK293 cells were cotransfected with pECE-RL and either pECE-FTL-IRE-LUX (D and F) or pECE-FTH1-IRE-LUX (E and G). Cells were treated with increasing concentrations of either dimethyl fumarate (D and E) or monomethyl fumarate (F and G). Data are presented as means ± standard deviations. * indicates a P value of <0.05 relative to vehicle treatment, as determined by Student's t test. (H and I) Western blots of HEK293 cells treated with increasing concentrations of DMF (H) or MMF (I) show dysregulated iron signaling, as indicated by simultaneously high protein levels of IRP2, FBXL5, FTL, and FTH1. ACTB was used as a loading control. (J) Western blots of UOK262-FH^−/− control cells and UOK262-FH^res cells. FH reconstitution leads to decreased NRF2, IRP2, FBXL5, FTH1, and FTL protein levels. ACTB was used as a loading control.

FIG 5

NRF2 activation by fumarate contributes to ferritin accumulation. (A) Quantitative reverse transcription-PCR shows that FH reconstitution in UOK262 (UOK262-FH^res) cells decreases relative FTL and FTH1 transcript levels. Changes in expression levels were normalized to values for UOK262-FH^res cells. GCLM is an NRF2 target gene known to be upregulated in HLRCC. Error bars represent standard deviations. * indicates a P value of <0.05 as determined by Student's t test. (B) Quantitative reverse transcription-PCR of HEK293-NRF2^−/− and HEK293-NRF2^+/+ cells treated with 40 μM DMF or the vehicle (Veh). Changes in expression levels were normalized to values for vehicle treatment. Error bars represent standard deviations. * indicates a P value of <0.05 as determined by Student's t test. (C) Western blot of HEK293 cells transfected with pIRESpuro3 (vector) or NRF2-Myc and treated with the vehicle or 80 μM MMF. ACTB was used as a loading control. (D) Western blot of HEK293-NRF2^−/− and HEK293-NRF2^+/+ cells treated with the vehicle, 80 μM MMF, or 40 μM DMF. ACTB was used as a loading control. (E) Western blot of HEK293-NRF2^−/− cells treated with increasing concentrations of DMF, 200 μM DFO, or 200 μM Fe.

FIG 6

FH inactivation promotes FOXM1 signaling. (A) Western blotting of the FH-inactive UOK262-FH^−/− cell line and the FH-reconstituted UOK262-FH^res cell line. (B) Quantitative reverse transcription-PCR shows that FH reconstitution in UOK262 cells decreases the relative transcript levels of FOXM1 and its downstream targets AURKA, AURKB, and CDK1. Changes in expression levels were normalized to values for UOK262-FH^res cells. GCLM is an NRF2 target gene known to be upregulated in HLRCC. Data are presented as means ± standard deviations of results from a representative experiment. * indicates a P value of <0.05 as determined by Student's t test. (C) Western blots of UOK262-FH^res cells treated with the vehicle or 80 μM monomethyl fumarate show increased FOXM1 signaling with MMF. (D) Western blots of UOK262-FH^res cells and HEK293 cells transduced with the pLKO-CMV empty vector (control) or pLKO-CMV-FTL and pLKO-CMV-FTH1 (ferritin) show that ferritin overexpression increases FOXM1 protein levels. (E) Western blots of UOK262 cells transduced with shRNAs targeting FTL (shRNA-FTL) or FTH1 (shRNA-FTH1) show decreased FOXM1 signaling relative to shRNA-Scr, a nontargeting shRNA control. (F) Cell viability measured by formazan production in UOK262 cells after transduction with shRNA-Scr, shRNA-FTL, and shRNA-FTH1. Data are presented as means ± standard deviations of results from a representative experiment. shRNA-FTL and shRNA-FTH1 knockdown curves were statistically significantly different from the shRNA-Scr curve for shRNA treatment (P < 0.001) but not for time (P > 0.05), as determined by two-way analysis of variance. (G) Relative mRNA levels of ferritin genes (FTL and FTH1), FOXM1, FOXM1 target genes (CDK1, AURKA, and AURKB), and nonferritin NRF2 targets (GCLM and AKR1B10) in HLRCC tumors and in normal kidney tissues. FDR, false discovery rate.

FIG 7

FH inactivation confers chronic proliferative signaling. The diagram shows how FH inactivation increases ferritin levels to promote proliferation. FH, a TCA cycle enzyme that catalyzes the hydration of fumarate (represented by green hexagons) to form malate (represented by yellow hexagons), is inactivated (represented by a red cross) in HLRCC. FH inactivation causes fumarate accumulation. The accumulated fumarate activates the NRF2 transcription factor, which transcribes the ferritin genes FTL and FTH1. Concurrently, the accumulated fumarate also inhibits IRP2, which normally represses the translation of FTL and FTH1, leading to a net increase in the ferritin level and subsequent FOXM1-dependent proliferative signaling.