Sorafenib, an oral multi-kinase inhibitor, is a first-generation targeted therapy that has demonstrated survival benefits in patients with advanced HCC (1, 2). Despite recent advances in immunotherapies for HCC, sorafenib monotherapy remains the most effective single drug (3). However, the majority of patients develop drug resistance within approximately 6 months of starting treatment (4). Thus, combination treatment with different drugs is considered to be a potential approach to improving the antitumor effects of sorafenib (5, 6).

Ferroptosis is an iron-dependent form of cell death characterized by increased production of reactive oxygen species (ROS) via the Fenton chemistry, as well as accumulation of lipid peroxidation products (7). Ferroptosis is induced by inhibiting the cystine-glutamate antiporters Xc- or xCT, which import cystine for de novo glutathione synthesis. Accumulating evidence shows that sorafenib induces ferroptosis by inhibiting cysteine uptake; therefore, pathways that mitigate sorafenib-induced ferroptosis induce drug resistance (8, 9). Recently, the challenges of inducing ferroptosis have attracted attention as a potent therapeutic strategy to reverse cancer resistance to sorafenib.

The orphan nuclear receptor estrogen-related receptor (ERR)γ plays an important role in mitochondrial metabolism and redox balance in tissues that have high metabolic demand, including heart, skeletal muscle, liver, and brown adipose tissue (10). ERRγ is also closely related to progression of breast cancer and anaplastic thyroid cancer (11, 12). Singh et al. suggested the potential clinical applicability of ERRγ inhibition by DN200434, an orally bioavailable and highly selective ERRγ inverse agonist, for the treatment of anaplastic thyroid cancer (12). Previously, we reported that ERRγ has clinical significance with respect to the diagnosis and treatment of HCC (13). However, the role of ERRγ as a modulator of ferroptosis in sorafenib resistance remains unclear.

Here, we examined whether ERRγ is involved in sorafenib resistance of HCC, and whether the ERRγ inverse agonist DN200434 induces ferroptosis and re-sensitizes sorafenib-resistant HCC to sorafenib.

Sorafenib-resistant HCC cells overexpress ERRγ and fail to induce ferroptosis

Sorafenib-resistant Huh7 and SK-Hep1 cells were established as previously described (14). As shown in Fig. 1A, sorafenib-resistant Huh7 and SK-Hep1 cells became less sensitive to sorafenib over time, or when exposed to different concentrations for 24 h. To determine whether sorafenib induces ferroptosis in HCC, we monitored elevated lipid peroxidation using C11-BODIPY. As expected, treatment with sorafenib increased lipid peroxidation in sorafenib-sensitive HCC cells but not in sorafenib-resistant HCC cells (Fig. 1B). The publicly available gene dataset GSE73571 was extracted from the Gene Expression Omnibus (GEO) database to obtain data on the phenotypes of four sorafenib-acquired resistant and three sorafenib-sensitive HCC cells. Analysis of this dataset showed that sorafenib treatment significantly increases ERRγ levels in HCC cells (Fig. 1C). Also, the data suggest that ERRγ levels are significantly higher in sorafenib-resistant HCC cells than in sorafenib-sensitive HCC cells, suggesting that ERRγ is involved in sorafenib resistance (Fig. 1D, E).

DN200434 enhances sorafenib-induced ferroptosis in sorafenib-resistant HCC cells

Given the upregulation of ERRγ and the failure of ferroptosis induction in sorafenib-resistant HCC cells, we next explored whether inhibiting ERRγ sensitizes sorafenib-resistant HCC cells through recovery of ferroptosis. To this end, we first confirmed that ERRγ protein levels in sorafenib-resistant HCC cells were downregulated by DN200434, which is already known to significantly decrease endogenous ERRγ expression in CAL62 cells (12) (Fig. 2A). Next, we assessed accumulation of lipid ROS in sorafenib-resistant HCC cells in the presence/absence of DN 200434. We found that sorafenib or DN200434 alone did not increase accumulation of lipid ROS in sorafenib-resistant HCC cells, whereas a combination of sorafenib and DN200434 induced ferroptosis, as evidenced by increased lipid peroxidation and intensity of 4-hydrxoynoneal (4-HNE) positive staining (Fig. 2B-E). In parallel with ferroptosis induction, sorafenib-resistant HCC cell death assessed by the LDH release assay was increased more by combined treatment with sorafenib and DN200434 than by sorafenib or DN200434 treatment alone; this was prevented by the presence of ferrostatin-1 (Fig. 2F). Given that sorafenib or DN200434 alone did not increase ferroptosis or cell death in sorafenib-resistant HCC cells, these findings indicate that, rather than by acting synergistically with sorafenib in sorafenib-resistant HCC cells, DN200434 renders sorafenib-resistant HCC cells more susceptible to ferroptosis.

DN200434 increases mitochondrial ROS, and downregulates mTORC1 activity and GPX4 levels

ERRγ is involved in mitochondrial homeostasis (15); therefore, we investigated whether induction of DN200434-mediated ferroptosis was associated with mitochondrial ROS production. As shown in Fig. 3A, B, sorafenib or DN200434 alone did not increase the amount of mitochondrial-derived superoxide stained by MitoSOX; however, the combination of sorafenib and DN200434 increased mitochondrial ROS significantly in sorafenib-resistant HCC cells (Fig. 3A, B). Based on previous results showing a higher glutathione (GSH)/glutathione disulfide (GSSG) ratio in sorafenib-resistant HCC cells (14), we measured the GSH/GSSG ratio in sorafenib-resistant cells in the absence/presence of DN200434. As shown in Fig. 3C, sorafenib or DN200434 alone did not alter the GSH/GSSG ratio; however, the combination of DN200434 and sorafenib decreased the GSH/GSSG ratio in sorafenib-resistant HCC cells significantly, compared with sorafenib or DN200434 treatment alone (Fig. 3C). Glutathione peroxidase-4 (GPX4) is the key upstream regulator of ferroptosis (7). Given that mitochondrial ROS reduces mTORC1 (16, 17), and that mTORC1 protects against ferroptosis by promoting GPX4 protein synthesis (18), we investigated the effects of DN200434 on mTORC1 activity and GPX4 levels in sorafenib-resistant HCC cells. Co-treatment of sorafenib-resistant HCC cells with DN200434 and sorafenib decreased mTORC1 activity, as measured by detection of phosphorylated mTOR (Ser-2448) and 4E binding protein 1(4E-BP1), and GPX4 levels, whereas this was not observed in the group treated with sorafenib or DN200434 alone (Fig. 3D). Notably, supplementation with GSH reversed the increase in mitochondrial ROS and decrease in mTORC1 activity and GPX4 levels induced by combined treatment of sorafenib-resistant HCC cells with sorafenib and DN200434 (Fig. 3A, B, D). In accordance with these results, and as shown in Fig. 2F, the increase in HCC cell death in the presence of DN200434 and sorafenib was reversed significantly when ferroptosis was inhibited by GSH. Collectively, these findings indicate that DN200434 potentiates ferroptosis via mitochondrial ROS-induced inhibition of mTORC1 and GPX4, thereby improving the response of sorafenib-resistant HCC cells to sorafenib.

DN200434 improves the response to sorafenib in sorafenib-resistant HCC tumor

Finally, we examined the effect of DN200434 on tumor growth in nude mice bearing subcutaneous sorafenib-resistant Huh7 and SK-Hep1 xenografts. Tumors were significantly larger in mice injected with Huh7 SR than in mice injected with Huh7 SS (Fig. 4A). The reduction in tumor growth in mice treated with sorafenib plus DN200434 was greater than that in mice treated with sorafenib alone, with no difference in body weight between the groups (Fig. 4B-D). The efficiency of combined treatment with sorafenib and DN200434 on tumor growth was confirmed in nude mice bearing sorafenib-resistant SK-Hep1 cells (Fig. 4E-H).

Here, we showed that ERRγ is significantly upregulated in sorafenib-resistant HCC cells, and that inhibition of ERRγ by DN200434 increases their susceptibility to sorafenib. Co-treatment with DN200434 and sorafenib significantly increased mitochondrial ROS, while at the same time decreasing mTORC1 activity and GPX4 levels, thereby potentiating sorafenib-induced ferroptosis in HCC cells. The efficacy of combined treatment with DN200434 and sorafenib was confirmed in xenograft models.

Previously, we reported that the degree of ERRγ expression in surgically resected human HCC tissues is associated with advanced clinical stage and invasive pathologic grade (13), and provided a rationale for ERRγ as a possible therapeutic target for HCC by showing that an ERRγ inverse agonist-induced ROS generation and suppressed HCC cell proliferation (13). Here, analysis of the data from sorafenib-resistant HCC cells and the Gene expression Omnibus (GEO) database revealed that ERRγ is also overexpressed in sorafenib resistance in HCC cells and tissues, suggesting that ERRγ plays a role in the development of sorafenib resistance. During oxidative remodeling, ERRγ plays an essential role in supporting mitochondrial metabolic fitness by activating a transcriptional network that regulates mitochondrial biogenesis, oxidative phosphorylation, and the electron transport chain in various cellular processes (15). Tissue-specific deletion of ERRγ results in a reduced oxygen consumption rate and elevated mitochondrial ROS production (19, 20). Consistent with these findings, we showed herein that inhibition of ERRγ by DN200434 increased mitochondrial ROS, supporting the hypothesis that ERRγ is implicated in drug resistance through impaired mitochondrial function.

Growing evidence shows that changes in cellular metabolism, redox homeostasis, and various signaling pathways are involved in regulating cell sensitivity to ferroptosis (18, 21). Recent findings demonstrate that mTORC1, a central regulator of physiology and cell growth, is a regulator of ferroptosis (18, 22). Inactivation of mTORC1 by PI3K/AKT inhibitors or mTOR inhibitors suppresses synthesis of GPX4 protein, which plays a crucial role in protection against ferroptosis, via downstream 4EBP; this triggers autophagy and sensitizes cells to lipid peroxidation and ferroptosis (22, 23). The work presented herein shows that mTORC1 activity in sorafenib-resistant HCC cells was decreased by combined treatment with DN200434 and sorafenib, which is consistent with previous reports showing that mitochondrial ROS reduce mTORC1 (16, 17). Notably, co-treatment with DN200434 and sorafenib suppressed mTORC1 activity and GPX4 levels. When combined with results showing that co-treatment with DN200434 and sorafenib leads to significant tumor regression in vitro and in vivo, these findings indicate that DN200434 improves the response to sorafenib by enhancing ferroptosis.

In conclusion, we showed that ERRγ contributes to the resistance to sorafenib, and that inhibition of ERRγ with DN200434 recovers sorafenib sensitivity by inducing ferroptosis. Our findings further emphasize that DN200434 may be a useful adjuvant agent in new treatment protocols aimed at increasing the susceptibility of HCC to sorafenib.

Cell culture and chemical treatments

Human HCC Huh7 and SK-Hep1 cells were obtained from the American Type Culture Collection (Manassas, VA, USA). To generate sorafenib-resistant HCC cells (Huh7 (SR) and SK-Hep1 (SR)), Huh7 and SK-Hep1 cells were cultured in the presence of increasing concentrations (up to a maximum concentration of 10 μM) of sorafenib (Cayman Chemical, Ann Arbor, MI, USA) for 8 months, as previously described (14 Sorafenib-resistant HCC cells (Huh7 (SR) and SK-Hep1 (SR) cells) and their parenteral cells (Huh7 and SK-Hep1 cells) were cultured in Dulbecco’s modified Eagle’s medium (Biological Industries, Kibbutz Beit Haemek, Israel) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. Sorafenib-resistant HCC cells and their parenteral cells were seeded at 3 − 5 × 10³ cells per well in 96-well plates. The cells were serum-starved for 24 h and incubated with 10% FBS with or without sorafenib for 24 h, and cell viability was measured using CCK8 Solution Reagent (CK04; Dojindo). The absorbance of each well at 495 nm was measured on a VERSA MAX ELISA reader (Molecular Devices). The proportion of viable cells in each treatment group was normalized against that in control wells. To determine the effects of ferroptosis on HCC cells, cells were treated with 10 μM sorafenib (Cayman), 12 μM DN200434, 1 μM ferrostatin-1 (Sigma, St. Louis, MO, USA), or 2 mM glutathione reduced-ethyl ester (Sigma) for 24 h.

Detection of lipid ROS

For fluorescence detection of lipid ROS, cells were stained for 30 minutes with C11-BODIPY (10 μM; Thermo Fisher Scientific) and washed three times with PBS, and the nuclei were stained using NucBlueTM Live ReadyProbesTM Reagent (Thermo Fisher Scientific). Data were presented as the ratio of the intensity of green/red fluorescence in cells. For 4-HNE staining, cells on glass slides were fixed for 15 min in 4% paraformaldehyde and then permeabilized with 0.3% Triton X-100 for 10 min at room temperature. Following 1 h of blocking in 5% normal goat serum (Vector Laboratories, Burlingame, USA), cells were incubated overnight at 4°C with a primary anti-4-HNE (Abcam, 1:100) antibody. After washing, the cells were incubated for 1 h at room temperature with an Alexa Fluor^TM 568 goat-anti-rabbit (Thermo Fisher Scientific) secondary antibody. Nuclei were stained with DAPI (Vector Laboratories). Fluorescence intensity was determined by obtaining the integrated signal density/cell.

Lactate dehydrogenase release assay

Cells were seeded at 1 × 10⁴/well into a 96-well plates and incubated at the indicated concentrations in cell culture medium for 24 h at 37°C. LDH release into the medium was detected using the LDH-Glo^TM Cytotoxicity Assay (Promega, Madison WI, USA), as recommended by the manufacturer. The release of LDH from untreated cells was used as a control.

Immunoblot analysis

Cells were incubated on ice for 30 min in IPH lysis buffer (50 mM Tris [pH 8.0], 150 mM NaCl, 5 mM EDTA, 0.1 mM phenylmethylsulphonyl fluoride [PMSF], and 0.5% NP-40) containing a protease inhibitor cocktail (Sigma) and dithiothreitol. Lysates were clarified by centrifugation at 12,000 × g for 10 min. Supernatants were collected, and the protein concentration was measured using the Bio-Rad protein assay (Bio-Rad, Richmond, CA, USA). Proteins in cell lysates were separated by SDS-PAGE and transferred to PVDF membranes (Millipore Corporation, Bedford, MA, USA). The membranes were incubated with primary antibodies specific for the following proteins: phospho-mTOR (S2448), mTOR, phospho-4E-BP1, 4E-BP1, GPX4 (Cell Signaling Technology), and actin (Sigma). After three washes in TBST, membranes were incubated with horseradish peroxidase-conjugated secondary antibodies (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Immunoreactive proteins were visualized by chemiluminescence (UVITec, Cambridge, UK).

Quantitative RT-PCR

Total RNA was prepared using the QIAzol lysis reagent (Qiagen, Hilden, Germany), and cDNA was synthesized using the RevertAid First Strand cDNA Synthesis kit (Thermo Fisher Scientific, Inc.). The resultant cDNA was amplified on a 7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). Gene expression was normalized to that of the corresponding level of β-actin mRNA. The primer sequences were as follows: ERRγ forward, CGATGCCCAAGAGACTGTGTT; reverse, AGACGCACCCCTTCTTTCAG.

GEO data processing

The gene expression profile dataset GSE73571 was downloaded from the Gene expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/gds/). R 3.6.3 (R Core Team; Auckland, New Zealand) was used for statistical analysis and to calculate fold-changes between sorafenib-sensitive and sorafenib-resistant tumors.

Measurement of the GSH/GSSG ratio

Cells were seeded at 1 × 10⁴/well in a 96-well plate and incubated at the indicated concentrations in cell culture medium for 24 h at 37°C. The experiments were repeated independently three times. Briefly, to measure the glutathione (GSH)/glutathione disulfide (GSSG) ratio, the GSH/GSSG-Glo^TM Assay (Promega) was performed according to the manufacturer’s protocol using 50 μl per well of total glutathione lysis reagent or oxidized glutathione lysis reagent. A vehicle control (cells incubated with DMSO) and a no-cell control (PBS) were also included. The glutathione concentrations (μM) were determined in the final reaction volume (200 μl), and the GSH/GSSG ratio was calculated using the following equation: Total GSH RLU × GSSG RLU)/GSSG RLU.

MitoSox

Mitochondrial ROS generation was assessed using the MitoSOX Red Mitochondrial Superoxide indicator (Invitrogen, Carlsbad, CA, USA). Huh7 (SR) and SK-Hep1 (SR) cells on a confocal dish were treated with 10 μM sorafenib in the presence/absence of 12 μM DN200434 for 24 h. Cells were then treated with 5 μM MitoSOX reagent working solution and incubated for 10 min at 37°C in the dark. The cells were then washed gently three times with warm HBSS buffer. Finally, cells were counterstained with NucBlue Live Cell Stain ReadyProbes (Invitrogen) and mounted in warm buffer for imaging. MitoSOX fluorescence intensity was quantified using ImageJ software.

Animal experiments

Huh7 (SR) or SK-Hep1 (SR) cells (6 × 10⁶), and Huh7 (SS) or SK-Hep1 (SS) cells (6 × 10⁶), were injected subcutaneously into the right flank of 6-week-old athymic male BALB/c nude (nu/nu) mice. Mice were randomized into three groups (n = 5 in the Huh7/Huh7 (SR) group and n = 4 in the SK-Hep1/SK-Hep1 (SR) group, respectively). When the tumor volume was around 100 mm³, mice received sorafenib (10 mg/kg via i.p. injection), DN200434 (10 mg/kg via oral gavage), or vehicle control (DMSO in PBS) every day for 15 days. Tumors were measured every 3 days using calipers and the volume was calculated as length × width² × 0.5 (mm³). All animal procedures were approved by the Institutional Animal Care and Use Committee of Kyungpook National University School of Medicine (KNU-2018-0015 and KNU-2021-0001).

Statistical analysis

Data were expressed as the mean ± standard error of the mean (SEM) of at least three independent experiments. Statistical difference was determined by Student’s t-test (Prism software); a P value < 0.05 was considered statistically significant.

This work was supported by National Research Foundation of Korea (NRF) grants NRF-2021R1A2C3005603, NRF-2017M3A 9G7073086, 2022R1A2C1008591, and NRF-2020R1A5A201 7323, funded by the Ministry of Science and the ICT; and by grant HI16C1501 from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute, funded by the Ministry of Health and Welfare.

The authors have no conflicting interests.