KEAP1 mutations activate the NRF2 pathway to drive cell growth and migration, and attenuate drug response in thyroid cancer

The KEAP1/NRF2 pathway, a major regulator of the cellular oxidative stress response, is frequently activated in human cancers. Often mediated by loss-of-function mutations in KEAP1, this activation causes increased NRF2 transcriptional activity and constitutive activation of the antioxidant response. While KEAP1 mutations have been well documented in various cancers, their presence and role in thyroid carcinoma have remained largely unexplored. In this study, we sequenced pediatric thyroid tumors and analyzed publicly available datasets, identifying 81 KEAP1 mutations in tumors across a range of histologies. In these tumors, we further identified frequent biallelic loss of KEAP1 via 19p13.2 loss of heterozygosity (LOH). MAPK-activating alterations were found in a subset of KEAP1-mutant cases, but they were mutually exclusive with 19p13.2 LOH. Transcriptome analysis also revealed significant activation of the NRF2 pathway in KEAP1-mutant tumors. Four additional cases with similar transcriptional profiles but lacking mutational data were identified, likely representing putative KEAP1 mutants. Using in vitro cell line models, we then profiled the functional consequences of KEAP1 knockout in cells with and without known driver alterations. In these models, we show that KEAP1 loss leads to an NRF2-dependent upregulation of AKR1C3, GCLC, NQO1, and TXNRD1, along with increased proliferation and migration irrespective of MAPK mutational status. We also demonstrate that loss of KEAP1 reduces sensitivity of RET fusion-positive cells to selpercatinib, consistent with previous reports that these alterations promote drug resistance in other malignancies. In this study, we comprehensively profile KEAP1 mutations in thyroid tumors, showing that they are more prevalent and functionally significant than previously recognized. These findings position KEAP1 mutations as novel oncogenic variants in thyroid cancer and support the integration of KEAP1/NRF2 pathway profiling into future studies and clinical frameworks.

1 Introduction

The kelch-like ECH associated protein 1 (KEAP1)/nuclear factor erythroid 2-related factor 2 (NRF2) pathway is the master regulator of cellular defense against chemical toxicity and oxidative stress. Under normal physiological conditions, reactive oxygen species (ROS) are produced and play critical regulatory roles in immune function, inflammation, cell division, and stress response (1). However, uncontrolled production of these oxidant species impairs metabolic function, causes toxicity, and contributes to the pathophysiology of disease, including tumorigenesis (2, 3). Accumulation of ROS is therefore counterbalanced by a highly regulated and complex antioxidant response, primarily mediated by the KEAP1-NRF2 pathway.

KEAP1, encoded on chromosome 19p13.2, is a Broad complex Tramtrack and Bric à Brac (BTB)-Kelch protein that functions as a negative regulator of NRF2 by facilitating its ubiquitination by the CUL3-based E3 ubiquitin ligase complex (4, 5). Under normal conditions, KEAP1 sequesters NRF2 in the cytoplasm, allowing for its polyubiquitination and subsequent degradation by the 26S proteasome (6). This maintains low basal levels of NRF2 and avoids inappropriate activation of the chemoprotective response (7). Under oxidative and electrophilic stress, however, radicals and electrophiles covalently modify reactive cysteine residues on KEAP1, diminishing the KEAP1-CUL3 complex’s ubiquitin ligase activity and stabilizing NRF2 (8–13). As it accumulates, NRF2 translocates to the nucleus, binds the antioxidant response element (ARE), a cis-acting enhancer sequence in the 5’-promoter region of cytoprotective genes, and induces transcription (10, 14–16).

While activation of NRF2 and its target genes has been shown to prevent cancer by eliminating carcinogenic oxidants and toxins, its constitutive activation can also drive tumorigenesis (17–20). In fact, NRF2 overexpression has been found to promote malignancy, tumorigenesis, proliferation, and migration, as well as confer resistance to cytotoxic agents (21–28). Cancers frequently achieve overexpression of NRF2 and these resulting phenotypes via somatic loss-of-function mutations in KEAP1. This has been well documented in renal, ovarian, esophageal, head and neck, gallbladder, and prostate cancer, and occurs at particularly high frequency in non-small cell lung cancer (18, 19, 25, 29–32). Aberrant activation of NRF2 via KEAP1 mutation is also associated with poor clinicopathological outcomes, including low-response rate to platinum-based chemotherapy, increased metastasis (long distance, lymph node, and tumor node), poor overall survival rate, and poor progression-free survival (33, 34). In addition to somatic mutations, epigenetic silencing of KEAP1 through promoter hypermethylation has also been reported in colorectal cancer, breast cancer, lung cancer, and kidney cancer (35–40). This hypermethylation is associated with increased NRF2 transcriptional activity, increased disease progression, higher mortality rate, and decreased response to chemotherapeutics (35–40).

In thyroid cancer, mutations that activate the MAPK pathway are the most common oncogenic drivers. However, approximately 10% of thyroid cancers have unidentified driver mutations (41, 42). Due to the prevalence of KEAP1 alterations in other malignancies, we sought to determine whether these mutations could represent an alternative driver of thyroid cancer, accounting for these unidentified mutations, or alternatively cooperate with other MAPK-activating mutations to further promote tumorigenesis. Limited evidence has implicated the KEAP1/NRF2 pathway in thyroid tumorigenesis; however, this connection remains incompletely characterized. In rare cases of familial non-toxic multinodular goiter, KEAP1 mutations have been identified and linked to increased NRF2 accumulation and transcriptional activity (43, 44). In papillary thyroid carcinoma (PTC), a handful of KEAP1 alterations have been identified and associated with increased nuclear NRF2 expression, correlating the mutations with NRF2 stabilization (45). Two retrospective immunohistochemical analyses also identified high nuclear NRF2 levels in PTC relative to benign lesions and normal tissue, indicating NRF2 pathway activation as a potential contributor to thyroid tumorigenesis (46, 47). Two studies also implicated the pathway in thyroid cancer progression through the inhibition of ferroptosis (48, 49). Additionally, an analysis of PTC samples from The Cancer Genome Atlas (TCGA) revealed epigenetic silencing of KEAP1 through promoter hypermethylation, further linking KEAP1/NRF2 pathway dysregulation to thyroid carcinogenesis (50).

While these previous reports identify KEAP1 mutations and implicate the KEAP1-NRF2 pathway in thyroid cancer, this connection has remained incompletely characterized. To our knowledge, this is the first study to comprehensively report the presence of inactivating KEAP1 mutations across all thyroid cancer subtypes. We characterize these mutations in the broader genomic landscape and identify for the first time a frequent genetic mechanism for biallelic KEAP1 loss. This study is also the first to comprehensively explore the functional consequences KEAP1 loss in thyroid carcinoma. We investigate the impact of KEAP1 loss on NRF2 pathway activation, as well as its effects on cell growth, migration, and drug response. Our findings indicate that KEAP1 mutations are functionally relevant and may represent a potential driver of thyroid tumorigenesis, as observed in other malignancies (21–28).

2 Materials and methods

2.1 Patient samples and targeted DNA/RNA sequencing

Patient samples were collected at the time of surgery and stored under the infrastructure of the Child and Adolescent Thyroid Consortium (CATC) Biorepository (IRB# 17–014224 and IRB# 20-018240). Histopathologic results, tumor staging, and demographic information including age and sex were collected from each patient and deidentified within the CATC Central REDCap data repository (51). This study involved human-derived biospecimens and was approved by the CHOP Institutional Review Board (IRB) with two arms: (a) for patients who had surgery before the time of initial IRB approval, a waiver of consent was approved as these were existing biospecimens from previous surgeries; and (b) for patients who had surgery after the time of initial IRB approval, enrollment is considered prospective and thus consent and assent were obtained.

Tumors were sequenced using the CHOP Division of Genomic Diagnostics’ Comprehensive Solid Tumor Panel (CSTP), which evaluates sequence and copy number (CN) alterations in 238 cancer associated genes, as previously described (52, 53). Briefly, genomic DNA was extracted from tumor samples and libraries were generated using probes targeting 238 genes. Libraries were sequenced on the Illumina MiSeq or HiSeq platforms using 150 bp paired-end sequencing. Short variants and CN alterations were called using the institutional ConcordS v2 software and NextGENe v2 NGS Analysis Software (SoftGenetics, State College, PA), respectively. mRNA expression was profiled using the HTG EdgeSeq Oncology Biomarker Panel (OBP), as previously described (54).

2.2 Data sources

Genetic data and clinical information for additional cases were collected from the following studies: AACR Project GENIE, Global Anaplastic Thyroid Cancer Initiative (GATCI), ICGC-THCA, MSK-IMPACT, MSK-MET, TCGA-THCA, Danilovic et al., Eszlinger et al., Ganly et al., Gopal et al., Kunstman et al., Landa et al., Lee et al., Nishihara et al., Pozdeyev et al., and Teshiba et al. (41, 43–45, 55–65). Data from these studies were accessed and downloaded from publicly available databases, including the Genomic Data Commons (GDC), cBioPortal, and COSMIC, or from the original publication.

For applicable studies, segmented CN data were downloaded and visualized in the Integrative Genomics Viewer (IGV). A log₂(CN) ≤ -0.2 threshold was set to designate cases with heterozygous segment loss at the relevant locus. mRNA expression data was downloaded and extracted from the Gene Expression Omnibus (GEO) using the GEOquery package (v2.74.0) in RStudio (v4.4.2).

2.3 Generation of clonal KEAP1 knock-out cell lines

TPC1 and SDAR2 cells were transfected with three distinct sgRNAs (Synthego) targeting sites in exon 2 of KEAP1 using Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer’s instructions. Ten experimental and two control transfections were performed per cell line. Sanger sequencing and Inference of CRISPR Edits (ICE, Synthego) were used to evaluate knock-out efficiency. Cells were then sorted using a FACSMelody Cell Sorter (BD Biosciences) and seeded at a density of 1 cell per well in a 96-well plate. Cells were cultured in growth medium supplemented with 20% FBS, 1% penicillin-streptomycin, 1X piperacillin-ciprofloxacin, and 1X amphotericin B, and then screened for mRNA expression of NRF2 target genes and Sanger sequenced to confirm successful knockout.

2.4 DNA extraction and Sanger sequencing

DNA was isolated using the Quick-DNA Miniprep Kit (Zymo Research, Irvine, CA) according to the manufacturer’s instructions. 50ng of DNA was PCR amplified, and the resulting product was purified with the ZR DNA Sequencing Clean-up Kit (Zymo Research). 10ng of the purified PCR product was Sanger sequenced using the forward and reverse primers. The sequence was extracted from the chromatogram using 4Peaks (Nucleobytes), and sequence alignment was performed using the Basic Local Alignment Search Tool (BLAST, NCBI). Primer information is listed in Supplementary Materials and Methods.

2.5 RNA isolation and RT-qPCR

RNA was extracted using TRIzol Reagent (Thermo Fisher Scientific) and the Direct-zol RNA Miniprep Kit (Zymo Research). 1000ng of RNA was reverse transcribed using the Verso cDNA Synthesis Kit and PCR reactions were performed using PowerTrack SYBR Green Master Mix (Applied Biosystems). Gene expression was measured in triplicate using the Quant Studio Real-Time PCR System (Applied Biosystems). β-actin (ACTB) was used to normalize the expression of all samples. Primer information is listed in Supplementary Materials and Methods.

2.6 Immunofluorescence imaging

Cells were fixed and permeabilized with a solution of 4% paraformaldehyde, 5% (w/v) sucrose and 0.1% Triton X-100. The fixed cells were blocked with 3% BSA for 1 hour at room temperature, incubated with the primary antibody overnight at 4°C, and incubated with the secondary antibody for 1 hour at room temperature. Nuclei were counterstained with 4’,6-diamidino-2-phenylindole (DAPI, 1µg/mL) for 15 minutes. Cells were imaged at 20X via confocal microscopy using the ZEISS LSM980. NRF2 nuclear intensity was quantitated using ImageJ. The following primary antibodies were used: NRF2 (1:200, Cell Signaling), Rhodamine Phalloidin-TRITC (1:400, Invitrogen). The following secondary antibody was used: Alexa Fluor™ Plus 555 Goat anti-Rabbit IgG (H+L) (1:500, Invitrogen).

2.7 Western blotting

Cells were lysed with radioimmunoprecipitation assay (RIPA) buffer supplemented with 1X Halt Protease and Phosphatase Inhibitor Cocktail and phenylmethylsulfonyl fluoride (PMSF) (Thermo Scientific). Protein concentration was determined using bicinchoninic acid (BCA) assay and 10µg of protein was loaded on a 10% polyacrylamide gel and resolved via SDS-PAGE. Proteins were transferred to a PVDF membrane (Millipore) and blocked with 5% BSA for 1 hour at room temperature. Membranes were incubated with the primary antibody overnight at 4 °C, secondary antibody for 1 hour at room temperature, and were visualized using SuperSignal™ West Femto Maximum Sensitivity Substrate (Thermo Scientific). The following primary antibodies were used: KEAP1 (1:1000, Cell Signaling), β-actin (1:5000, Cell Signaling). The following secondary antibody was used: HRP-linked Goat anti-Rabbit IgG (1:5000, Cell Signaling).

2.8 Cell proliferation assays

Growth curves were performed using the CellTiter-Glo luminescent assay. TPC1 cells (1500 cells/well in 5% FBS) and SDAR2 cells (1500 cells/well in 10% FBS) were seeded in 96-well plates. At the indicated time points, CellTiter-Glo reagent was added to each well at a 1:1 ratio with culture medium and the luminescence was measured using a Varioskan LUX microplate reader (Thermo Scientific). Luminescence was averaged across 10 replicates.

2.9 Transwell migration assays

Cells (10⁵/well) were seeded in the upper chamber of a 24-well transwell chamber (8µm pore size, Corning) in the recommended FBS-free medium containing 0.1% BSA. Complete medium containing 10% FBS was added to the bottom chamber of the insert as a chemoattractant. After 16 hours of incubation, non-migrating cells were removed from the upper chamber using a cotton swab, and the remaining cells were fixed and stained using 70% ethanol and 0.5% (w/v) crystal violet (Fisher Chemical). The number of migrated cells was counted from five randomly selected fields.

2.10 Dose-response curves

TPC1 cells (1500/well in 5% FBS) were seeded in 96-well plates. After 24 hours, cells were dosed with the indicated concentrations of selpercatinib diluted in culture medium. Cells were cultured for 72 hours and nuclei were stained with NucBlue Live ReadyProbes Reagent (Invitrogen). The stained nuclei were then imaged with the EVOS M7000 and quantified using ImageJ.

Cell counts were normalized to percent of control and the IC₅₀ was calculated using GraphPad’s four-parameter logistic regression.

2.11 Colony formation assays

TPC1 cells (500/well in 5% FBS) were seeded at single cell density in 6-well plates. After 24 hours, cells were dosed with the indicated concentration of drug diluted in culture medium. Drug treatments were refreshed every 3 days. After 12 days in culture, colonies were fixed and stained using 100% ice-cold methanol and 0.5% (w/v) crystal violet (Fisher Chemical). Wells were imaged, then crystal violet was eluted using 10% acetic acid. Absorbance was recorded at 595 nm on a Varioskan LUX microplate reader (Thermo Scientific).

2.12 Statistical analyses

All statistical tests were performed with GraphPad Prism 10 (v10.4.1) or RStudio (v4.4.2). Fisher’s Exact Test, one-way ANOVA, and two-way ANOVA were all conducted, and a two-sided p-value < 0.05 was considered statistically significant.

3 Results

3.1 KEAP1 mutations occur in both pediatric and adult thyroid tumors

Mutations in KEAP1 have previously been reported in various pediatric and adult cancers, but have not been widely reported or well characterized in the context of thyroid tumors. To assess the prevalence of these mutations in pediatric tumors, we sequenced samples from the CHOP Biorepository (n = 191) using the CSTP and identified five papillary carcinomas and one benign multinodular goiter harboring KEAP1 mutations (Figure 1, Supplementary Table S1). These six cases represented 3.14% of the total tumors in our biorepository and included four females with a mean age at surgery of 15.19 years (SD = 2.23). Four of the KEAP1 mutations were missense, while two were truncating. We next sought to determine if KEAP1 mutations are also present in non-pediatric cases. We examined sequencing data from publicly available datasets and published literature, and identified an additional 75 tumor samples and two thyroid cancer cell lines with KEAP1 mutations (Figure 1, Supplementary Table S2). The identified cell lines were previously established from primary tumors and not reported in tumor databases, thus representing independent cases. These cases included 36 females with an average age at surgery of 58.04 (SD = 13.63). Of these cases, 51 had missense mutations, 23 had truncating mutations, and one had an in-frame deletion.

Figure 1

Figure 1. KEAP1 mutations occur in both pediatric and adult thyroid tumors. Oncoprint detailing 81 KEAP1-mutant thyroid tumors, including cases from the current study and publicly available datasets. Clinicopathologic features include histology, age, sex, tumor origin, type of KEAP1 variant, chromosome 19p13.2 loss, and MAPK alteration. The predicted consequence of each KEAP1 mutation is annotated based on its OncoKB prediction. ATC, anaplastic thyroid carcinoma; FTC, follicular thyroid carcinoma; MNG, multinodular goiter; OTC, oncocytic thyroid carcinoma; PDTC, poorly-differentiated thyroid carcinoma; PTC, papillary thyroid carcinoma; LOH, loss of heterozygosity.

Between our CHOP cohort and publicly mined data, we identified 81 total cases representing a range of histologies: 3 multi-nodular goiters (MNGs) (3.70%), 7 follicular thyroid carcinomas (FTCs) (8.64%), 26 papillary thyroid carcinomas (PTCs) (32.1%), 12 oncocytic thyroid carcinomas (OTCs) (14.8%), 11 poorly-differentiated thyroid carcinomas (PDTCs) (13.6%), 18 anaplastic thyroid carcinomas (ATCs) (22.2%), and 4 cases with unspecified histologies (4.94%). Mutation mapping of these 81 KEAP1 mutations revealed that variants were distributed across all annotated domains of KEAP1, showing no evidence of clustering within a specific region (Supplementary Figure S1). Additionally, in cases where mutational status could be determined, 58 (92.1%) mutations were confirmed somatic, while 5 (7.94%) were confirmed germline. These five germline cases included three from our CHOP cohort, which were confirmed to be germline variants via Sanger sequencing or whole exome sequencing of paired normal tissue (Supplementary Figure S2). As non-adjacent specimens were not collected from these patients at the time of surgery, adjacent normal tissue was confirmed to be histologically benign and used for sequencing. For the remaining 18 cases, mutation status was unable to be evaluated due to limited data in publicly available cases or absence of paired normal tissue in the CHOP cohort.

We next examined CN segment data to determine if any KEAP1-mutant cases had concurrently lost their remaining wild-type allele. In 14 (50%) of cases with reported CN data, we identified loss of heterozygosity (LOH) at chromosome 19p13.2, indicating biallelic loss of wild-type KEAP1. In 24.7% of the KEAP1-mutant tumors, we also identified several concurrent MAPK-activating mutations: BRAF p.V600E, KRAS p.G12V/D, NRAS p.Q61R/K, RET, or ETV6::NTRK3 fusions. Notably, these MAPK alterations were mutually exclusive with 19p13.2 LOH in KEAP1-mutant tumors (OR = 0.049, p = 0.0044). Figure 1 displays an oncoprint plot detailing all cases harboring KEAP1 mutations, 19p13.2 loss, concurrent MAPK alteration, and associated demographic and histopathologic features. Supplementary Table S2 provides detailed molecular and demographic information for each sample.

3.2 KEAP1 mutations activate the NRF2 pathway in thyroid tumors

To determine whether KEAP1 mutations correlated with NRF2 pathway activation, we profiled mRNA expression of five canonical NRF2 target genes: AKR1C3, GCLC, GSR, NQO1, and TXNRD1. Six KEAP1-mutant tumors—four PTCs, one OTC, and one PDTC—had associated mRNA expression data. Four of these cases had missense KEAP1 mutations, while the others had truncating alterations (Figure 2A). In addition, we identified one FTC-derived cell line (described below) with a truncating KEAP1 mutation. Consistent with our prior observations, two of the six tumor cases exhibited concurrent MAPK pathway alterations: one carried a CCDC186::RET fusion, while the other had a KRAS p.G12V point mutation (Supplementary Table S2). Notably, three tumors also harbored concurrent 19p13.2 LOH. In five of the six cases, we found significant overexpression of canonical NRF2 target genes (Figures 2B–E). The only case lacking NRF2 pathway activation carried a germline KEAP1 p.C226P missense mutation, which co-occurred with the CCDC186::RET fusion and was computationally predicted to be tolerated by SIFT (Supplementary Figure S2, Supplementary Table S2).

Figure 2

Figure 2. KEAP1 mutations activate the NRF2 pathway in both pediatric and adult thyroid tumors. (A) Lollipop plot detailing histology and KEAP1 alterations in tumor samples and FTC cell line with paired RNA profiling and mutational analysis. (B-E) RNA expression of NRF2 target genes: AKR1C3, GCLC, GSR, NQO1, and TXNRD1 in (B) pediatric PTC/adenoma CHOP samples (n = 98), (C) adult PTC TCGA-THCA samples (n = 481), (D) adult PDTC/ATC samples from Landa, et al., 2016 (GSE76039) (n = 37), and (E) adult OTC samples from Ganly et al., 2018 (PRJNA445446) (n = 49). Samples of interest are colored based on their known or suspected KEAP1/NRF2 pathway alterations. Z-scores were calculated relative to all samples in the cohort. (F) RT-qPCR expression of NRF2 target genes AKR1C3, GCLC, NQO1, and TXNRD1 in SDAR1 (KEAP1^mut) and SDAR2 (KEAP1^wt) cell lines. * p < 0.05, ** p < 0.01, *** p < 0.001. Statistics were performed using a one-way ANOVA.

We also identified one PTC and one follicular adenoma in our CHOP cohort for which we had RNA sequencing data which demonstrated increased expression of the same NRF2 target genes (Figure 2B). There was no remaining sample available to perform KEAP1 mutational analysis, but the expression profiles support a loss-of-function KEAP1 mutation or other NRF2-activating alteration. We further observed two PTCs in the TCGA cohort that also had significant overexpression of AKR1C3, GCLC, GSR, NQO1, and TXNRD1 (Figure 2C). One case harbored a PAX8::NFE2L2 fusion, causing markedly increased expression of NFE2L2, the gene encoding NRF2, resulting in a similar functional effect as KEAP1 loss (Supplementary Figure S3). The second case lacked mutational data but exhibited CN loss of chromosome 19p13.2. This, in combination with its expression profile, supports a putative loss-of-function KEAP1 mutation.

In addition to patient samples, we analyzed two independent FTC cell lines, SDAR1 and SDAR2. These cell lines originated from the same patient but from different sites: a primary tumor and a neck metastasis, respectively. It has been previously reported that SDAR1 harbors a homozygous p.Q74* mutation in KEAP1, while SDAR2 does not (42). We sequenced these cell lines to evaluate their mutational status and confirmed that SDAR1 indeed carries a homozygous p.Q74* mutation, while SDAR2 is wild type for KEAP1 (Supplementary Figure S4). Consistent with KEAP1 loss and NRF2 pathway activation, SDAR1 showed increased expression of AKR1C3, GCLC, NQO1, and TXNRD1 compared to SDAR2 (Figure 2F).

3.3 Knockout of KEAP1 causes NRF2 pathway activation in PTC and FTC cells

Since we determined that KEAP1 mutations occur in tumors irrespective of MAPK driver status, we next sought to assess the effects of KEAP1 loss in cell lines with and without MAPK pathway alterations. Using the PTC-derived cell line TPC1, which carries a CCDC6::RET fusion, we knocked out KEAP1 to generate two clonal cell lines. Successful knockout was confirmed via Sanger sequencing and Western blotting (Supplementary Figure S5). Since NRF2 transcriptional activity is dependent on its nuclear localization, we performed immunofluorescence to determine the localization of NRF2. We found that both knockout cell lines exhibited increased levels of nuclear-localized NRF2 protein relative to the control cell line (Figures 3A, B). To further confirm increased NRF2 transcriptional activity, we evaluated the expression of the canonical NRF2 target genes AKR1C3, GCLC, NQO1, and TXNRD1 and found significant upregulation in response to KEAP1 knockout (Figure 3C). To ensure that induction of these genes was NRF2-dependent, we treated cells with the NRF2 inhibitor ML385 and observed that expression levels of these target genes decreased in the knockout cells, with minimal effect on the control cell line (Figure 3C).

Figure 3

Figure 3. Knockout of KEAP1 causes NRF2 pathway activation in PTC and FTC cells. (A) Representative immunofluorescence images of NRF2 (green), F-actin (red), and nuclei (DAPI, blue) in TPC1 control and KEAP1 knockout clones (KO1 and KO2), showing increased NRF2 nuclear localization upon KEAP1 knockout (scale bar = 20µm). (B) Quantitation of NRF2 nuclear intensity in TPC1 control (n = 84), TPC1 KO1 (n = 79), and TPC1 KO2 (n = 44) cells. (C, D) RT-qPCR expression of NRF2 target genes: AKR1C3, GCLC, NQO1, and TXNRD1 in (C) TPC1 KEAP1 knockout clones in the presence and absence of 5µM ML385 and (D) SDAR2 KEAP1 knockout clones in the presence and absence of 5µM ML385, showing NRF2 dependent upregulation of target genes in knockout cell lines (black/gray bars) relative to control cell lines (white bars). All experiments were repeated in triplicate. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001 versus the control cell line in the absence of ML385. # p < 0.001 versus the respective knockout cell line in the absence of ML385. Statistics were performed using a two-way ANOVA test.

We next examined the effects of KEAP1 loss in cells lacking a MAPK driver. Using SDAR2, an FTC cell line with no known driver mutation, we knocked out KEAP1 and generated two clonal cell lines. We again confirmed successful knockout via Sanger sequencing and Western blotting (Supplementary Figure S5). SDAR2 cell lines did not tolerate the serum starvation necessary for assessing NRF2 protein localization via immunofluorescence. However, we were still able to determine NRF2 transcriptional activity by evaluating expression of AKR1C3, GCLC, NQO1, and TXNRD1. We found significant upregulation of these genes in the knockout clones relative to the control cell line (Figure 3D). To confirm this expression pattern was NRF2-dependent, we again treated cells with ML385 and saw a marked reduction in expression of the target genes in the knockout cell lines, while control cells remained largely unaffected (Figure 3D).

3.4 Loss of KEAP1 promotes proliferation and migration in PTC and FTC cells

We next sought to assess the functional consequences of KEAP1 loss in cell lines with and without MAPK driver alterations by evaluating proliferation and migration. Using TPC1 KEAP1 knockout clones, we first evaluated proliferation utilizing the CellTiter-Glo Assay. We found that loss of KEAP1 increased growth, even in the presence of a MAPK driver mutation (Figure 4A). To then determine if this effect persisted in cells without a known driver alteration, we performed growth assays using SDAR2 KEAP1 knockout clones and observed a similar increase in proliferation relative to control cells (Figure 4B). We next assessed whether KEAP1 loss affects migration in MAPK-mutant and MAPK-wildtype cells by performing transwell migration assays. We observed that both TPC1 and SDAR2 cells exhibited increased cell migration upon KEAP1 knockout, suggesting that KEAP1 loss promotes cell migration independent of MAPK mutational status (Figures 4C–F).

Figure 4

Figure 4. Loss of KEAP1 stimulates proliferation and migration in PTC and FTC cells. Growth curves of (A) TPC1 control and KEAP1 knockout cell lines (KO1 and KO2) and (B) SDAR2 control and KEAP1 knockout cell lines (KO1 and KO2), showing increased proliferation in the knockout clones relative to controls. (C–F) Transwell migration assays of (C) TPC1 control and KEAP1 knockout clones (KO1 and KO2) and (E) SDAR2 control and KEAP1 knockout clones (KO1 and KO2) with relative quantifications (D, F) after crystal violet staining, showing increased migration upon KEAP1 loss. The number of migrated cells was averaged from five randomly selected 10X fields. ** p < 0.01, *** p < 0.001, **** p < 0.0001. Statistics were performed using a one-way ANOVA test.

3.5 KEAP1 loss attenuates the response of RET fusion-positive cells to targeted receptor tyrosine kinase inhibition

The KEAP1/NRF2 pathway regulates the transcriptional output of antioxidant and drug-metabolizing genes and has been implicated in drug resistance across multiple cancers (10, 28). We therefore investigated whether KEAP1 loss modulates the sensitivity of thyroid cancer cells to targeted RET inhibition. Selpercatinib is a selective RET tyrosine kinase inhibitor approved for use in RET fusion-positive thyroid cancers (66). Previous studies have shown that TPC1, which harbors a CCDC6::RET fusion, is sensitive to selpercatinib (67–69). We tested the efficacy of selpercatinib in TPC1 cells with and without KEAP1 knockout. We found that selpercatinib inhibited TPC1 parental cells with an IC₅₀ of 6.2 nM, consistent with prior reports (67–69). Loss of KEAP1 significantly reduced sensitivity to selpercatinib, with IC₅₀ values of 22.4 and 25.7 nM (Figure 5A). In the KEAP1 knockout cells, we observed a plateau in response to selpercatinib from 50-1000nM, indicating reduced sensitivity even at high inhibitor concentrations. To determine whether this effect persisted during extended selpercatinib exposure, we performed 12-day colony formation assays at high inhibitor concentrations. The TPC1 KEAP1 knockout cells remained less sensitive to selpercatinib than control cells, even after extended exposure (Figures 5B, C).

Figure 5

Figure 5. Loss of KEAP1 desensitizes RET fusion-positive cells to targeted receptor tyrosine kinase inhibition. (A) Dose-response graph showing TPC1 control and TPC1 KEAP1 knockout clones’ response to varying concentrations of selpercatinib after 3 days of treatment. Cell numbers were counted after nuclei staining, and percent viability was calculated relative to the 0nM condition for each cell line. (B) 12-day colony formation assay showing clonogenic potential of TPC1 control and TPC1 KEAP1 knockout cell lines in the presence of high-dose selpercatinib. Colonies were stained with crystal violet and each panel is an entire well of a 6-well dish. (C) Quantitation via crystal violet elution. OD measurements were taken, and percent viability was calculated relative to the 0nM condition for each cell line. All experiments were performed in triplicate. * p < 0.05, ** p < 0.01, *** p < 0.001 versus control cell line at each concentration of selpercatinib. Statistics were performed using a one-way ANOVA test.

4 Discussion

While KEAP1 mutations have been widely reported and studied in malignancies such as non-small cell lung cancer, their presence and role in thyroid cancer have been largely overlooked (70, 71). Although a few KEAP1 mutations have been reported in thyroid tumors, these studies lacked systematic evaluation of NRF2 pathway activation and in vitro modeling of mutation consequence (43–45). Having observed KEAP1 mutations in a number of pediatric thyroid cancer cases, we sought to determine whether they promote thyroid tumorigenesis, as has been reported in other malignancies. Here, we report the first comprehensive analysis of KEAP1 mutations in thyroid tumors, revealing that these alterations are far more prevalent and broadly distributed than previously recognized. Although prior reports have only identified a small number of KEAP1 mutations in PTCs and benign MNGs, we identified 81 such alterations across a diverse spectrum of thyroid cancers. Remarkably, we identified KEAP1 mutations in all thyroid cancer histologies, ranging from well-differentiated forms (PTC, FTC) to PDTC and ATC, with no subtype-specific mutational pattern. We also report the first instance of KEAP1 mutations occurring in pediatric thyroid tumors. These KEAP1/NRF2 pathway alterations occurred at a frequency of 3.1% in our pediatric cohort, as compared to 0.6% in adult PTCs (TCGA), 2.0% in adult OTC (PRJNA445446), and 2.7% in adult PDTC/ATC (GSE76039).

In both pediatric and adult thyroid tumors, we identified KEAP1 mutations with concurrent MAPK driver mutations, indicating potential cooperation between NRF2 and MAPK pathways in thyroid tumorigenesis. This group of dual KEAP1- and MAPK-mutant cases represents approximately 25% of all identified KEAP1-altered samples. We also identified a distinct subset of cases harboring both a KEAP1 mutation and loss of heterozygosity (LOH) at chromosome 19p13.2, revealing a concerted genetic mechanism by which both alleles of KEAP1 are inactivated through combined mutation and chromosomal loss. This biallelic loss occurred at a strikingly high frequency (approximately 50% of cases with CN data), consistent with the classical two-hit model that underlies inactivation of many tumor suppressors. Interestingly, biallelic loss of KEAP1 was mutually exclusive with MAPK alterations, suggesting two distinct modes by which KEAP1 may promote tumorigenesis. In tumors harboring initiating MAPK driver alterations, activation of the NRF2 pathway via heterozygous KEAP1 mutation may act as a cooperative event, enhancing tumor progression. Conversely, in tumors lacking MAPK mutations, biallelic inactivation of KEAP1, through combined single copy mutation and 19p13.2 LOH, could act as an initiating oncogenic event. Indeed, we identified seven cases of advanced disease (PDTC and ATC) and six PTC cases with no known driver mutations but with full inactivation of KEAP1, supporting the theory that homozygous KEAP1 loss may be sufficient to promote tumorigenesis. These previously unclassified tumors highlight the importance of re-evaluating historically “driverless” cases with updated genomic tools. As sequencing technologies advance, our ability to uncover cryptic or novel oncogenic drivers and characterize these “dark matter” tumors continues to expand. Although the majority of identified KEAP1-mutant cases lacked known driver mutations, limited CN data prevented us from determining whether these tumors also harbored biallelic KEAP1 loss. Even in cases with CN data, we recognize that tumor purity and ploidy may confound LOH calling. As a result, our analysis may include false negatives that underestimate the true prevalence of 19p13.2 LOH in KEAP1-mutant samples. Notably, we identified several cases in which LOH was not called by our criteria, yet, the KEAP1 mutation exhibited a high allele frequency, suggesting underlying LOH that was not formally identified. This high allele frequency was also observed in many samples that lacked CN data, further highlighting potential cases with unidentified LOH. More comprehensive genomic profiling is essential to completely dissect the importance of biallelic KEAP1 loss in thyroid tumors, particularly in cases that are genetically ambiguous.

For 78% of the cases in this report, we were able to confirm the germline or somatic status of the identified KEAP1 alteration. Only five cases had confirmed germline alterations, three of which were benign multinodular goiters. The presence of these mutations is consistent with prior reports implicating KEAP1 germline variants in familial multinodular goiter (43, 44, 72). Notably, the vast majority (92%) of alterations were somatic, further suggesting that KEAP1 mutations may be functionally relevant in tumorigenesis. Although we recognize that some of the somatic KEAP1 mutations listed may be passenger alterations, nearly half (48%) were predicted to be oncogenic or likely oncogenic by OncoKB. Evaluation of NRF2 pathway activation can help better profile the functionality of these mutations. Indeed, in 6 out of 7 cases where expression data was available, we observed upregulation of NRF2 target genes, indicating loss-of-function alterations. Notably, two of these mutations were not predicted by OncoKB to be oncogenic, but still showed significant NRF2 pathway activation, underscoring the value of a multifaceted approach to more accurately assess the pathogenicity of these mutations. The single KEAP1 mutation lacking evidence of NRF2 pathway activation was a germline variant, co-occurred with a CCDC186::RET fusion driver mutation, and was predicted to be functionally neutral by SIFT, all supporting its likely status as a non-functional variant. While this MAPK-altered case lacked a functionally significant KEAP1 mutation, the case carrying concurrent KEAP1 and KRAS p.G12V mutations had robust NRF2 pathway activation, supporting that the mutation is functionally significant. These findings provide further support to the idea that in certain molecular contexts, NRF2 pathway activation may cooperate with MAPK mutations to enhance tumor progression. Notably, the co-occurrence of KRAS and KEAP1 mutations has been linked to poorer clinical outcomes and reduced therapeutic efficacy in non-small cell lung cancer, further supporting a potential cooperative role of KEAP1 alterations and MAPK activation (73). The strong correlation between NRF2 target gene expression and loss-of-function KEAP1 mutations suggests the ability to infer KEAP1 mutational status via an mRNA expression signature when accompanying DNA sequencing data is absent. Indeed, we identified four additional tumors with elevated NRF2 pathway activation in our transcriptomic analysis. While one case harbored a known PAX8::NFE2L2 fusion, causing significant overexpression of NRF2 and its target genes, the remaining three lacked mutational profiling. Despite the absence of sequencing data for these tumors, their expression signatures showing NRF2 pathway activation strongly suggest underlying KEAP1 inactivation, further emphasizing the importance of paired genomic and transcriptomic information to determine both mutation presence and functional consequence.

We also report the first functional characterization of KEAP1 mutations in thyroid cancer using in vitro models. To model the consequences of KEAP1 loss both independently of and concurrently with MAPK driver alterations, we generated clonal KEAP1 knockout cell lines from both TPC1, which harbors a CCDC6::RET fusion, and SDAR2, which lacks a known driver. In line with the patterns observed in patient tumors, we confirmed that KEAP1 loss indeed induces NRF2 target gene expression in an NRF2-dependent manner in both MAPK-mutant and MAPK-wildtype cells, validating the strong association between KEAP1 inactivation and NRF2 pathway activation. Consistent with the role of KEAP1 mutations in driving tumorigenesis, we found that loss of KEAP1 promoted both cell proliferation and migration. These effects were observed in both cell lines, directly showing that NRF2 pathway activation via KEAP1 loss can act independently of or cooperate with MAPK alterations to promote tumor progression. These findings further implicate KEAP1 mutations as functionally significant mutations in thyroid cancer, directly show their role in promoting tumorigenic cellular phenotypes across distinct genetic backgrounds, and position them as a potential therapeutic target in this disease.

In other cancers, overactivation of the NRF2 pathway has been shown to confer chemotherapeutic resistance; however, this resistance has not been explored in the context of MAPK inhibition (25, 28, 74). Here, we report the first instance of KEAP1 mutations attenuating drug response in thyroid cancer cells. We show that loss of KEAP1 reduces sensitivity of RET fusion-positive PTC cells to the receptor tyrosine kinase inhibitor selpercatinib, a key therapeutic agent for RET-driven thyroid cancers. These findings suggest that KEAP1 inactivation may represent a previously unrecognized mechanism of resistance to RET inhibition in thyroid tumors. Our novel in vitro findings highlight the clinical relevance of KEAP1 mutations in thyroid cancer and underscore the importance of mutational profiling to better inform therapeutic strategies.

There are several limitations to this study, highlighting the importance of future work with more robust clinical and functional analysis of identified KEAP1 variants. The retrospective nature of this study precluded us from validating the pathogenicity of missense variants in cases that lacked paired RNA sequencing. Although we employed computational algorithms to predict the impact of these missense variants, we show that these algorithms are not always accurate. Additionally, more robust CN data in KEAP1 mutant tumors is necessary to further dissect the prevalence and functional relevance of biallelic versus monoallelic loss. Future studies should employ RNA sequencing in addition to mutational and CN analysis to fully profile all KEAP1 mutations in patient samples. In this study, we also perform in vitro characterization of KEAP1 loss using knockout cell line models. While these models show convincing evidence that KEAP1 mutations promote proliferation, migration, and drug resistance, it will be important to confirm these observations in vivo. Development of models in the context of monoallelic or biallelic loss will also be necessary to determine the functional relevance of partial versus complete KEAP1 inactivation in tumorigenesis. Additionally, generation of both in vivo and in vitro knockout models in the background of different driver mutations will be important to determine if KEAP1 promotes resistance to other clinically relevant inhibitors, such as dabrafenib or larotrectinib. These models will also facilitate detailed mechanistic studies of how KEAP1 loss drives drug resistance.

Our comprehensive report of KEAP1 mutations in thyroid tumors highlights its previously unappreciated prevalence and broad distribution across thyroid cancer subtypes. By combining multi-omic analyses with in vitro characterization of KEAP1 loss in thyroid cancer, we demonstrate that these mutations are functionally relevant and capable of promoting tumorigenesis across diverse genetic landscapes. Additionally, we identify a highly prevalent and concerted genetic mechanism in which both copies of KEAP1 are lost. This two-hit mechanism, which is mutually exclusive with classical MAPK driver alterations, may represent an initiating oncogenic event in some tumors. This potential dual role—as either an initiating event or cooperating alteration—highlights the previously unrecognized importance of KEAP1 mutations in thyroid cancer. Moreover, our finding that KEAP1 mutations not only drive tumor growth and migration but also attenuate cellular response to selpercatinib in RET fusion-positive cells highlights their clinical significance. These findings indicate that KEAP1 mutations play an important role in promoting tumorigenesis and modulating therapeutic response, suggesting that profiling of this pathway should be integrated into future studies and clinical testing.

Data availability statement

The data presented in this study were previously deposited in the NCBI's Sequence Read Archive (SRA), BioProject Accession: PRJNA934932 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA934932).

Ethics statement

The studies involving humans were approved by the Children’s Hospital of Philadelphia Institutional Review Board (IRB #17-014224 and IRB #20-018240). The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by the participants’ legal guardians/next of kin.

Author contributions

NB: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing – original draft, Writing – review & editing. JR-F: Conceptualization, Data curation, Methodology, Writing – original draft, Writing – review & editing. ER: Data curation, Software, Writing – review & editing. CH-P: Investigation, Writing – review & editing. KH: Data curation, Writing – review & editing. AI: Data curation, Writing – review & editing. AB: Funding acquisition, Resources, Writing – review & editing. AF: Conceptualization, Data curation, Funding acquisition, Methodology, Resources, Supervision, Writing – original draft, Writing – review & editing.

Funding

The author(s) declared that financial support was received for this work and/or its publication. This work was supported in part by a grant from The Children’s Hospital of Philadelphia Frontier Programs (AB, AF), NIH R01CA214511 (AF), and Department of Defense W81XWH2210655 (AF).

Acknowledgments

We would like to acknowledge Zachary Spangler, Michele Scheerer, Kay Labella, and the entire Franco Laboratory for their support throughout the duration of this study and preparation of this manuscript.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc.2025.1685379/full#supplementary-material

References

1. Finkel T. Signal transduction by reactive oxygen species. J Cell Biol. (2011) 194:7–15. doi: 10.1083/jcb.201102095

PubMed Abstract | Crossref Full Text | Google Scholar

2. Balaban RS, Nemoto S, and Finkel T. Mitochondria, oxidants, and aging. Cell. (2005) 120:483–95. doi: 10.1016/j.cell.2005.02.001

PubMed Abstract | Crossref Full Text | Google Scholar

3. Ma Q. Transcriptional responses to oxidative stress: pathological and toxicological implications. Pharmacol Ther. (2010) 125:376–93. doi: 10.1016/j.pharmthera.2009.11.004

PubMed Abstract | Crossref Full Text | Google Scholar

4. Cullinan SB, Gordan JD, Jin J, Harper JW, and Diehl JA. The keap1-btb protein is an adaptor that bridges nrf2 to a cul3-based E3 ligase: oxidative stress sensing by a cul3-keap1 ligase. Mol Cell Biol. (2004) 24:8477–86. doi: 10.1128/mcb.24.19.8477-8486.2004

PubMed Abstract | Crossref Full Text | Google Scholar

5. Kobayashi A, Kang MI, Okawa H, Ohtsuji M, Zenke Y, Chiba T, et al. Oxidative stress sensor keap1 functions as an adaptor for cul3-based E3 ligase to regulate proteasomal degradation of nrf2. Mol Cell Biol. (2004) 24:7130–9. doi: 10.1128/mcb.24.16.7130-7139.2004

PubMed Abstract | Crossref Full Text | Google Scholar

6. Tian W, Rojo de la Vega M, Schmidlin CJ, Ooi A, and Zhang DD. Kelch-like ech-associated protein 1 (Keap1) differentially regulates nuclear factor erythroid-2-related factors 1 and 2 (Nrf1 and nrf2). J Biol Chem. (2018) 293:2029–40. doi: 10.1074/jbc.RA117.000428

PubMed Abstract | Crossref Full Text | Google Scholar

7. Nguyen T, Sherratt PJ, Huang HC, Yang CS, and Pickett CB. Increased protein stability as a mechanism that enhances nrf2-mediated transcriptional activation of the antioxidant response element. Degradation of Nrf2 by the 26 S Proteasome. J Biol Chem. (2003) 278:4536–41. doi: 10.1074/jbc.M207293200

PubMed Abstract | Crossref Full Text | Google Scholar

8. Dinkova-Kostova AT, Holtzclaw WD, and Wakabayashi N. Keap1, the sensor for electrophiles and oxidants that regulates the phase 2 response, is a zinc metalloprotein. Biochemistry. (2005) 44:6889–99. doi: 10.1021/bi047434h

PubMed Abstract | Crossref Full Text | Google Scholar

9. Ma Q and He X. Molecular basis of electrophilic and oxidative defense: promises and perils of nrf2. Pharmacol Rev. (2012) 64:1055–81. doi: 10.1124/pr.110.004333

PubMed Abstract | Crossref Full Text | Google Scholar

10. Ma Q. Role of nrf2 in oxidative stress and toxicity. Annu Rev Pharmacol Toxicol. (2013) 53:401–26. doi: 10.1146/annurev-pharmtox-011112-140320

PubMed Abstract | Crossref Full Text | Google Scholar

11. Dinkova-Kostova AT, Holtzclaw WD, Cole RN, Itoh K, Wakabayashi N, Katoh Y, et al. Direct evidence that sulfhydryl groups of keap1 are the sensors regulating induction of phase 2 enzymes that protect against carcinogens and oxidants. Proc Natl Acad Sci U.S.A. (2002) 99:11908–13. doi: 10.1073/pnas.172398899

PubMed Abstract | Crossref Full Text | Google Scholar

12. Eggler AL, Liu G, Pezzuto JM, van Breemen RB, and Mesecar AD. Modifying specific cysteines of the electrophile-sensing human keap1 protein is insufficient to disrupt binding to the nrf2 domain neh2. Proc Natl Acad Sci U.S.A. (2005) 102:10070–5. doi: 10.1073/pnas.0502402102

PubMed Abstract | Crossref Full Text | Google Scholar

13. Hong F, Sekhar KR, Freeman ML, and Liebler DC. Specific patterns of electrophile adduction trigger keap1 ubiquitination and nrf2 activation. J Biol Chem. (2005) 280:31768–75. doi: 10.1074/jbc.M503346200

PubMed Abstract | Crossref Full Text | Google Scholar

14. Keum YS. Regulation of nrf2-mediated phase ii detoxification and anti-oxidant genes. Biomol Ther (Seoul). (2012) 20:144–51. doi: 10.4062/biomolther.2012.20.2.144

PubMed Abstract | Crossref Full Text | Google Scholar

15. Rushmore TH, Morton MR, and Pickett CB. The antioxidant responsive element. Activation by oxidative stress and identification of the DNA consensus sequence required for functional activity. J Biol Chem. (1991) 266:11632–9. doi: 10.1016/S0021-9258(18)99004-6

PubMed Abstract | Crossref Full Text | Google Scholar

16. Nguyen T, Sherratt PJ, and Pickett CB. Regulatory mechanisms controlling gene expression mediated by the antioxidant response element. Annu Rev Pharmacol Toxicol. (2003) 43:233–60. doi: 10.1146/annurev.pharmtox.43.100901.140229

PubMed Abstract | Crossref Full Text | Google Scholar

17. Ramos-Gomez M, Dolan PM, Itoh K, Yamamoto M, and Kensler TW. Interactive effects of nrf2 genotype and oltipraz on benzo[a]Pyrene-DNA adducts and tumor yield in mice. Carcinogenesis. (2003) 24:461–7. doi: 10.1093/carcin/24.3.461

PubMed Abstract | Crossref Full Text | Google Scholar

18. Ohta T, Iijima K, Miyamoto M, Nakahara I, Tanaka H, Ohtsuji M, et al. Loss of keap1 function activates nrf2 and provides advantages for lung cancer cell growth. Cancer Res. (2008) 68:1303–9. doi: 10.1158/0008-5472.Can-07-5003

PubMed Abstract | Crossref Full Text | Google Scholar

19. Padmanabhan B, Tong KI, Ohta T, Nakamura Y, Scharlock M, Ohtsuji M, et al. Structural basis for defects of keap1 activity provoked by its point mutations in lung cancer. Mol Cell. (2006) 21:689–700. doi: 10.1016/j.molcel.2006.01.013

PubMed Abstract | Crossref Full Text | Google Scholar

20. Iida K, Itoh K, Kumagai Y, Oyasu R, Hattori K, Kawai K, et al. Nrf2 is essential for the chemopreventive efficacy of oltipraz against urinary bladder carcinogenesis. Cancer Res. (2004) 64:6424–31. doi: 10.1158/0008-5472.CAN-04-1906

PubMed Abstract | Crossref Full Text | Google Scholar

21. DeNicola GM, Karreth FA, Humpton TJ, Gopinathan A, Wei C, Frese K, et al. Oncogene-induced nrf2 transcription promotes ros detoxification and tumorigenesis. Nature. (2011) 475:106–9. doi: 10.1038/nature10189

PubMed Abstract | Crossref Full Text | Google Scholar

22. Bauer AK, Cho HY, Miller-Degraff L, Walker C, Helms K, Fostel J, et al. Targeted deletion of nrf2 reduces urethane-induced lung tumor development in mice. PloS One. (2011) 6:e26590. doi: 10.1371/journal.pone.0026590

PubMed Abstract | Crossref Full Text | Google Scholar

23. Zhang HS, Zhang ZG, Du GY, Sun HL, Liu HY, Zhou Z, et al. Nrf2 promotes breast cancer cell migration via up-regulation of G6pd/hif-1α/notch1 axis. J Cell Mol Med. (2019) 23:3451–63. doi: 10.1111/jcmm.14241

PubMed Abstract | Crossref Full Text | Google Scholar

24. Ko E, Kim D, Min DW, Kwon SH, and Lee JY. Nrf2 regulates cell motility through rhoa-rock1 signalling in non-small-cell lung cancer cells. Sci Rep. (2021) 11:1247. doi: 10.1038/s41598-021-81021-0

PubMed Abstract | Crossref Full Text | Google Scholar

25. Homma S, Ishii Y, Morishima Y, Yamadori T, Matsuno Y, Haraguchi N, et al. Nrf2 enhances cell proliferation and resistance to anticancer drugs in human lung cancer. Clin Cancer Res. (2009) 15:3423–32. doi: 10.1158/1078-0432.CCR-08-2822

PubMed Abstract | Crossref Full Text | Google Scholar

26. Ma X, Zhang J, Liu S, Huang Y, Chen B, and Wang D. Nrf2 knockdown by shrna inhibits tumor growth and increases efficacy of chemotherapy in cervical cancer. Cancer Chemother Pharmacol. (2012) 69:485–94. doi: 10.1007/s00280-011-1722-9

PubMed Abstract | Crossref Full Text | Google Scholar

27. Was H, Cichon T, Smolarczyk R, Rudnicka D, Stopa M, Chevalier C, et al. Overexpression of heme oxygenase-1 in murine melanoma: increased proliferation and viability of tumor cells, decreased survival of mice. Am J Pathol. (2006) 169:2181–98. doi: 10.2353/ajpath.2006.051365

PubMed Abstract | Crossref Full Text | Google Scholar

28. Wang XJ, Sun Z, Villeneuve NF, Zhang S, Zhao F, Li Y, et al. Nrf2 enhances resistance of cancer cells to chemotherapeutic drugs, the dark side of nrf2. Carcinogenesis. (2008) 29:1235–43. doi: 10.1093/carcin/bgn095

PubMed Abstract | Crossref Full Text | Google Scholar

29. Ooi A, Wong JC, Petillo D, Roossien D, Perrier-Trudova V, Whitten D, et al. An antioxidant response phenotype shared between hereditary and sporadic type 2 papillary renal cell carcinoma. Cancer Cell. (2011) 20:511–23. doi: 10.1016/j.ccr.2011.08.024

PubMed Abstract | Crossref Full Text | Google Scholar

30. Shibata T, Kokubu A, Gotoh M, Ojima H, Ohta T, Yamamoto M, et al. Genetic alteration of keap1 confers constitutive nrf2 activation and resistance to chemotherapy in gallbladder cancer. Gastroenterology. (2008) 135:1358–68. doi: 10.1053/j.gastro.2008.06.082

PubMed Abstract | Crossref Full Text | Google Scholar

31. Sjoblom T, Jones S, Wood LD, Parsons DW, Lin J, Barber TD, et al. The consensus coding sequences of human breast and colorectal cancers. Science. (2006) 314:268–74. doi: 10.1126/science.1133427

PubMed Abstract | Crossref Full Text | Google Scholar

32. Singh A, Misra V, Thimmulappa RK, Lee H, Ames S, Hoque MO, et al. Dysfunctional keap1-nrf2 interaction in non-small-cell lung cancer. PloS Med. (2006) 3:e420. doi: 10.1371/journal.pmed.0030420

PubMed Abstract | Crossref Full Text | Google Scholar

33. Solis LM, Behrens C, Dong W, Suraokar M, Ozburn NC, Moran CA, et al. Nrf2 and keap1 abnormalities in non-small cell lung carcinoma and association with clinicopathologic features. Clin Cancer Res. (2010) 16:3743–53. doi: 10.1158/1078-0432.Ccr-09-3352

PubMed Abstract | Crossref Full Text | Google Scholar

34. Wang Q, Xu L, Wang G, Chen L, Li C, Jiang X, et al. Prognostic and clinicopathological significance of nrf2 expression in non-small cell lung cancer: A meta-analysis. PloS One. (2020) 15:e0241241. doi: 10.1371/journal.pone.0241241

PubMed Abstract | Crossref Full Text | Google Scholar

35. Fabrizio FP, Costantini M, Copetti M, Torre A, Sparaneo A, Fontana A, et al. Abstract 2397: epigenetic silencing in clear renal cell carcinoma: keap1 promoter hypermethylation. Cancer Res. (2017) 77:2397. doi: 10.1158/1538-7445.Am2017-2397

Crossref Full Text | Google Scholar

36. Maiello E, Parrella P, Latiano TP, Barbano R, Muscarella LA, Pasculli B, et al. Hypermethylation of the keap1 gene in colorectal cancer and association with disease progression. J Clin Oncol. (2013) 31:e14655–e. doi: 10.1200/jco.2013.31.15_suppl.e14655

Crossref Full Text | Google Scholar

37. Hanada N, Takahata T, Zhou Q, Ye X, Sun R, Itoh J, et al. Methylation of the keap1 gene promoter region in human colorectal cancer. BMC Cancer. (2012) 12:66. doi: 10.1186/1471-2407-12-66

PubMed Abstract | Crossref Full Text | Google Scholar

38. Barbano R, Muscarella LA, Pasculli B, Valori VM, Fontana A, Coco M, et al. Aberrant keap1 methylation in breast cancer and association with clinicopathological features. Epigenetics. (2013) 8:105–12. doi: 10.4161/epi.23319

PubMed Abstract | Crossref Full Text | Google Scholar

39. Fabrizio FP, Sparaneo A, Centra F, Trombetta D, Storlazzi CT, Graziano P, et al. Methylation density pattern of keap1 gene in lung cancer cell lines detected by quantitative methylation specific pcr and pyrosequencing. Int J Mol Sci. (2019) 20:2697. doi: 10.3390/ijms20112697

PubMed Abstract | Crossref Full Text | Google Scholar

40. Wang R, An J, Ji F, Jiao H, Sun H, and Zhou D. Hypermethylation of the keap1 gene in human lung cancer cell lines and lung cancer tissues. Biochem Biophys Res Commun. (2008) 373:151–4. doi: 10.1016/j.bbrc.2008.06.004

PubMed Abstract | Crossref Full Text | Google Scholar

41. Cancer Genome Atlas Research N. Integrated genomic characterization of papillary thyroid carcinoma. Cell. (2014) 159:676–90. doi: 10.1016/j.cell.2014.09.050

PubMed Abstract | Crossref Full Text | Google Scholar

42. Landa I, Pozdeyev N, Korch C, Marlow LA, Smallridge RC, Copland JA, et al. Comprehensive genetic characterization of human thyroid cancer cell lines: A validated panel for preclinical studies. Clin Cancer Res. (2019) 25:3141–51. doi: 10.1158/1078-0432.CCR-18-2953

PubMed Abstract | Crossref Full Text | Google Scholar

43. Nishihara E, Hishinuma A, Kogai T, Takada N, Hirokawa M, Fukata S, et al. A novel germline mutation of keap1 (R483h) associated with a non-toxic multinodular goiter. Front Endocrinol (Lausanne). (2016) 7:131. doi: 10.3389/fendo.2016.00131

PubMed Abstract | Crossref Full Text | Google Scholar

44. Teshiba R, Tajiri T, Sumitomo K, Masumoto K, Taguchi T, and Yamamoto K. Identification of a keap1 germline mutation in a family with multinodular goitre. PloS One. (2013) 8:e65141. doi: 10.1371/journal.pone.0065141

PubMed Abstract | Crossref Full Text | Google Scholar

45. Danilovic DLS, de Mello ES, Frazzato EST, Wakamatsu A, de Lima Jorge AA, Hoff AO, et al. Oncogenic mutations in keap1 disturbing inhibitory nrf2-keap1 interaction: activation of antioxidative pathway in papillary thyroid carcinoma. Head Neck. (2018) 40:1271–8. doi: 10.1002/hed.25105

PubMed Abstract | Crossref Full Text | Google Scholar

46. Ziros PG, Manolakou SD, Habeos IG, Lilis I, Chartoumpekis DV, Koika V, et al. Nrf2 is commonly activated in papillary thyroid carcinoma, and it controls antioxidant transcriptional responses and viability of cancer cells. J Clin Endocrinol Metab. (2013) 98:E1422–7. doi: 10.1210/jc.2013-1510

PubMed Abstract | Crossref Full Text | Google Scholar

47. Stuchi LP, Castanhole-Nunes MMU, Maniezzo-Stuchi N, Biselli-Chicote PM, Henrique T, Padovani Neto JA, et al. Vegfa and nfe2l2 gene expression and regulation by micrornas in thyroid papillary cancer and colloid goiter. Genes (Basel). (2020) 11:954. doi: 10.3390/genes11090954

PubMed Abstract | Crossref Full Text | Google Scholar

48. Li S, Zhang Y, Zhang J, Yu B, Wang W, Jia B, et al. Neferine exerts ferroptosis-inducing effect and antitumor effect on thyroid cancer through nrf2/ho-1/nqo1 inhibition. J Oncol. (2022) 2022:7933775. doi: 10.1155/2022/7933775

PubMed Abstract | Crossref Full Text | Google Scholar

49. Li W, Huang G, Wei J, Cao H, and Jiang G. Alkbh5 inhibits thyroid cancer progression by promoting ferroptosis through tiam1–nrf2/ho-1 axis. Mol Cell Biochem. (2023) 478:729–41. doi: 10.1007/s11010-022-04541-x

PubMed Abstract | Crossref Full Text | Google Scholar

50. Martinez VD, Vucic EA, Pikor LA, Thu KL, Hubaux R, and Lam WL. Frequent concerted genetic mechanisms disrupt multiple components of the nrf2 inhibitor keap1/cul3/rbx1 E3-ubiquitin ligase complex in thyroid cancer. Mol Cancer. (2013) 12:124. doi: 10.1186/1476-4598-12-124

PubMed Abstract | Crossref Full Text | Google Scholar

51. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, and Conde JG. Research electronic data capture (Redcap)–a metadata-driven methodology and workflow process for providing translational research informatics support. J BioMed Inform. (2009) 42:377–81. doi: 10.1016/j.jbi.2008.08.010

PubMed Abstract | Crossref Full Text | Google Scholar

52. Surrey LF, MacFarland SP, Chang F, Cao K, Rathi KS, Akgumus GT, et al. Clinical utility of custom-designed ngs panel testing in pediatric tumors. Genome Med. (2019) 11:32. doi: 10.1186/s13073-019-0644-8

PubMed Abstract | Crossref Full Text | Google Scholar

53. Franco AT, Ricarte-Filho JC, Isaza A, Jones Z, Jain N, Mostoufi-Moab S, et al. Fusion oncogenes are associated with increased metastatic capacity and persistent disease in pediatric thyroid cancers. J Clin Oncol. (2022) 40:1081–90. doi: 10.1200/jco.21.01861

PubMed Abstract | Crossref Full Text | Google Scholar

54. Ricarte-Filho JC, Casado-Medrano V, Reichenberger E, Spangler Z, Scheerer M, Isaza A, et al. Dicer1 rnase iiib domain mutations trigger widespread mirna dysregulation and mapk activation in pediatric thyroid cancer. Front Endocrinol (Lausanne). (2023) 14:1083382. doi: 10.3389/fendo.2023.1083382

PubMed Abstract | Crossref Full Text | Google Scholar

55. Consortium APG. Aacr project genie: powering precision medicine through an international consortium. Cancer Discov. (2017) 7:818–31. doi: 10.1158/2159-8290.CD-17-0151

PubMed Abstract | Crossref Full Text | Google Scholar

56. Eszlinger M, Khalil M, Gillmor AH, Huang H, Stewardson P, McIntyre JB, et al. Histology-based molecular profiling improves mutation detection for advanced thyroid cancer. Genes Chromosomes Cancer. (2021) 60:531–45. doi: 10.1002/gcc.22949

PubMed Abstract | Crossref Full Text | Google Scholar

57. Ganly I, Makarov V, Deraje S, Dong Y, Reznik E, Seshan V, et al. Integrated genomic analysis of hürthle cell cancer reveals oncogenic drivers, recurrent mitochondrial mutations, and unique chromosomal landscapes. Cancer Cell. (2018) 34:256–70.e5. doi: 10.1016/j.ccell.2018.07.002

PubMed Abstract | Crossref Full Text | Google Scholar

58. Gopal RK, Kübler K, Calvo SE, Polak P, Livitz D, Rosebrock D, et al. Widespread chromosomal losses and mitochondrial DNA alterations as genetic drivers in hürthle cell carcinoma. Cancer Cell. (2018) 34:242–55.e5. doi: 10.1016/j.ccell.2018.06.013

PubMed Abstract | Crossref Full Text | Google Scholar

59. Kunstman JW, Juhlin CC, Goh G, Brown TC, Stenman A, Healy JM, et al. Characterization of the mutational landscape of anaplastic thyroid cancer via whole-exome sequencing. Hum Mol Genet. (2015) 24:2318–29. doi: 10.1093/hmg/ddu749

PubMed Abstract | Crossref Full Text | Google Scholar

60. Landa I, Ibrahimpasic T, Boucai L, Sinha R, Knauf JA, Shah RH, et al. Genomic and transcriptomic hallmarks of poorly differentiated and anaplastic thyroid cancers. J Clin Invest. (2016) 126:1052–66. doi: 10.1172/jci85271

PubMed Abstract | Crossref Full Text | Google Scholar

61. Lee Y, Moon S, Seok JY, Lee J-H, Nam S, and Chung YS. Characterization of the genomic alterations in poorly differentiated thyroid cancer. Sci Rep. (2023) 13:19154. doi: 10.1038/s41598-023-46466-5

PubMed Abstract | Crossref Full Text | Google Scholar

62. Nguyen B, Fong C, Luthra A, Smith SA, DiNatale RG, Nandakumar S, et al. Genomic characterization of metastatic patterns from prospective clinical sequencing of 25,000 patients. Cell. (2022) 185:563–75.e11. doi: 10.1016/j.cell.2022.01.003

PubMed Abstract | Crossref Full Text | Google Scholar

63. Pozdeyev N, Gay LM, Sokol ES, Hartmaier R, Deaver KE, Davis S, et al. Genetic analysis of 779 advanced differentiated and anaplastic thyroid cancers. Clin Cancer Res. (2018) 24:3059–68. doi: 10.1158/1078-0432.Ccr-18-0373

PubMed Abstract | Crossref Full Text | Google Scholar

64. Zehir A, Benayed R, Shah RH, Syed A, Middha S, Kim HR, et al. Mutational landscape of metastatic cancer revealed from prospective clinical sequencing of 10,000 patients. Nat Med. (2017) 23:703–13. doi: 10.1038/nm.4333

PubMed Abstract | Crossref Full Text | Google Scholar

65. Zeng PYF, Prokopec SD, Lai SY, Pinto N, Chan-Seng-Yue MA, Clifton-Bligh R, et al. The genomic and evolutionary landscapes of anaplastic thyroid carcinoma. Cell Rep. (2024) 43:113826. doi: 10.1016/j.celrep.2024.113826

PubMed Abstract | Crossref Full Text | Google Scholar

66. Duke ES, Bradford D, Marcovitz M, Amatya AK, Mishra-Kalyani PS, Nguyen E, et al. Fda approval summary: selpercatinib for the treatment of advanced ret fusion-positive solid tumors. Clin Cancer Res. (2023) 29:3573–8. doi: 10.1158/1078-0432.CCR-23-0459

PubMed Abstract | Crossref Full Text | Google Scholar

67. Administration USFaD. Selpercatinib multi-disciplinary review and evaluation. Drugs@FDA (2020). Available online at: https://www.accessdata.fda.gov/drugsatfda_docs/nda/2020/213246Orig1s000MultidisciplineR.pdf (Accessed July 2, 2025).

Google Scholar

68. Chen W, Dream S, Leung PY, Wu PK, Wong S, and Park JI. Selpercatinib combination with the mitochondria-targeted antioxidant mitoq effectively suppresses ret-mutant thyroid cancer. NPJ Precis Oncol. (2024) 8:39. doi: 10.1038/s41698-024-00536-7

PubMed Abstract | Crossref Full Text | Google Scholar

69. Amitani M and Ito K-i. Abstract 3278: analysis of the mechanism of resistance to selpercatinib in thyroid cancer cells with ret/ptc1 fusion gene. Cancer Res. (2024) 84:3278. doi: 10.1158/1538-7445.AM2024-3278

Crossref Full Text | Google Scholar

70. Zavitsanou A-M, Pillai R, Hao Y, Wu WL, Bartnicki E, Karakousi T, et al. Keap1 mutation in lung adenocarcinoma promotes immune evasion and immunotherapy resistance. Cell Rep. (2023) 42:113295. doi: 10.1016/j.celrep.2023.113295

PubMed Abstract | Crossref Full Text | Google Scholar

71. Li J, Shi D, Li S, Shi X, Liu Y, Zhang Y, et al. Keap1 promotes anti-tumor immunity by inhibiting pd-L1 expression in nsclc. Cell Death Dis. (2024) 15:175. doi: 10.1038/s41419-024-06563-3

PubMed Abstract | Crossref Full Text | Google Scholar

72. Nishihara E, Fukata S, Hirokawa M, Higuchi M, Ito M, Nishikawa M, et al. Nodule-specific nrf2-targeted upregulation in patients with keap1 mutations and familial nontoxic multinodular goiter. J Clin Endocrinol Metab. (2024) 110:973–982. doi: 10.1210/clinem/dgae699

PubMed Abstract | Crossref Full Text | Google Scholar

73. Arbour KC, Jordan E, Kim HR, Dienstag J, Yu HA, Sanchez-Vega F, et al. Effects of co-occurring genomic alterations on outcomes in patients with kras-mutant non-small cell lung cancer. Clin Cancer Res. (2018) 24:334–40. doi: 10.1158/1078-0432.CCR-17-1841

PubMed Abstract | Crossref Full Text | Google Scholar

74. Islam SS, Qassem K, Islam S, Parag RR, Rahman MZ, Farhat WA, et al. Genetic alterations of keap1 confers chemotherapeutic resistance through functional activation of nrf2 and notch pathway in head and neck squamous cell carcinoma. Cell Death Dis. (2022) 13:696. doi: 10.1038/s41419-022-05126-8

PubMed Abstract | Crossref Full Text | Google Scholar