Mutant Isocitrate Dehydrogenase 1 Sensitizes Intrahepatic Cholangiocarcinoma Cells to MDM2 Inhibitors

Abstract

Mutations in isocitrate dehydrogenase 1 (IDH1) occur in 10% to 25% of intrahepatic cholangiocarcinoma (iCCA) cases. Despite significantly prolonged progression-free survival, the mutant IDH1 (mIDH1) inhibitor ivosidenib achieved only a 3% response rate in clinical trials, highlighting the need for new therapeutic options for IDH1 mutation (IDH1^mut) iCCA. Our in silico analysis demonstrated that IDH1^mut and TP53 mutation (TP53^mut) were mutually exclusive in iCCA cells and that IDH1^mut iCCA cells expressed higher mouse double minute 2 homolog (MDM2) levels than IDH1^wt iCCA cells. Chromatin immunoprecipitation quantitative polymerase chain reaction assay showed enrichment of histone-3-lysine-4 tri-methylation (H3K4me3), an indicator of active gene transcription, at the MDM2 promoter in IDH1^mut iCCA cells, confirming the data from ENCODE histone-seq. Treatment with a mIDH1 inhibitor reduced 2-hydroxyglutarate (2-HG) levels, enhanced lysine-specific demethylase 5 (KDM5) activity, and attenuated the H3K4me3/H3K4me1 ratio at the MDM2 promoter, which was accompanied by a reduction in MDM2 expression and an increase in wild-type TP53 (wtTP53) protein levels in IDH1^mut iCCA cells. The effect of the mIDH1 inhibitor on MDM2 mRNA levels was reversed by treatment with KDOAM-25 citrate, a pan-KDM5 inhibitor. In addition, MDM2 inhibitors that could block MDM2-mediated wtTP53 degradation selectively induced TP53 reactivation, cell-cycle arrest, and growth inhibition in IDH1^mut iCCA cells. The combination of mIDH1 and MDM2 inhibitors synergistically suppressed the proliferation of IDH1^mut iCCA cells. Our study delineated a novel mIDH1-MDM2-wtTP53 axis and its potential application for wtTP53 reactivation therapy in IDH1^mut iCCA.

Significance:

IDH1 mutation enhances MDM2 expression by inhibiting KDM5 activity to promote the proliferation of TP53^wt iCCA cells. Cotargeting MDM2 and mIDH1 yields a synergistic effect on growth inhibition, providing a new strategy for treating patients with iCCA with IDH1 mutations.

Introduction

Biliary tract cancer (BTC) is a heterogeneous group of malignancies arising from epithelial cells of the biliary tract. Based on their primary location, BTCs are divided into intrahepatic, perihilar, and distal cholangiocarcinoma; gallbladder cancer; and ampulla of Vater carcinoma, which have different trends of incidence changes, geographic distribution, risk factors, clinical outcomes, and genetic alterations (1–3). Intrahepatic cholangiocarcinoma (iCCA), next to hepatocellular carcinoma, is the second most common hepatic malignancy. Owing to the lack of specific symptoms and signs, iCCA is usually diagnosed at a late stage with local invasion and/or intrahepatic or extrahepatic dissemination (1). According to the Surveillance, Epidemiology, and End Results Program database, the 5-year overall survival rate of iCCA diagnosed between 2012 and 2018 in the United States was 9%, with 23%, 9%, and 3% for patients with localized, regional, and metastatic diseases, respectively (4).

Before the era of precision medicine, gemcitabine plus cisplatin was the standard systemic treatment for patients with unresectable and metastatic iCCA, similar to other BTCs (5, 6). The isocitrate dehydrogenase (IDH) protein is a metabolic enzyme responsible for converting isocitrate into α-ketoglutarate (α-KG). Somatic mutations in IDH1 or IDH2 (IDH1/2^mut) are frequently identified in several cancers, including 10% to 25% of iCCA cases (7). The mutant IDH1/2 (mIDH1/2) protein encoded by a gain-of-function IDH1/2 mutation converts α-KG to 2-hydroxyglutarate (2-HG), an oncometabolite that competitively suppresses the activity of α-KG–dependent dioxygenases (α-KGD), including DNA and histone demethylases, to promote epigenetic reprogramming and tumorigenesis (8–10). Based on the structural differences between the catalytic pockets of wild-type (wt) IDH1/2 (wtIDH1/2) and mIDH1/2, several inhibitors targeting mIDH1 and/or mIDH2 have been developed and have undergone clinical trials (11). AG120 (ivosidenib) is the first FDA-approved agent for IDH1^mut iCCA. However, despite significant improvement in progression-free survival (PFS) versus placebo control, the confirmed response rate and absolute PFS increase of second-line ivosidenib were 2.4% and 0.7 months, respectively, in the pivotal, ClarIDHy trial for IDH1^mut, chemotherapy-refractory iCCA (12–14). In addition, patients who received long-term ivosidenib treatment might develop acquired resistance to ivosidenib through acquired resistance mutations that impair ivosidenib binding affinity, such as the conversion of catalytic arginine substitution and secondary IDH1^mut at the dimer-interface region or isoform switch by acquired IDH2^mut (15–18). Taken together, ivosidenib treatment could result in a 22% 12-month PFS rate but a modest tumor response rate and might acquire secondary resistance mutations after long-term use, suggesting that other therapeutic interventions beyond mIDH1 inhibition are warranted for patients with IDH1^mut iCCA. Herein, we report a novel mechanism by which mIDH1-derived 2-HG inhibits lysine-specific demethylase 5 (KDM5) to upregulate the expression of MDM2 via histone methylation and selectively promotes the proliferation of IDH1^mut iCCA cells. In addition, the combination of mIDH1 and MDM2 inhibitors may provide a new strategy for treating patients with IDH1^mut iCCA.

Materials and Methods

Clinical sample gene mutation profile analysis

Mutation profiling datasets were obtained using cBioPortal software (RRID:SCR_014555; refs. 19–21). Odds ratios (OR) were calculated using the following formula: (IDH1^wt, TP53^wt sample number × IDH1^mut, TP53^mut sample number)/(IDH1^wt, TP53^mut sample number × IDH1^mut, TP53^wt sample number). The P value of the OR was calculated using Fisher’s exact test.

Immunohistochemistry staining

The use of human iCCA tissue samples in this study was approved by the Institutional Review Board of the National Cheng Kung University Hospital (IRB A-ER-111-019). Formalin-fixed, paraffin-embedded iCCA samples were sectioned and stained with hematoxylin and eosin. For MDM2 IHC staining, sections were incubated with MDM2 antibody, secondary mouse antibody, and diaminobenzidine for colorimetric detection. Three independent pathologists who were unaware of the genetic alterations evaluated the IHC staining results. The histochemical scoring (H-score) assessment was based on the staining intensity of MDM2 and the proportion of stained cancer cells.

Cell lines

The following cell lines were obtained from the respective sources: HuCCT1 (JCRB0425, RRID:CVCL_0324) and HuH28 (JCRB0426, RRID: CVCL_2955) were obtained from the Japanese Collection of Research Bioresources Cell Bank, SNU1079 (No.01079, RRID:CVCL_5008) was purchased from the Korean Cell Line Bank, and RBE (RRID:CVCL_4896) and SSP25 (RRID:CVCL_4902) were kindly provided by Dr. Chien-Feng Li (Chi Mei Medical Center, Tainan, Taiwan) and Dr. Chun-Nan Yeh (Chang Gung University, Taoyuan, Taiwan), respectively. HuCCT1, RBE, SSP25, and SNU1079 cells were cultivated in RPMI-1640 medium supplemented with 10% FBS and 1% 100× penicillin–streptomycin–glutamine. HuH28 cells were cultivated in DMEM with high glucose supplemented with 20% FBS and 1% 100× penicillin–streptomycin–glutamine. All cell lines were incubated in a 37°C humidified incubator with 5% CO₂ and passaged by trypsinization with 0.25% trypsin and protease solution. All cell lines were verified by short tandem repeat testing, and experiments were performed on cells cultivated for less than 10 passages.

Intracellular 2-HG measurement

To detect intracellular 2-HG concentration, the D-2-HG assay kit (Sigma-Aldrich, MAK320) was used following the manufacturer’s protocol. Cells were seeded in 10 cm dishes (5 × 10⁵ cells per dish) for 24 hours and treated with either mIDH1 inhibitors or dimethyl sulfoxide (DMSO) control for 48 hours. After treatment, cells were washed with ice-cold PBS, scraped, and collected by centrifugation. The collected cells were lysed in CelLytic M buffer (Sigma-Aldrich, C2978), and the protein concentration was quantified. The lysate was deproteinized by perchloric acid (PCA) precipitation. After PCA precipitation, the supernatants for 2-HG detection were collected and stored at −80°C. The supernatants were mixed with D-2-HG complete reaction and incubated for 60 minutes at 37°C. The fluorescence was measured at λ_ex = 540 nm and λ_em = 590 nm using CLARIOstar Plus (RRID:SCR_026330). The intracellular D-2-HG concentrations were normalized to their corresponding protein concentrations.

RNA extraction and complementary DNA synthesis

RNA was extracted using the Total RNA Miniprep Purification Kit (GeneMark, TR01), following the manufacturer’s protocol. Cells were first cultivated in six-well plates (4 × 10⁴ cells/well) for 24 hours and then treated with inhibitors or DMSO control for 48 hours. After treatment, cells were washed with cold PBS and lysed in an RNA lysis solution containing 1% 2-mercaptoethanol. Lysates were purified using a two-step wash with RNA wash solution I and ethanol/RNA wash solution II. The purified RNA was eluted in nuclease-free water and stored at −80°C. RNA quality was measured using a nanophotometer. Complementary DNA (cDNA) was reverse-transcribed from purified RNA using the ReverTra Ace -α- kit (TOYOBO, FSK-101), following the manufacturer’s instructions. One microgram of total purified RNA was mixed with oligo-dT primers, RNase inhibitor, dNTP mixture, RT buffer, and RNase-free water. cDNA was amplified in a thermal cycler using the following program: 42°C for 20 minutes, 99°C for 5 minutes, and 4°C for 5 minutes. The cDNA was stored at −20°C.

Real-time polymerase chain reaction

For real-time polymerase chain reaction (PCR), we followed the manufacturer’s protocol and used a GoTaq qPCR Master Mix kit (Promega, A6001). GoTaq qPCR master mix, forward primer, reverse primer, and nuclease-free water were mixed with cDNA. The mixture was quantified on a QuantStudio 5 Real-Time PCR System (RRID:SCR_020240), using a standard qPCR amplification and dissociation program. The 2^ΔΔCT value was calculated and normalized to the corresponding control group. Actin was used as the internal control for qPCR analyses. The sequences of the primers are as follows: ACTIN forward: 5′-CTGGAACGGTGAAGGTGACA-3′, ACTIN reverse: 5′-AAGGGACTTCCTGTAACAACGC-3′, MDM2 forward: 5′-GAATCATCGGACTCAGGTACATC-3′, and MDM2 reverse: 5′-TCTGTCTCACTAATTGCTCTCCT-3′.

Western blotting

Cells were first cultivated in a six-well plate (4 × 10⁴ cells/well) for 24 hours and treated with inhibitors or DMSO control for 72 hours. To detect wtTP53 protein expression in IDH1 mutation iCCA cells, cells were pretreated with the proteasome inhibitor MG-132 (15 μmol/L) for 2 hours before harvest. Cells were lysed in CelLytic M buffer (Sigma-Aldrich, C2978) containing a protease inhibitor cocktail and phosphatase inhibitors. The lysates were purified by centrifugation at 16,000 × g for 15 minutes at 4°C. The supernatants were collected and quantified using the Bradford protein assay (Bio-Rad, 5000006). The lysates were mixed with 4× loading dye buffer and heated at 95°C for 15 minutes. Protein lysates were loaded onto SDS-PAGE gels. The gels were run at 80 V for 2 hours and then transferred overnight at 30 V at 4°C. The membranes were sequentially blocked with 5% non-fat milk/TBST buffer, primary antibodies, and secondary antibodies. Chemiluminescence was activated by enhanced chemiluminescence and detected by exposure to X-ray film or the ChemiDoc Imaging System (RRID:SCR_021693). Target protein expression was normalized to the expression of the internal control proteins.

Nuclear protein extraction and KDM5 activity measurement

To detect KDM5 activity, we performed nuclear extraction using a nuclear extraction kit (Abcam, ab113474) and the KDM5 activity quantification assay kit (Abcam, ab113464), following the manufacturer’s protocol. Cells were seeded in 10 cm dishes (5 × 10⁵ cells per dish) for 24 hours and treated with either mIDH1 inhibitors or DMSO control for 48 hours. The cells were washed with ice-cold PBS, and cell pellets were collected by scraping and centrifuging. Cell pellets were lysed in 1× Pre-Extraction Buffer containing 0.1% DTT and 0.1% 100× protease inhibitor cocktail, vortexed, and centrifuged to collect the nuclear pellets. Nuclear pellets were lysed in Extraction Buffer containing 0.1% DTT and 0.1% 100× protease inhibitor cocktail, vortexed, and centrifuged to collect nuclear protein. The nuclear extracts were collected, quantified, and stored at −80°C until KDM5 activity measurement. For KDM5 activity measurement, 10 μg of purified nuclear extract was mixed with reaction buffer containing completed assay buffer and substrate, incubated at 37°C for 120 minutes, and washed with wash buffer. Capture antibody was added to the reaction wells, covered with Parafilm M, and incubated at room temperature for 60 minutes. After incubation, the capture antibody solution was removed and washed with wash buffer. Detection antibody was added to the reaction wells, covered with Parafilm M, and incubated at room temperature for 30 minutes. After incubation, the detection antibody solution was removed and washed with wash buffer. Fluoro developer was then added to the reaction wells and incubated at room temperature for 4 minutes in the dark. The fluorescence signal was measured at λ_ex = 530 nm and λ_em = 590 nm, using CLARIOstar Plus (RRID:SCR_026330).

Native chromatin immunoprecipitation

Cells were seeded in 10 cm dishes (5 × 10⁵ cells per dish) for 24 hours and then treated with either mIDH1 inhibitors or DMSO control for 48 hours. The cells were washed with ice-cold PBS, scraped, and centrifuged to collect cell pellets. The cell pellet was resuspended in lysis buffer containing 1 mmol/L CaCl₂, 0.2% Triton X-100, and 50 mmol/L Tris-HCl (pH 7.6) on ice for 15 minutes; homogenized by passing through a 27G needle; and sonicated to shear the chromatin. Five microliter of DNA fragments were used for agarose gel analysis to verify the sonication results. The sheared chromatin supernatants were mixed with IgG control or the indicated antibodies (tri-methyl-histone 3 lysine 4 and mono-methyl-histone 3 lysine 4) at 4°C overnight on a rotary mixer. Protein A/G magnetic beads were then added to the complex solution and incubated at 4°C for 4 hours on a rotary mixer. The complex solution was placed on a magnetic stand, and the supernatant was removed. The magnetic beads were washed on a rotary mixer with the following buffers and times: RIPA buffer twice, RIPA buffer + 0.3 mol/L NaCl twice, LiCl buffer twice, TE buffer + 0.2% Triton X-100 twice, and TE buffer once. The magnetic beads were resuspended in 10% SDS/proteinase K/TE buffer and incubated overnight at 65°C. The samples were vortexed briefly and placed on a magnetic stand, and the supernatants were collected. The magnetic beads were washed with 0.5 mol/L NaCl/TE buffer and placed on a magnetic stand, and the supernatants were collected. The second collected supernatant was combined with the first collected solution and stored at −20°C until DNA purification.

DNA fragment purification

DNA fragments were purified from the collected solutions using a GenepHlow Gel/PCR Kit (Geneaid, DFH300), according to the manufacturer’s instructions. The collected solutions were mixed with gel/PCR buffer, transferred to collection tubes, and centrifuged at 16,000 × g for 30 seconds. The collection tubes were washed with ethanol/wash buffer and centrifuged at 16,000 × g for 30 seconds. The remaining ethanol was removed by centrifuging the collection tubes at 16,000 × g for 3 minutes. The purified DNA was eluted in 70°C-heated elution buffer and stored at −20°C until real-time PCR.

Cell-cycle analysis

Cells were seeded in 10 cm dishes (5 × 10⁵ cells per dish) for 24 hours and then treated with MDM2 inhibitors or DMSO control for one doubling time. The cells were harvested and washed with 2% FBS/PBS. The cells were fixed in 70% ethanol at 4°C for 1 hour and then stored at −20°C overnight. The fixed cells were washed with cold PBS and stained with propidium iodide/RNase/Triton X-100 solution. FL-2 intensity was detected using the BD FACSCalibur Flow Cytometry System (RRID:SCR_000401) and analyzed using ModFit LT (RRID:SCR_016106).

Proliferation assay

Cells were seeded in 24-well plates (3 × 10³ cells/well). For single-drug treatment, increasing doses of mIDH1 inhibitors, MDM2 inhibitors, or DMSO controls were added. For combined treatment, mIDH1 and MDM2 inhibitors were added at different ratios. Cells were harvested at 3 doubling times for each cell line. Cells were stained with a 0.5% methylene blue ethanol solution for 2 hours, washed with flowing tap water, and air-dried overnight. Methylene blue-stained cells were dissolved in a 1% sarkosyl/PBS buffer, and the absorbance was read at 595 nm using a microplate reader. The data were normalized to those of DMSO controls. The combined index was calculated by adjusting the effective concentration of each drug combination. The data were analyzed and visualized using SynergyFinder 3.0 (22).

TUNEL assay

To detect apoptotic cells, the TUNEL assay kit (MedChemExpress, HY-K1078, HY-K1079) was used following the manufacturer’s protocol. Cells were seeded in eight-well chamber slides (5 × 10² cells per well) for 24 hours and then treated with MDM2 inhibitors for one doubling time. After incubation, cells were washed twice with PBS for 5 minutes and then fixed in fixation solution containing 4% paraformaldehyde in PBS, pH 7.4, for 30 minutes at 4°C. After fixation, cells were washed twice with PBS for 5 minutes and incubated with permeabilization solution containing 0.3% Triton X-100 in PBS, pH 7.4, for 5 minutes at room temperature. Cells were washed twice with PBS for 5 minutes and incubated with 50 μL of TUNEL working solution containing TdT enzyme and FITC-12-dUTP labeling mix at 37°C for 60 minutes in the dark. After staining, cells were washed 3 times with PBS, added mounting medium with DAPI to the sample, and then detected the fluorescence signal by fluorescence microscopy.

Antibodies and reagents

Antibodies used in this article are as follows: p53 (7F5) Rabbit mAb (Cell Signaling Technology cat. #2527, RRID:AB_10695803), Mic-1 (D2A3) Rabbit mAb (Cell Signaling Technology cat. #8479, RRID:AB_11129236), MDM2 antibody [SMP14] (GeneTex cat. # GTX70278, RRID:AB_372294), Anti-Glyceraldehyde-3-Phosphate Dehydrogenase Antibody, clone 6C5 (Millipore cat. # MAB374, RRID:AB_2107445), Rabbit (DA1E) mAb IgG XP Isotype Control (Cell Signaling Technology cat. # 3900, RRID:AB_1550038), Mono-Methyl-Histone H3 (Lys4) (D1A9) XP Rabbit mAb (Cell Signaling Technology cat. # 5326, RRID:AB_10695148), Tri-Methyl-Histone H3 (Lys4) (C42D8) Rabbit mAb (Cell Signaling Technology cat. # 9751, RRID:AB_2616028), JARID1A (D28B10) XP Rabbit mAb (Cell Signaling Technology cat. # 3876, RRID:AB_2129055), JARID1B (E2X6N) Rabbit mAb (Cell Signaling Technology cat. # 15327, RRID:AB_2798737), JARID1C (D29B9) Rabbit mAb (Cell Signaling Technology cat. # 5361, RRID:AB_10706165), JARID1D (E4D4B) Rabbit Monoclonal Antibody ((Cell Signaling Technology cat. # 78995, RRID:AB_3740212)), alpha Tubulin antibody [GT114] (GeneTex cat. # GTX628802, RRID:AB_2716636). Reagents and chemicals used in this study were DMSO (Sigma-Aldrich, D2650), ivosidenib (synonym: AG-120; MedChemExpress, HY-18767), IDH-305 (MedChemExpress, HY-104036), idasanutlin (synonym: RG7388; MedChemExpress, HY-15676), siremadlin (synonyms: NVP-HDM201; HDM201; MedChemExpress, HY-18658), KDOAM-25 citrate (MedChemExpress, HY-102047B), and MG-132 (synonyms: Z-Leu-Leu-Leu-al; MG132; MedChemExpress, HY-13259).

Statistical analysis

RStudio (RRID:SCR_000432), R Project for Statistical Computing version 4.2 (RRID:SCR_001905), and the R packages ggplot2 (RRID:SCR_014601) and ggsignif (RRID:SCR_023047) were used to analyze and visualize the data. Paired t tests and Wilcoxon signed-rank tests were used (PsyArXiv 7awm6_v1; refs. 23, 24).

Results

Mutations in IDH1/2 and TP53 are mutually exclusive in human iCCA tissues and cell lines

To identify other potential therapeutic targets in iCCA with IDH1/2 mutations, we first reviewed the comprehensive genomic profiling of BTC, in which IDH1/2 mutations were only identified in iCCA and were mutually exclusive with other major genetic alterations, namely TP53, KRAS, and SMAD4 mutations; actionable FGFR2 fusion and BRAF^V600E mutation; and ERBB2 and MDM2 amplifications (25, 26). With the emergence of wtTP53 reactivation therapy, we examined the mutual exclusion of IDH1/2^mut and TP53 alterations. Of the 1,031 iCCA cases included from seven studies in the cBioPortal database, IDH1/2 mutations were detected in 146 (14.1%; Fig. 1A; refs. 19–21). The TP53 mutation (TP53^mut) rate was 6.1% (9/146) in IDH1/2^mut and 29.9% (265/885) in IDH1/2^wt tumors, with an Odds ratio (OR) of 0.15 (95% confidence interval, 0.08–0.31), P < 0.0001 (Fisher’s exact test). Trends of mutual exclusivity between IDH1/2 and TP53 mutations from individual studies with case numbers greater than 100, three within cBioPortal and four other recent publications, are shown in Fig. 1B (27–32). Interestingly, both iCCA cell lines with pathogenic IDH1^mut (SNU1079 and RBE) in the Cancer Cell Encyclopedia were TP53^wt further supporting our observation (Supplementary Fig. S1A; ref. 33).

Figure 1.

IDH1/2 mutations and TP53 mutations are mutually exclusive in iCCA samples. A, Genetic alterations of IDH1, IDH2, TP53, KRAS, SMAD4, FGFR2, BRAF, ERBB2, and MDM2 in cBioPortal iCCA datasets. Each column represents one individual’s mutation profiling data. The color in each column represents different genetic alterations. Dark brown represents in-frame mutation (putative driver), light brown represents in-frame mutation (unknown significance), dark green represents missense mutation (putative driver), light green represents missense mutation (unknown significance), orange represents splice mutation (putative driver), black represents truncated mutation (putative driver), gray represents truncated mutation (unknown significance), purple represents structural variant (putative driver), red represents amplification, and blue represents deep deletion. B, Odds ratios and 95% confidence interval (CI) of IDH1/2 and TP53 mutations in six iCCA mutation profiling datasets. The P value was evaluated by the Fisher exact test.

Mutant IDH1 upregulates the expression of ubiquitin ligase MDM2 at the transcription level

To elucidate the functions and utilities of molecules and signaling pathways involved in IDH1^mut/TP53^wt iCCA, we compared the expression profiles of IDH1^mut/TP53^wt and IDH1^wt/TP53^mut iCCA cells from the Cancer Cell Line Encyclopedia (CCLE) mRNA dataset and found the upregulation of the E3 ubiquitin ligase MDM2, which ubiquitinates and negatively modulates TP53 protein expression in IDH1^mut/TP53^wt iCCA cells (Supplementary Fig. S2A; refs. 33, 34). The augmentation of MDM2 in IDH1^mut iCCA cells was confirmed in iCCA tissues and cell lines (Fig. 2A–C). Real-time PCR and Western blotting confirmed the increases in MDM2 mRNA and protein levels in both IDH1^mut iCCA cell lines (SNU1079 and RBE; Fig. 2B and C). To validate the association between mIDH1 and MDM2 expression, IDH1^mut iCCA cells were treated with two mIDH1-specific inhibitors, AG120 and IDH305. Pharmacologic inhibition of mIDH1 led to a reduction in intracellular 2-HG levels, accompanied by a decrease in MDM2 mRNA and protein expression (Fig. 2D–F). As expected, attenuation of MDM2 was accompanied by an increase in wtTP53 protein levels (Fig. 2F). In contrast, mIDH1 inhibitor treatment did not alter MDM2 and mTP53 expression in IDH1^wt iCCA cell lines (HuH28, HuCCT1, and SSP25; Supplementary Fig. S2B and S2C). These data suggested that 2-HG upregulates the expression of MDM2 at the transcriptional level in IDH1^mut iCCA cells.

Figure 2.

Mutant IDH1 upregulates MDM2 mRNA and protein expression. A, iCCA tissue sections were stained with hematoxylin and eosin (H&E) or MDM2 IHC. The stained sections were divided into three groups: (i) IDH1/2 wild-type (IDH11/2^wt), (ii) IDH1/2 mutation (IDH1/2^mut), and (iii) IDH1/2 wild-type, MDM2/4 amplification (IDH11/2^wt, MDM2/4^AMP). Statistical significance was evaluated by the Wilcoxon rank sum test (B) MDM2 mRNA expression in five iCCA cell lines. The 2^ΔΔCT value was normalized to HuH28. C, MDM2 and TP53 protein expression in five iCCA cell lines. GAPDH was used as an internal control. D, Intracellular 2-HG levels of two IDH1^mut iCCA cell lines after DMSO or mIDH1 inhibitors treatment for 48 hours. Fluorescent values of 2-HG were normalized to the DMSO control. E,MDM2 mRNA expression of two IDH1^mut iCCA cell lines after DMSO or mIDH1 inhibitors treatment for 48 hours. The 2^ΔΔCT value was normalized to the DMSO control. F, MDM2 and TP53 protein expression in two IDH1^mut iCCA cell lines after DMSO or mIDH1 inhibitor treatment for 72 hours. α-Tubulin was used as an internal control. MG-132 was used to stop proteasome degradation before harvest. IDH1 mutation iCCA cell lines: RBE and SNU1079. IDH1 wild-type iCCA cell lines: HuCCT1, HuH28, and SSP25. mIDH1 inhibitors: AG-120 and IDH-305. All data are represented as mean ± SD. All experiments were performed and repeated in three independent biological replicates. Statistical significance was evaluated by paired t test.

Mutant IDH1 suppresses KDM5 activity and enhances H3K4 methylation levels at the MDM2 promoter region

2-HG inhibits the activity of KDMs, a group of α-KGDs that modulate histone lysine methylation and gene transcription (35, 36). We first searched the histone chromatin immunoprecipitation (ChIP) sequencing data from ENCODE on the UCSC Genome Browser to explore the molecular mechanisms underlying the epigenetic modulation of MDM2 transcription in IDH1^mut iCCA cells (37–39). The histone-3-lysine-4 tri-methylation (H3K4me3), an active histone code, was highly enriched in three different regions near the promoter region and transcription start site (TSS) of MDM2 in pan-cell types (Fig. 3A; refs. 40, 41). ChIP-qPCR was performed to detect H3K4 mono-methylation (H3K4me1) and H3K4me3 levels in three H3K4me3-enriched promoter and TSS regions of MDM2 in IDH1^mut and IDH1^wt iCCA cells. Among them, the methylation score (H3K4me3/H3K4me1 ratio) of region 2 was the highest in the two IDH1^mut iCCA cells compared with regions 1 and 3 within the same cells or region 2 of the IDH1^wt iCCA cells (Fig. 3B). AG120 and IDH305 treatment profoundly reduced the H3K4 methylation level and H3K4me3/H3K4me1 ratio in region 2 of both IDH1^mut iCCA cells (Fig. 3C).

Figure 3.

Mutant IDH1 enhanced H3K4 methylation levels at the MDM2 promoter. A, H3K4 tri-methylation and mono-methylation signals at the MDM2 promoter region. Histone-seq data were obtained from the ENCODE project and visualized on the UCSC Genome Browser. B, Ratio of mono-methyl-H3K4 and tri-methyl-H3K4 at the MDM2 promoter region in five iCCA cell lines. C, H3K4 methylation level at the MDM2 promoter region after DMSO or mIDH1 inhibitor treatment for 48 hours in two IDH1^mut iCCA cell lines. IDH1 mutation iCCA cell lines: RBE and SNU1079. IDH1 wild-type iCCA cell lines: HuCCT1, HuH28, and SSP25. mIDH1 inhibitors: AG-120 and IDH-305. All data are represented as mean ± SD. All experiments were performed and repeated in three independent biological replicates. Statistical significance was evaluated by paired t test: NS, not significant.

The KDM5 family (KDM5A–5D) specifically recognizes and demethylates H3K4me3 via α-KGD oxidation, which is antagonized by 2-HG (35). We hypothesized that high H3K4 methylation at the MDM2 promoter region in IDH1^mut iCCA cells is caused by low KDM5 activity. We measured total KDM5 activity in the nuclear extract of iCCA cell lines and found a significant decrease in IDH1^mut iCCA cells compared with IDH1^wt iCCA cells (Fig. 4A). Treatment with AG120 and IDH305 enhanced total KDM5 activity and reduced the H3K4me3 level, which was accompanied by a reduction in MDM2 mRNA expression in IDH1^mut iCCA cells (Fig. 4B–E). Cotreatment with KDOAM-25 citrate, a pan-KDM5 inhibitor, restored H3K4me3 and MDM2 levels in AG120- and IDH305-treated IDH1^mut iCCA cells (Fig. 4C–E). These results suggested that mIDH1-derived 2-HG enhances H3K4me3 levels at the MDM2 promoter by suppressing KDM5 activity. To further evaluate which KDM5 protein family member is expressed in IDH1^mut iCCA cells, we screened the endogenous protein levels of KDM5A, KDM5B, KDM5C, and KDM5D across five iCCA cell lines. Surprisingly, KDM5B was the only member that was expressed in all five iCCA cell lines (Fig. 4F), indicating that KDM5B is the predominant KDM5 member involved in mIDH1-MDM2 modulation.

Figure 4.

Mutant IDH1 suppresses KDM5 activity to enhance H3K4 methylation level at the MDM2 promoter and MDM2 expression. A, Methyl-H3K4 demethylase KDM5 activity in five iCCA cell lines. Fluorescent values were normalized to corresponding protein concentration and incubation time. B, Methyl-H3K4 demethylase KDM5 activity in two IDH1^mut iCCA cell lines after DMSO or mIDH1 inhibitor treatment for 48 hours. Fluorescent value was first normalized to the corresponding protein concentration and incubation time. C, Global H3K4 methylation level in two IDH1^mut iCCA cell lines after DMSO, mIDH1 inhibitor, or KDM5s inhibitor treatment for 48 hours. Histone 3 was used as a histone control. D,MDM2 mRNA expression of two IDH1^mut  iCCA cell lines after DMSO, mIDH1 inhibitor, or KDM5s inhibitor treatment for 48 hours. The 2^ΔΔCT value was normalized to the DMSO control. E, MDM2 protein expression of two IDH1^mut  iCCA cell lines after DMSO, mIDH1 inhibitor, or KDM5s inhibitor treatment for 72 hours. GAPDH was used as an internal control. F, KDM5 family member protein expression in five iCCA cell lines. α-Tubulin was used as an internal control. IDH1 mutation iCCA cell lines: RBE and SNU1079. IDH1 wild-type iCCA cell lines: HuCCT1, HuH28, and SSP25. mIDH1 inhibitors: AG-120 and IDH-305. Pan methyl-H3K4 demethylases KDM5 inhibitors, KDOAM-25 citrate. All data are represented as mean ± SD. All experiments were performed and repeated in three independent biological replicates. Statistical significance was evaluated by paired t test.

MDM2 inhibitors restore wtTP53 function and suppress the growth of IDH1^mut iCCA cells

We demonstrated that IDH1 and TP53 mutations are mutually exclusive in iCCAs and that mIDH1 enhances MDM2 expression through the modulation of KDM5 activity by 2-HG in IDH1^mut iCCA cells. These findings suggest that IDH1^mut iCCA is a potential candidate for MDM2 inhibitor treatment, which is under extensive clinical evaluation for MDM2-amplified TP53^wt tumors, including BTCs (42, 43). We treated iCCA cells with different concentrations of MDM2 inhibitors (RG7388 and HDM201) or mIDH1 inhibitors (AG120 and IDH305) for three doubling times and assessed the the cell proliferation. Consistent with previous studies and clinical trials, mIDH1 inhibitors had little growth-inhibitory effect on iCCA cell lines. However, IDH1^mut/TP53^wt iCCA cells were more susceptible to MDM2 inhibitors than IDH1^wt/TP53^mut iCCA cells (Fig. 5A). To clarify the mechanism by which MDM2 inhibitors affect IDH1^mut iCCA cells, we treated SNU1079 and RBE cells with RG7388 and HDM201 for one doubling time and harvested the cells for analysis. Both inhibitors disrupted the interaction between MDM2 and TP53, inducing the accumulation of these two proteins in IDH1^mut/TP53^wt iCCA cells (Fig. 5B). In addition, the expression of macrophage inhibitory cytokine-1 (MIC-1), a TP53 target gene, was also increased (Fig. 5B; ref. 44). This led to cell-cycle arrest in the G₁-phase and apoptosis in the SNU1079 and RBE cells (Fig. 5C and D). Conversely, treatment with the MDM2 inhibitor did not significantly inhibit the growth or accumulation of MDM2 and mTP53 proteins in IDH1^wt/TP53^mut iCCA cells under the same experimental conditions (Fig. 5A and B). As mIDH1 inhibitors suppressed MDM2 expression, it would be interesting to explore the effect of MDM2 and mIDH1 inhibitor combination treatment in IDH1^mut/TP53^wt iCCA cells. We cotreated iCCA cell lines with various combinations of MDM2 and mIDH1 inhibitors and calculated the combination index value. The combination index estimation indicated that the MDM2 and mIDH1 inhibitors showed a synergistic effect (combination index < 1) in suppressing the growth of IDH1^mut/TP53^wt iCCA cell lines (Fig. 6A and B). These results suggest that MDM2 is a therapeutic target in IDH1^mut/TP53^wt iCCA cells.

Figure 5.

MDM2 inhibitors reactivate wtTP53 to suppress cell growth, arrest the cell cycle, and induce cell death in IDH1^mut iCCA cells. A, Cell viability after mIDH1 inhibitors or MDM2 inhibitors treatment for 3 doubling times in five iCCA cell lines. Data was normalized to the corresponding DMSO control. B, MDM2, TP53, and MIC1 protein expression in two IDH1^mut and three IDH1^wt iCCA cell lines after DMSO or MDM2 inhibitors treatment for 1 doubling time. MIC1 was used as an indicator of wtTP53 activation. GAPDH was used as an internal control. C, Cell-cycle analysis of two IDH1^mut iCCA cell lines after MDM2 inhibitors treatment for 1 doubling time. D, TUNEL staining in two IDH1^mut iCCA cell lines after MDM2 inhibitors treatment for 1 doubling time. Blue staining represents DAPI. Red staining represents TUNEL positive cells. The white arrows point to TUNEL-positive/DAPI colocalized cells. IDH1 mutation iCCA cell lines: RBE and SNU1079. IDH1 wild-type iCCA cell lines: HuCCT1, HuH28, and SSP25. mIDH1 inhibitors: AG-120 and IDH-305. MDM2 inhibitors: RG7388 and HDM201. All data are represented as mean ± SD. All experiments were performed and repeated in three independent biological replicates. Statistical significance was evaluated by paired t test: NS, not significant.

Figure 6.

Synergistic effect of MDM2 inhibitors and mIDH1 inhibitors combination treatment in two IDH1^mut iCCA cell lines. A, Synergistic effect of MDM2 inhibitors and mIDH1 inhibitors combination treatment in SNU1079. Cells were treated with different combined doses of MDM2 inhibitors and mIDH1 inhibitors for 3 doubling times. Data was first normalized to DMSO control and then calculated for cytotoxicity and the highest single agent  (HSA) synergy score. B, Synergistic effect of MDM2 inhibitors and mIDH1 inhibitors combination treatment in RBE. Cells were treated with different combined doses of MDM2 inhibitors and mIDH1 inhibitors for 3 doubling times. Data were first normalized to DMSO control and then calculated for cytotoxicity and HSA synergy score. C, A working model shows the mechanism by which mIDH1 activates MDM2 expression to promote TP53 degradation and proliferation in iCCA cells. IDH1 mutation iCCA cell lines: RBE and SNU1079. mIDH1 inhibitors: AG-120 and IDH-305. MDM2 inhibitors: RG7388 and HDM201. All data are represented as mean ± SD. All experiments were performed and repeated in three independent biological replicates.

Discussion

Large-scale next-generation sequencing has contributed to the discovery of pathogenic somatic mutations in cancer genomes. Comprehensive mutation profiling provides a new perspective for studying interactions between gene alterations. Our earlier iCCA genomic profiling dataset analyses revealed mutual exclusion of IDH1/2 and TP53 mutations, and CCLE mRNA dataset analysis showed IDH1^mut/TP53^wt cells expressing higher levels of MDM2 than IDH1^wt/TP53^mut cells. Our laboratory work further demonstrated that both IDH1^mut/TP53^wt iCCA (SNU1079 and RBE) cell lines express higher MDM2 protein levels than IDH1^wt/TP53^mut iCCA cells. Mutant IDH1 inhibitor treatment reduced the intracellular level of 2-HG, which was accompanied by a reduction in MDM2 mRNA and protein expression, and enhanced wtTP53 protein expression in both IDH1^mut/TP53^wt iCCA cells. As MDM2 is a known TP53 negative regulator that binds TP53 to facilitate its degradation, we further explored the molecular mechanisms underlying the modulation of MDM2 transcription by 2-HG, which is a mIDH1/2-derived oncometabolite known to reprogram gene expression and cellular metabolism through suppressing α-KGD activity (34, 45). KDM5 histone lysine demethylase is a target enzyme of 2-HG (35). Our study showed that KDM5 activity was lower and the level of H3K4me3 in part of the promoter region of MDM2 was higher in IDH1^mut iCCA cells than in IDH1^wt iCCA cells. All the events—reduced KDM5 activity, increased H3K4me3 at the MDM2 promoter region, enhanced MDM2 expression, and reduced wtTP53 expression in IDH1^mut iCCA cells—could be partially reversed by mIDH1 inhibitor treatment. These findings indicate that wtTP53 inactivation plays a role in the carcinogenesis of IDH1^mut iCCA. In addition, combining the findings of the mutual exclusion of IDH1/2 and TP53 mutations, and the enhanced MDM2 expression in IDH1^mut iCCA tissues and cells also provides a rationale to evaluate the potential use of MDM2 inhibitors in this subgroup of patients.

MDM2 is an E3 ubiquitin ligase that directly targets wtTP53 to promote TP53 degradation through monoubiquitination, polyubiquitination, NEDD8 NEDDylation, and SUMO-1 modification (34). These post-translational modifications govern TP53 cellular localization, activity, downstream target gene selection, and protein stability. Based on structural studies of the MDM2-TP53 binding motif (Phe19, Trp23, and Leu26), small-molecule inhibitors targeting MDM2 were designed and developed (46, 47). Currently, the targeted population for MDM2 inhibitor therapy in solid tumors mainly focuses on those with frequent MDM2 amplification/TP53^wt genotype, and encouraging therapeutic efficacy has been observed in patients with advanced BTCs (48). In this study, we used two MDM2 inhibitors that had undergone clinical trials to test the cytotoxic effect of IDH1^mut iCCA (49, 50). These two MDM2 inhibitors showed more effective cytotoxicity and wtTP53 reactivation than the mIDH1 inhibitor in IDH1^mut iCCA cells. The finding is clinically relevant because it provides a rationale to expand the potential targeted iCCA population for MDM2 inhibitor treatment. According to the cBioportal dataset, MDM2 amplification and IDH1/2 mutations were detected in 3.5% (15/430) and 25.6% (109/430) of iCCA tissue, respectively. Of these, only two tumors had co-occurrence of IDH1 mutation and MDM2 amplification. Furthermore, MDM2 inhibitors may overcome the mIDH1 inhibitor resistance in IDH1^mut iCCA with acquired IDH2 mutation (15).

Moreover, cotreatment with mIDH1 and MDM2 inhibitors synergistically suppressed the proliferation of IDH1^mut iCCA cells. Although mIDH1 inhibitors did not directly target wtTP53 or MDM2, indirect epigenetic regulation of MDM2 still provides a new therapeutic strategy for wtTP53 reactivation. One of the most important obstacles to the development of MDM2 in cancer therapy is dose-dependent hematologic toxicity, notably thrombocytopenia (51). The finding of synergism between IDH1 inhibitors and MDM2 inhibitors provides a potential strategy for combination therapy aimed at reducing dose-dependent adverse events of MDM2 inhibitors without compromising therapeutic efficacy in IDH1^mut/TP53^wt iCCA. Moreover, this mutually exclusive phenomenon was also observed in IDH1/2^mut acute myeloid leukemia, revealing that MDM2 inhibitors or other agents that reactivate wtTP53 are a potential treatment strategy for IDH1/2^mut AML (Supplementary Fig. S3).

The KDM5 family comprises four proteins: KDM5A, KDM5B, KDM5C, and KDM5D. Members of the KDM5 family catalyze the demethylation or oxidation of methyl groups from the activating histone codes H3K4me3 and H3K4me2, which facilitate the repressive H3K4me1 code (52). KDM5 protein members were the key targets inhibited by 2-HG (35). IDH1 mutation or cell-permeable 2-HG enhanced H3K4me3 levels by inhibiting KDM5A/C/D, promoted cell viability, and contributed to transformation in AML and glioma (35). We demonstrated that KDM5A/C/D were downregulated in IDH1^mut iCCA, whereas KDM5B was upregulated, suggesting that altered expression of the KDM5 family occurs in the transformation and carcinogenesis of iCCA. Pan-KDM5 inhibitors restored both H3K4 demethylation levels and MDM2 suppression after treatment with mIDH1 inhibitors, suggesting that KDM5 plays a role in the regulation between IDH1 and MDM2.

In summary, we confirmed that mIDH1-derived 2HG enhances MDM2 transcription by decreasing KDM5 activity and enriching H3K4me3 at the MDM2 promoter region. Increased MDM2 promotes the degradation of the wtTP53 protein in IDH1^mut iCCA, and pharmacologic inhibition of MDM2 reduces wtTP53 degradation and reactivates wtTP53-triggered growth inhibition and cell death in IDH1^mut iCCA (Fig. 6C). We identified a novel mIDH1–MDM2–wtTP53 axis that mIDH1 indirectly regulates wtTP53 protein expression through epigenetic upregulation of MDM2 transcription. This renders a wtTP53 reactivation therapeutic option for IDH1^mut iCCA.

Supplementary Material

Supplementary Figure 1

Cell lines used in this study.

Supplementary Figure 2

MDM2 expression after mutant IDH1 inhibitor treatment in IDH1WT iCCA cells.

Supplementary Figure 3

Mutation profile of IDH1, IDH2 and TP53 in AML

Data Availability

The data were generated by the first author and are available from the corresponding author upon request. Mutation profiling datasets analyzed in this study are publicly available in the following online datasets and studies (19–21, 27–32, 53–57).

Authors’ Disclosures

C.-T. Liu reports grants from the National Science and Technology Council, Executive Yuan, Taiwan, and the National Institute of Cancer Research, National Health Research Institutes, Taiwan, during the conduct of the study. L.-T. Chen reports grants from the National Science and Technology Council and grants from the National Health Research Institute during the conduct of the study as well as personal fees and nonfinancial support from ACTgenomics, TTY, and ONO; grants and personal fees from Novartis; grants, personal fees, and nonfinancial support from Ipsen; grants from Pfizer; personal fees from MSD, AstraZeneca, Revolution Medicine, Bristol Myers Squibb, Astellas, AbbVie, Taivex, and AcadeMab Bio; personal fees and other from Onward; and grants and nonfinancial support from SynCoreBio outside the submitted work and a patent to Anti–ENO-1 monoclonal antibody issued and licensed and a patent to Anti-CXCR2 monoclonal antibody issued. No disclosures were reported by the other authors.

Authors’ Contributions

C.-T. Liu: Data curation, formal analysis, investigation, validation, visualization, writing–original draft, writing–review and editing. Y.-Y. Su: Resources, data curation, investigation. N.-J. Chiang: Resources, data curation, investigation. C.-F. Li: Resources, formal analysis. Y.-C. Ma: Resources, formal analysis. K.-C. Chang: formal analysis. Y.-H. Hung: Conceptualization, project administration, supervision. W.-C. Hung: Conceptualization, resources, supervision, funding acquisition, writing–original draft, writing–review and editing. L.-T. Chen: Conceptualization, resources, supervision, funding acquisition, writing–original draft, writing–review and editing.