TMTP1‐Modified Small Extracellular Vesicles Target BRAF Mutation in Anaplastic Thyroid Cancer Reversing Vemurafenib Resistance With CRISPR/Cas9 Delivery

ABSTRACT

This study investigates a novel approach to overcome Vemurafenib resistance in BRAF‐mutant Anaplastic thyroid carcinoma (ATC) using CRISPR/Cas9 gene editing and TMTP1‐modified extracellular vesicles (TMTP1‐sgBRAF‐EVs). By knocking out the BRAF gene, the study elucidates Vemurafenib‐induced ferroptosis mechanisms involving lipid peroxidation and reactive oxygen species (ROS) generation in ATC cells. The developed TMTP1‐sgBRAF‐EVs system demonstrates superior tumour‐targeting and drug delivery capabilities, significantly enhancing Vemurafenib efficacy in both in vitro and in vivo models. This innovative combination of gene editing technology with a nanoparticle delivery system shows promising potential as a therapeutic strategy for treating aggressive BRAF‐mutant ATC.

Abbreviations

AFM

atomic force microscopy

ATC

anaplastic thyroid carcinoma

CHX

cycloheximide

DEGs

differentially expressed genes

EVs

extracellular vesicles

FBS

fetal bovine serum

KEGG

Kyoto encyclopaedia of genes and genomes

MDA

malondialdehyde

NEB

new England biolabs

NTA

nanoparticle tracking analysis

PFA

paraformaldehyde

PI

propidium Iodide

PS

Penicillin‐Streptomycin

PVDF

polyvinylidene fluoride

RNA‐Seq

RNA sequencing

ROS

reactive oxygen species

TEM

transmission electron microscopy

TFRC

transferrin receptor

TMTP1

tumour metastasis targeting peptide 1

TMTP1‐sgBRAF‐EVs

CRISPR/Cas9 gene editing and tumour metastasis targeting Peptide 1‐modified Extracellular Vesicle

1. Introduction

Anaplastic thyroid carcinoma (ATC) is a rare but highly aggressive type of thyroid cancer. Despite its low incidence among all thyroid cancer cases, the prognosis for ATC is extremely poor, with a typical survival period of less than 6 months (Baloch et al. 2022; Boucai et al. 2024; Bible et al. 2021). The highly invasive nature and rapid progression of ATC make it a particularly challenging disease to manage (Liu et al. 2022; Califano et al. 2023). Traditional treatments, such as radiation and chemotherapy, are largely ineffective against ATC, and patients often exhibit significant resistance to these conventional therapies (Ding et al. 2024; Wang et al. 2022c). In recent years, advances in molecular biology and genomics have led to a deeper understanding of the molecular mechanisms underlying ATC, particularly the role of key genetic mutations in its development and progression (Grabowska et al. 2023; Alzahrani et al. 2023). However, effective therapeutic strategies for ATC remain lacking, highlighting the urgent need for novel treatments to improve patient survival and quality of life.

ATC is the most aggressive and lethal type of thyroid cancer. BRAF mutations, particularly the V600E variant, occur frequently in ATC and are closely associated with tumour invasiveness, metastatic potential, and poor prognosis (Vladimirova et al. 2021; Nasimi et al. 2021). This mutation leads to the continuous activation of the MAPK signalling pathway, promoting the growth and survival of tumour cells (Johnson et al. 2022; Lu et al. 2022). Additionally, BRAFV600E may reshape the tumour microenvironment and suppress immune cell activity, leading to decreased responsiveness to immunotherapy. The BRAF^V600E mutation is also found in various other cancers, particularly melanoma, colorectal cancer, and thyroid cancer (Cook and Lochhead 2022; Shimoi et al. 2024). Although BRAF inhibitors such as Vemurafenib have shown efficacy in treating these cancers, their effectiveness in BRAF‐mutant ATC is limited (Yuan and Guo 2022; Xu et al. 2024). ATC cells rapidly develop resistance to Vemurafenib, significantly reducing its therapeutic effectiveness. Studies suggest that BRAFV600E‐mutant tumour cells can upregulate PD‐L1 to suppress T cell activity, thereby evading immune surveillance (Ma et al. 2022). Various combinational strategies—including immunomodulators, targeted therapies, and chemotherapy—are being explored to enhance the efficacy of immunotherapy (Zeng et al. 2023b). In addition, tumour microenvironment modulation, such as using anti‐angiogenic agents, may further improve therapeutic outcomes (Ou et al. 2024). A better understanding of resistance mechanisms may enable the development of more effective immunotherapy strategies for BRAFV600E‐mutant colorectal cancer patients (Bible et al. 2021; Scheffel et al. 2021; Maniakas et al. 2022).

Ferroptosis is a form of iron‐ and lipid peroxidation‐dependent cell death that has recently emerged as a promising mechanism in cancer suppression (Zou et al. 2022; Chen et al. 2023; Sun et al. 2021). Unlike apoptosis or necrosis, ferroptosis regulates cell viability through iron metabolism and lipid reactive oxygen species (ROS) accumulation (Yan et al. 2021; Zhang et al. 2022; Yang et al. 2022). Modulating ferroptosis‐related genes significantly impacts tumour cell survival and drug sensitivity, offering a novel therapeutic avenue (Hao et al. 2021; Kim et al. 2022b). Studies have shown that in BRAFV600E‐mutant melanoma cell lines A375 and G‐361, classical ferroptosis inducers such as Erastin and RSL3 effectively trigger ferroptotic cell death. Furthermore, it has been demonstrated that BRAF‐activated ARSI inhibits EREG‐mediated ferroptosis, promoting BRAFV600E‐mutant thyroid cancer progression and resistance to Sorafenib (Chen et al. 2025). This suggests that BRAF mutations alter cellular sensitivity to ferroptosis inducers, underscoring the influence of BRAF on ferroptosis susceptibility. Ongoing efforts aim to develop novel targeted therapies beyond current BRAF inhibitors, with tumour microenvironment‐targeting strategies showing potential to enhance efficacy and reduce resistance (Goswami et al. 2024). As the mechanisms underlying BRAFV600E mutation become clearer, more effective targeted therapies for colorectal cancer and other BRAF‐driven malignancies are anticipated.

CRISPR/Cas9 gene editing technology has been widely applied in cancer research in recent years, enabling precise targeting and knockout of specific genes to investigate their roles in tumour development and progression (Wang et al. 2022c; Gong et al. 2021, Happi Mbakam et al. 2022). Due to its efficiency, precision, and versatility, CRISPR/Cas9 has become a powerful tool for studying gene functions and developing gene therapies (Wang et al. 2022c; Park and Bao 2021; Happi Mbakam et al. 2022). extracellular vesicles (EVs), as natural nanocarriers, possess low immunogenicity and high biocompatibility, making them ideal tools for drug delivery (Zou et al. 2023; Mondal et al. 2023; Li et al. 2021c). EVs can carry various biomolecules, such as proteins, RNA, and DNA, playing a crucial role in intercellular communication and regulation of cellular functions (Hu et al. 2023; Mu et al. 2023; Li et al. 2021c). Tumour Metastasis Targeting Peptide 1 (TMTP1) can specifically target tumour cells, enhancing the efficiency of drug delivery. Therefore, in this study, we developed a TMTP1‐modified EV system to deliver CRISPR/Cas9 gene editing tools, aiming to increase the sensitivity of BRAF‐mutant ATC cells to Vemurafenib. The TMTP1‐modified EVs system achieves precise drug delivery by specifically recognising receptors on the surface of tumour cells, thereby improving therapeutic outcomes (Li et al. 2022; Wang et al. 2022c; Shen et al. 2022).

The primary objective of this study is to explore the mechanisms of reversing Vemurafenib resistance in BRAF‐mutant ATC using a CRISPR/Cas9 gene editing and TMTP1‐modified EV (TMTP1‐sgBRAF‐EVs) delivery system. First, we utilised CRISPR/Cas9 technology to generate stable BRAF knockout subclones in BRAF‐mutant ATC cell lines and analysed their sensitivity to Vemurafenib. Next, we performed the Kyoto Encyclopaedia of Genes and Genomes (KEGG) enrichment analysis to investigate the impact of BRAF knockout on ferroptosis‐related gene expression and conducted a series of experiments to validate its role in ferroptosis. Finally, we developed the TMTP1‐sgBRAF‐EVs delivery system, assessed its tumour‐targeting and delivery capabilities, and evaluated its effect on Vemurafenib efficacy through both in vitro and in vivo experiments. The results of this study have the potential to offer new therapeutic hope for patients with BRAF‐mutant ATC, improving their survival rates and quality of life, and providing new theoretical foundations and strategies for cancer treatment.

2. Materials and Methods

2.1. Ethical Statement

BALB/c nude mice (4–6 weeks old) were purchased from Hunan SJA Laboratory Animal Co., Ltd. (Hunan, China). All animals were housed in rooms with controlled temperature, humidity, and lighting, and were provided with water ad libitum. Animal welfare and experimental procedures were conducted in accordance with the Ethical Guidelines for the Care and Use of Laboratory Animals of Anhui Medical University and were approved by the University's Animal Experimentation Ethics Committee.

2.2. Cell Culture

Human normal thyroid cells Nthy‐ori3‐1 (BNCC340487, Beina Chuanglian Biotechnology Co., Ltd., China) and ATC cell lines 8505c (BRAF^V600E, BNCC360331, BeNa Culture Collection, China), BHT‐101 (BRAF^V600E, BNCC359868, BeNa Culture Collection, China), CAL‐62 (BRAF^WT, BNCC359829, BeNa Culture Collection, China), and KMH‐2 (BRAF^WT, BNCC359895, BeNa Culture Collection, China) were cultured in RPMI 1640 medium (22400089, Gibco, Shanghai, China) supplemented with 10% fetal bovine serum (FBS) (10099158, Gibco, Shanghai, China) and 1% penicillin‐streptomycin (PS) (15140122, Gibco, Shanghai, China). All cell lines were maintained in a humidified incubator at 37°C with 5% CO₂. All cell lines used in this study were authenticated by short tandem repeat (STR) analysis and confirmed to be free of Mycoplasma contamination.

2.3. In Vitro Cell Proliferation and Drug Resistance Assay

Cells were seeded at a density of 1 × 10⁴ cells per well in 24‐well plates and cultured in RPMI 1640 medium supplemented with 10% FBS. After 24 h, the cells were treated with RPMI 1640 medium containing 5% FBS and varying concentrations of Vemurafenib (HY‐12057, MedChemExpress (MCE), Shanghai, China). The proliferation and drug resistance of BHT‐101 and 8505C cells were assessed using the Cell Counting Kit‐8 (CCK‐8, Dojindo Chemical Technology Co., Ltd, Shanghai, China). Briefly, cells were seeded in 96‐well plates at a density of 5000 cells per well in 100 µL of complete medium and incubated at 37°C with 5% CO₂ for 24 h. The medium was then replaced with 100 µL of CCK‐8 working solution per well, and the cells were incubated for an additional 2 h. The viability of the cells in each well was determined by measuring the absorbance.

2.4. Generation of BRAF Knockout Cells

BRAF‐deficient cells (sgBRAF) were generated using CRISPR/Cas9 technology. The sgRNAs used were as follows: BRAF‐sgRNA1: 5'‐TAGCACTGAAAGGCTAGAAG‐3'; BRAF‐sgRNA2: 5'‐TGTCCCCGTTGAACAGAGCC‐3'; BRAF‐sgRNA3: 5'‐TTGGAGTGCATCTAGCTTGC‐3'. These sgRNAs were inserted into the Lenti‐CRISPR v2 vector (Hanbio Biotechnology Co., Ltd., Shanghai, China), which contains the Streptococcus pyogenes Cas9 nuclease gene. Cells were infected with lentivirus containing the Lenti‐CRISPR v2 vector and selected with 4 µg/ml puromycin (A1113803, Gibco, Thermo Fisher Scientific, Shanghai, China) to obtain stable BRAF knockout (sgBRAF) cells. The knockout effect was confirmed by Western blot analysis (Zeng et al. 2023b).

2.5. Isolation and Purification of EVs

EVs were isolated and purified from the conditioned media of cultured 8505c and BHT‐101 cells. The culture supernatant was first centrifuged at 300 × g for 10 min at 4°C to remove cell debris. The resulting supernatant was then centrifuged at 2000 × g for 15 min at 4°C, followed by centrifugation at 5000 × g for 15 min at 4°C to further remove residual debris. The supernatant was then filtered through a 0.45 µm membrane and layered on top of a 2 mL 60% sucrose cushion. This was followed by ultracentrifugation at 100,000 × g for 90 min at 4°C. Approximately 3 mL of the interface between the sucrose cushion and the conditioned media was collected. For a second round of flotation ultracentrifugation, the 10 mL sample was layered on top of a 0.75 mL 60% sucrose cushion and centrifuged at 110,000 × g for 12 h at 4°C using an SW41Ti rotor. Approximately 1 mL of the interface was collected from each tube after the second round of flotation ultracentrifugation. The samples were then loaded onto pre‐packed chromatography columns connected to the NGC Quest 10 chromatography system. The columns were equilibrated and eluted with PBS that had been filtered through a 0.22 µm membrane and degassed. Each sample was eluted with 75 fractions of 2 mL PBS each. Portions of the collected samples were used for nanoparticle tracking analysis (NTA) and BCA protein assay. Finally, the EVs were filtered through a 0.45 µm membrane, concentrated using an Amicon Ultra‐15 centrifugal filter unit at 3260 × g, and resuspended in PBS. The purified EVs were stored at −80°C until use. All procedures were carried out at 4°C.

2.6. Identification of EVs

EVs were diluted in PBS and fixed with 1% glutaraldehyde, then applied to carbon‐coated copper grids and stained with 1% phosphotungstic acid. The specimens were examined using a JEM‐2100 transmission electron microscope (JEOL, Tokyo, Japan). Atomic force microscopy (AFM) images were recorded using the BioScope Resolve system and ScanAsyst imaging mode (Bruker, Billerica, MA, USA). EV samples at a concentration of 5 µg/mL were prepared by placing a drop on the grid and incubating until dry. The images were processed using NanoScope analysis software (Bruker).

NTA was performed using the NanoSight NS300 system (Malvern Instruments, Malvern, UK). The Brownian motion of EVs suspended in PBS was recorded and tracked, and size distribution data were generated using the Stokes‐Einstein equation. The zeta potential of EVs was measured at 25°C in 0.1× PBS buffer using the Nano ZS90 instrument (Malvern).

The protein concentration of EVs was determined using the BCA protein assay kit (P0010S, Beyotime). Western blot analysis was conducted to detect EV markers ALIX, TSG101, Calnexin, and CD63.

2.7. Labelling of EVs

Fluorescent dyes 3,3'‐dioctadecyloxacarbocyanine perchlorate (DiO, D275) and 1,1'‐dioctadecyl‐3,3,3',3'‐tetramethylindodicarbocyanine iodide (DiR, D12731) were purchased from Invitrogen (USA) for labelling EVs. Purified EVs were incubated with each fluorescent dye (5 mM DiO or 5 mM DiR) at 37°C for 15 min, followed by ultracentrifugation at 120,000 g for 90 min to remove free dye. The labelled EVs were washed twice with PBS by centrifugation at 120,000 g and resuspended in PBS before use.

2.8. Homology of Tumour EVs to Parental Cell Lines In Vitro

8505c or BHT‐101 cells were seeded at a density of 2 × 10⁴ cells/mL on 4‐well culture slides. After 24 h, the medium was replaced with RPMI 1640 containing 7 × 10⁸ DiR‐ or DiO‐labelled 8505c‐EVs or BHT‐101‐EVs. The cells were then cultured for an additional 12 h, washed three times with PBS, fixed with 4% paraformaldehyde (PFA), and stained with DAPI (Life Technologies, R37606). Images were captured using a fluorescence microscope (Olympus IX81). Additionally, BHT‐101 and 8505c cells were seeded at a density of 2 × 10⁴ cells/mL in 96‐well plates. After 24 h, the medium was replaced with RPMI 1640 containing 1.75 × 10⁸ DiR‐labelled BHT‐101‐EVs or 8505c‐EVs, and the cells were cultured for 12 h. Fluorescence measurements were performed using a Fluoroskan Ascent FL microplate fluorometer (Thermo Fisher, 374 Fluoroskan Ascent FL).

A mixture of 100 µL RPMI 1640 containing 2 × 10⁶ 8505c cells and 100 µL PBS was subcutaneously injected into the flanks of immunodeficient BALB/c nude mice (Hunan SJA Laboratory Animal Co., Ltd., Hunan, China). After 2 weeks, the tumours grew to approximately 200 mm². The mice were then randomly divided into two groups: (1) intravenous injection of 3 × 10¹¹ DiR‐labelled 8505c‐EVs in 100 µL PBS, and (2) intravenous injection of 3 × 10¹¹ DiR‐labelled BHT‐101‐EVs in 100 µL PBS. After 24 h, the mice were euthanised, and the tumours and major organs were harvested. Fluorescence in each organ was detected using the IVIS Spectrum Imaging System (Caliper Lifesciences).

2.9. Modification of EVs With Tumour‐Targeting Peptide TMTP1

To establish EVs with tumour‐targeting functionality expressing the TMTP1 peptide, we fused the targeting peptide TMTP1 to the N‐terminal of Lamp2b, a protein reported to be abundantly present in EV membranes, creating the TMTP1‐Lamp2b‐HA construct. This construct was then inserted into the lentiviral vector pLVX‐IRES‐G418 (TMTP1‐Lamp2b‐HA). The lentiviral system was used to transfect 293T cells with TMTP1‐Lamp2b‐HA. Lentiviral particles were collected and used to infect 8505c or BHT‐101 cells at a density of 2.5 × 10⁵ cells per well in 6 cm culture dishes with varying concentrations of lentivirus. After 2 days of transduction, the medium was replaced with 4 mg/mL puromycin for 2 days of cell selection. The medium was then replaced with serum‐free RPMI 1640 for 48 h, and EVs were isolated by ultracentrifugation.

To prepare TMTP1‐sgBRAF‐EVs, purified EVs were resuspended in pre‐chilled electroporation buffer (1.15 mM potassium phosphate, pH 7.2; 25 mM potassium chloride; 21% OptiPrep working solution), and the concentration was adjusted to ∼10¹⁰ particles/mL. The EV suspension was filtered through a 0.22 µm filter and stored on ice for further use. To form EV ribonucleoprotein (RNP) complexes, Cas9 protein (New England Biolabs (NEB), Beverly, MA, USA) was thoroughly mixed with sgRNA to create the RNP complex. The RNP complex was then added to the EVs at a mass ratio of 1:3, and electroporation was performed to form the EV RNP complex. Electroporation was carried out at 110 V for 4 ms, with a 10 ms interval, a pulse duration of 50 ms, and a pulse separation time of 50 ms, using the Gene Pulser Xcell (Bio‐Rad) for 10 pulses. Following electroporation, the mixture was centrifuged at 800 × g for 5 min, 3000 × g for 15 min, and 10,000 × g for 1 h. The resulting supernatant was ultracentrifuged at 10,000 g for 2 h at 4°C, and the precipitate was resuspended in cold PBS solution.

2.10. Encapsulation Efficiency of Electroporation

To evaluate the encapsulation efficiency of electroporation, a standard curve was generated using Western blot. The Grey values were plotted on the Y‐axis, and the amounts of Cas9 protein on the X‐axis, resulting in the equation Y = 0.02095X − 0.2033. Subsequently, 400 ng of Cas9 protein was added to the electroporation reaction system. After electroporation, EVs were isolated using the aforementioned method, and the EV proteins were extracted for analysis.

2.11. Proteinase K and RNase Protection Assay

TMTP1‐sgBRAF‐EVs (25 mL, 1×10¹⁰ vesicles/mL) were treated with or without Proteinase K (0.5 mg/mL; Cat# 39450‐01‐6, MCE) and with or without 0.25% Triton X‐100 (Cat# 9036‐19‐5, Sigma) at 37°C for 30 min. The level of spCas9 within the EVs was analysed by Western blotting. For sgRNA detection, EVs were treated with RNase A (100 µg/mL; Thermo) with or without Triton X‐100 at 37°C for 30 min, followed by termination with RNA column purification buffer (Trans, Cat. No. ER101‐01). Total RNA was then extracted. Reverse transcription was performed with PrimeScript RT reagent kit (Takara) containing random primers to ensure effective reverse transcription of non‐polyA RNAs such as sgRNA (Wan et al. 2024).

2.12. Stability Measurement of TMTP1‐sgBRAF‐EVs

To assess the stability of TMTP1‐sgBRAF‐EVs over time, they were stored in phosphate‐buffered saline (PBS, 1x, pH 7.4) at 4°C for 7 days. To simulate their eventual in vivo environment, they were placed in fresh serum at 37°C for 24 h. The stability of TMTP1‐sgBRAF‐EVs in both conditions was monitored using the NanoSight system.

2.13. Internalisation Efficiency of TMTP1‐sgBRAF‐EVs

8505c cells were seeded onto coverslips and, once adhered, were co‐cultured with TMTP1‐sgBRAF‐EVs and unmodified EVs for 24 h. After fixing the cells, the nuclei were stained with DAPI, and the cytoskeleton was stained with Phalloidin (ab176753, Abcam, UK). The internalisation of Cas9‐GFP and sgBRAF‐Cy3 was observed using a fluorescence microscope to evaluate the colocalisation of Cas9 and sgRNA. Colocalisation analysis was performed by measuring the fluorescence intensity of GFP and Cy3 using image processing software, and a colocalisation curve was plotted. For quantitative analysis of EV internalisation, treated 8505c cells were washed with PBS and collected by trypsinisation with 0.25% trypsin. The cells were resuspended in PBS and analysed by flow cytometry to assess EV internalisation efficiency by measuring the fluorescence intensity of GFP within the cells.

Additionally, in vivo experiments involved preparing frozen sections of tumour tissues from each group. Cas9 was labelled with GFP, sgBRAF with Cy3, and EVs with DiR. The sections were observed under a fluorescence microscope.

2.14. Lentivirus Infection and Drug Treatment

Human transferrin receptor (TFRC)‐shRNA constructs were created using the vector pLKO.1‐puro (QYV0024, Qualityard Biological Technology Co., Ltd.). The sequences used were: Sequence 1: GCTGGTCAGTTCGTGATTAAA, Sequence 2: CCCAACAGATACTGGAAGTTT, and shNC: AGACCAGACCGCACGTAAACC, all purchased from Sigma‐Aldrich. TFRC‐shRNA lentivirus (sh‐TFRC) and control lentivirus (sh‐NC) were constructed using HEK293T cells (CBP60661, Nanjing Cobioer Biotechnology Co., Ltd., Jiangsu, China). The plasmids and lentivirus packaging services were provided by Sangon Biotech. The constructed plasmid carrying the TFRC‐shRNA and a luciferase reporter gene (sh‐TFRC‐luc) was co‐transfected with helper plasmids into HEK293T cells using Lipofectamine 2000 reagent (11668030, Thermo Fisher, USA). After validation, amplification, and purification, the packaged lentivirus was obtained.

For lentivirus‐mediated transfection, 5 × 10⁵ cells were seeded into 6‐well plates. When the cells reached 70%–90% confluency, they were transfected with lentivirus at a multiplicity of infection (MOI) of 10 (working titre: approximately 5 × 10⁶ TU/mL) and 5 µg/mL polybrene (TR‐1003, Merck, USA) in the culture medium. After 4 h of transfection, an equal volume of culture medium was added to dilute the polybrene. After 24 h of transfection, the medium was replaced with a fresh culture medium. Transfection efficiency was observed 48 h post‐transfection using the luciferase reporter gene. For stable cell line selection, 10 µg/mL puromycin (A1113803, Gibco, Grand Island, NY, USA) was used for antibiotic selection. Once the cells no longer died in the puromycin‐containing medium, they were collected for subsequent experiments.

For the drug and EV treatment, 8505c and BHT‐101 cells, with or without TFRC knockdown, were treated for 24 h with 10 nM equal amounts of DMSO, Vemurafenib, NAC (HY‐B0215, MCE, Shanghai, China), apoptosis inhibitor ZVAD‐fmk (HY‐16658B, MCE, Shanghai, China), necroptosis inhibitor necrostatin‐1 (HY‐15760, MCE, Shanghai, China), ferroptosis inhibitor ferrostatin‐1 (HY‐100579, MCE, Shanghai, China), iron chelator deferoxamine (DFO, HY‐B0988, MCE, Shanghai, China), autophagy inducer rapamycin (Rap) (HY‐10219, MCE, Shanghai, China), autophagy inhibitor chloroquine (CQ) (HY‐17589A, MCE, Shanghai, China), unloaded EVs (unmod‐EVs), and TMTP1‐sgBRAF‐EVs. Additionally, BHT‐101 and 8505c cells were continuously exposed to Erastin (HY‐15763, MCE, Shanghai, China) and RSL3 (HY‐100218A, MCE, Shanghai, China) to develop ferroptosis‐resistant cell lines.

2.15. Colony Formation and Soft Agar Assay

8505c and BHT‐101 cells (1 × 10³ cells per well) were seeded into 12‐well plates. The RPMI 1640 medium was supplemented with 3% Noble agar (BD Biosciences) and 10% FBS, and then an additional 0.6% agar RPMI 1640 medium was added. After 3 weeks of incubation, colonies with a diameter of at least 500 µm were counted using a digital microscope.

2.16. Transwell Assay

For the migration assay, 600 µL of 30% FBS medium was added to 24‐well plates, and Transwell chambers (Millipore, Darmstadt, Germany) were placed on top. Then, 300 µL of a cell suspension at a concentration of 1 × 10⁶ cells/mL was added to each chamber. Depending on the cell characteristics, the cells were incubated for 24–72 h. After incubation, the 24‐well plates were removed, and the medium in the upper chambers was discarded. The cells on the upper surface of the membrane were wiped off with a cotton swab, and the outer layer of the chamber was washed twice with PBS. The cells were fixed with 600 µL of 4% PFA for 30 min and stained with 600 µL of crystal violet staining solution at room temperature for 15 min. After washing with PBS, the cells were air‐dried and observed under a microscope. Six random fields were photographed, and the number of cells was counted.

For the invasion assay, ECM gel (ECM550, Chemicon, USA) was placed at 4°C overnight. On the following day, all pipette tips and Transwell chambers were pre‐cooled on ice for 30 min before the experiment. The ECM gel was diluted with a serum‐free medium at a ratio of 1:9 to a final concentration of 1 mg/mL. Forty microliters of the diluted ECM gel were added to the polycarbonate membrane of each upper chamber in a 24‐well Transwell plate. The plates were incubated in a 37°C incubator with 5% CO₂ for 5 h to allow the ECM gel to polymerise. Excess liquid was removed, and 70 µL of pure DMEM medium was added to each chamber, followed by incubation at 37°C for 0.5 h to rehydrate the matrix gel. The excess medium was then removed. The rest of the procedure was performed as described in the migration assay.

2.17. Cell Death Assay

Cells were seeded in 20 mm confocal dishes and incubated overnight, followed by drug treatment. The live/dead detection kit (PF00007, Proteintech, China), consisting of Calcein‐AM/PI, was used according to the manufacturer's instructions.

2.18. Reactive Oxygen Species and Lipid Peroxidation Assay

Cells were seeded in 6‐well plates or 20 mm glass‐bottom cell culture dishes (NEST, China) and incubated overnight, followed by drug treatment. Cells were incubated for 48 h, 72 h, or other specified times, and the C11 BODIPY 581/591 assay kit (GC40165, GLPBIO, USA) was used according to the manufacturer's instructions for detecting lipid peroxidation. The ROS detection kit (HY‐D0940, MCE, Shanghai, China) was used following the manufacturer's protocol. Lipid peroxidation levels were detected using flow cytometry (BD Accuri C6 Plus, BD Bioscience; BD LSRFortessa, USA) and a fluorescence microscope (Olympus Corporation, Japan). ROS levels were measured by flow cytometry.

2.19. Malondialdehyde Assay

Cells were seeded in cell culture flasks and incubated overnight, followed by drug treatment. After 72 h of incubation, the malondialdehyde (MDA) assay kit (M496, Dojindo, Japan) was used according to the manufacturer's instructions to measure MDA levels.

2.20. Reduced Glutathione/Oxidised Glutathione Activity Assay

Cells were seeded in 75 cm² cell culture flasks and incubated overnight, followed by drug treatment. After 72 h of incubation, the GSSG/GSH Quantification Kit II (G263, Dojindo, Japan) was used as per the manufacturer's instructions. The ratio of GSH/GSSG was measured using a microplate reader (Biotek, USA) through luminescence.

2.21. Transmission Electron Microscopy (TEM) Observation of Cells

Cells were harvested and prefixed with 2.5% glutaraldehyde (111‐30‐8, Sigma‐Aldrich, USA) in the dark, followed by fixation with formaldehyde (50‐00‐0, Sigma‐Aldrich, USA) for 1 h. The cells were then dehydrated using a graded series of acetone concentrations (48358, Sigma‐Aldrich, USA) and embedded in epoxy resin (1675‐54‐3, Sigma‐Aldrich, USA). The cells were gradually heated in an oven at 42°C for 2 h, then at 52°C overnight, and finally at 62°C for an additional night to complete polymerisation. Ultrathin sections (50–70 nm) were prepared using an ultramicrotome (EM UC6, Leica Biosystems, Germany). Sections were treated with a 1% uranyl acetate aqueous solution (XHL2375, Hubei Xinhongli Chemical Co., Ltd., Tianmen, China) for 5 min and placed in a CO₂‐free environment for 2–4 min to generate citrate. The sections were observed at 100 kV using a JEM‐2100 transmission electron microscope (JEOL, Tokyo, Japan) and digitally captured with an AMT V700 side‐mount camera.

2.22. FerroOrange Staining

Cells were fixed with 4% PFA solution for approximately 15 to 20 min. FerroOrange dye (MX4559‐48UG, Shanghai Maokang Biotechnology Co., Ltd., Shanghai, China) was diluted to the appropriate concentration and added to the cell culture dishes, following the recommended staining time according to the manufacturer's instructions. The cells were washed with PBS to remove excess dye. Staining results were observed and captured using a fluorescence microscope. FerroOrange emits fluorescence at a specific wavelength upon binding with Fe²⁺. The intensity of fluorescence was analysed to determine the content and distribution of Fe²⁺, using ImageJ for image processing and quantification. The number of positive cells was statistically analysed.

2.23. Protein Degradation Assay

Cells were pre‐seeded in 6‐well plates at a density of 5 × 10⁵ cells per well. After removing the old culture medium, 2 mL of fresh complete medium was added to each well. Cycloheximide (CHX) (HY‐12320, MCE, Shanghai, China) was added to each well at predetermined incubation time points (0, 3, 6, and 12 h). Specifically, 2 µL of 50 mg/mL CHX was added to each well at 12, 9, 6, and 0 h before the total incubation period of 12 h, resulting in a final concentration of 50 µg/mL. After a total incubation of 12 h, proteins were extracted from all samples for Western blot analysis. The relative grayscale values of the target protein were determined using β‐Actin expression as the internal control.

2.24. Western Blot

Proteins from cells were extracted using the Tissue Protein Rapid Extraction Kit (EX2410) and the Fixed Cell Protein Extraction Kit (EX2170) from Solarbio. Protein concentration was determined using the BCA Protein Assay Kit (Sigma, BCA1‐1KT). Equal amounts of protein (20 µg per lane) were separated by 10%–12% SDS‐PAGE and transferred to polyvinylidene fluoride (PVDF) membranes (EMD Millipore, Billerica, MA, USA). The membranes were blocked with 5% BSA for 2 h and washed with PBS. They were then incubated overnight at 4°C with primary antibodies (BRAF (ab33899, Abcam, Cambridge, UK); SLC7A11 (ab300667, Abcam, Cambridge, UK); ACSL4 (ab155282, Abcam, Cambridge, UK); GPX4 (ab125066, Abcam, Cambridge, UK); Alix (ab117600, Abcam, Cambridge, UK); TSG101 (ab30871, Abcam, Cambridge, UK); CD63 (ab271296, Abcam, Cambridge, UK); Calnexin (ab22595, Abcam, Cambridge, UK); HA (ab18181, Abcam, Cambridge, UK); Cas9 (ab189380, Abcam, Cambridge, UK); PCNA (ab29, Abcam, Cambridge, UK); Vimentin (Vim) (ab20346, Abcam, Cambridge, UK); N‐cadherin (ab18203, Abcam, Cambridge, UK); TFRC (ab84036, Abcam, Cambridge, UK); LC3B (ab51520, Abcam, Cambridge, UK); SQSTM1/p62 (ab91526, Abcam, Cambridge, UK); and β‐Actin (ab8227, Abcam, Cambridge, UK)) at dilutions recommended by the manufacturer. After washing, the membranes were incubated with goat anti‐rabbit HRP‐conjugated secondary antibody (Abcam, ab6721, 1:2000, UK) at room temperature for 2 h. Bands were visualised using the enhanced chemiluminescence system (Thermo Fisher, iBright FL1500). Band intensities were analysed using AlphaView SA software (Version 3.4.0). Relative protein expression levels were determined by normalising the target band intensity to the β‐Actin band intensity. Experiments were repeated three times.

2.25. RT‐qPCR Detection of Gene Expression in Tissues and Cells

Total RNA was extracted using Trizol reagent (15596026, Invitrogen, USA) according to the manufacturer's instructions. RNA was reverse transcribed into cDNA using the PrimeScript RT reagent Kit (RR047A, Takara, Japan). RT‐qPCR was performed on the synthesised cDNA using the Fast SYBR Green PCR Kit (11736059, Thermo Fisher Scientific, Shanghai, China) with triplicate for each sample. ACTB was used as the internal control. Relative gene expression levels were calculated using the 2^−ΔΔCt method. Experiments were repeated three times. The primer sequences used for RT‐qPCR are listed in Table S1, and they were synthesised by Takara.

2.26. T7 Endonuclease I (T7E1) Assay

Genomic DNA was extracted using 500 µL extraction buffer (10 mM Tris, pH 8; 2 mM EDTA; 0.2% Triton X‐100; 200 µg/mL Proteinase K) and incubated at 65°C for 15 min, followed by heat inactivation at 95°C for 5 min. Target regions were amplified by PCR using EconoTaq PLUS GREEN 2X Master Mix (Lucigen) or AccuPrime Taq DNA Polymerase High Fidelity (ThermoFisher) according to the manufacturers’ instructions. PCR products were denatured at 95°C for 10 min and re‐annealed by ramping down at −2°C/sec to 85°C and then −1°C/sec to 25°C. Heteroduplexed PCR products (5 µL) were incubated with 5 U T7E1 enzyme (New England Biolabs) at 37°C for 20 min. Digested products were analysed by electrophoresis on Novex 10% TBE gels (Invitrogen), and band density was quantified using ImageJ26 software. PCR products were also subjected to Sanger sequencing. The estimated percentage of NHEJ events was calculated using the following formula: %NHEJ events = 100 × [1 − (1 − fraction cleaved)], where the fraction cleaved is defined as: (density of digested products)/(density of digested products + undigested parental band).

2.27. Tumour Imaging, Biodistribution and In Vivo Targeting

8505C cells (1 × 10⁷ cells per mouse) were transplanted into the right flank of nude mice. When the 8505C tumours grew to approximately 10 mm in diameter, the mice were randomly divided into three groups for tumour targeting verification and in vivo distribution of TMTP1‐sgBRAF‐EVs (III). The control group, non‐targeted group, and targeted group were intravenously injected with saline, DiR‐labelled unmodified EVs, and DiR‐labelled TMTP1‐sgBRAF‐EVs (1 mg/mL, 200 µL per mouse), respectively. The protein concentration may vary depending on EV purity. Images of the mice were captured at 0, 1, 8, and 24 h post‐injection using the IVIS Spectrum Imaging System (Caliper Life Sciences, USA) (Wang et al. 2022c).

2.28. In Vivo Antitumour Efficacy

We established a nude mouse ATC model by subcutaneously injecting 8505C cells. When the tumours grew to approximately 8 mm in diameter, the tumour‐bearing nude mice were intravenously injected with equal amounts of saline (saline control), Vemurafenib (Vemurafenib‐treated group), sgRNA (EVs loaded with sgBRAF alone), sgBRAF (EVs loaded with pre‐formed Cas9/sgBRAF ribonucleoprotein [RNP] complexes), or TMTP1‐sgBRAF‐EVs (TMTP1‐sgBRAF‐EVs‐treated group) (1 mg/mL, 200 µL per mouse; n = 5). At the end of the treatment, the mice were euthanised, and the tumours were excised. The body weight and tumour volume of the mice were measured. Tumour volume was calculated using the formula V = (L × W²)/2, where L is the length and W is the width of the tumour measured with calipers. Tumour tissue samples were fixed in 4% PFA, dehydrated, cleared, paraffin‐embedded, and sectioned for histological analysis. The sections were subjected to H&E and IHC staining, using specific primary antibodies against BRAF (ab33899, Abcam, Cambridge, UK) and 4‐HNE (ab48506, Abcam, Cambridge, UK) (Wang et al. 2022).

2.29. Systemic Toxicity Assessment

The systemic toxicity of tumour EVs was further evaluated. We recorded the body weight changes of all nude mice after tumour implantation. Major organs were subjected to H&E staining to assess the potential adverse effects of tumour EVs. Blood samples were collected to perform a complete blood panel analysis and blood biochemical examination, including measurements of serum white blood cells (WBC), red blood cells (RBC), granulocytes (Gran), hematocrit (HCT), hemoglobin (HGB), mean corpuscular volume (MCV), monocytes (Mon), platelets (PLT), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), aspartate aminotransferase (AST), albumin (ALB), blood urea nitrogen (BUN), urea, and creatinine (CREA) levels. The concentrations of tumour necrosis factor‐α (TNF‐α) (JN22616, Jining Bio, Shanghai, China), IFN‐γ (JN18496, Jining Bio, Shanghai, China), interleukin‐10 (IL‐10) (JN20865, Jining Bio, Shanghai, China), and interleukin‐1β (IL‐1β) (JN19669, Jining Bio, Shanghai, China) were determined using ELISA. Absorbance at 450 nm was measured using a microplate reader.

2.30. Immunofluorescence

For cell samples, cells were fixed with 4% PFA at room temperature for 15 min, then blocked with 3% BSA at 37°C for 30 min to prevent nonspecific staining. Cells were incubated with primary antibodies overnight, washed three times with PBS for 3 min each, and then incubated with secondary antibodies at room temperature for 2 h. DAPI was used to counterstain the nuclei. For tumour tissue samples, tissues were fixed in 4% PFA, dehydrated, cleared, paraffin‐embedded, and sectioned. For immunofluorescence staining, sections were deparaffinised and rehydrated, followed by blocking with 2% BSA. Primary antibodies, including anti‐LAMP1 (ab278043, Abcam, Cambridge, UK) and anti‐TFRC (ab84036, Abcam, Cambridge, UK), were applied and incubated at 4°C overnight. The next day, sections were washed with PBS, incubated with appropriate secondary antibodies, and then observed under a fluorescence microscope. The next day, the sections were washed with PBS and incubated with goat anti‐mouse IgG (A10551, ThermoFisher Scientific Inc., Shanghai, China) or goat anti‐rabbit IgG (A‐11008, ThermoFisher Scientific Inc., Shanghai, China) secondary antibodies at room temperature for 1 h. After washing with PBS, the nuclei were stained with DAPI (C1002, Beyotime, China) for 5 min, followed by three washes with PBS for 5 min each to remove excess DAPI. Cells were carefully lifted from the culture plates using a fine curved needle and small tweezers and placed onto slides with an anti‐fade mounting medium (cells facing down). The slides were observed and photographed under a fluorescence microscope (FV‐1000/ES, Olympus, Japan). Quantification was performed by measuring the fluorescence coverage area in six fixed fields per group under a 40× objective lens, and the average value was calculated.

2.31. RNA Sequencing (RNA‐Seq)

RNA‐Seq (Illumina HiSeq) was performed on 8505c cells treated with sgNC and sgBRAF. Sequencing libraries were generated using the NEBNext Ultra RNA Library Prep Kit for Illumina (E7530, NEB, Beverly, MA). Quality control was performed using RNA‐SeQC v1.1.8, and read counts were obtained using HTSeqcounts v0.7.2. Differentially expressed genes (DEGs) between the two groups were identified using the limma package in R, with |log₂FoldChange| > 1 and p‐value < 0.05 as the cutoff criteria. KEGG enrichment analysis was conducted using the ClusterProfiler package (version 3.18.1, Bioconductor, USA) to identify biological processes and pathways related to ferroptosis.

2.32. Mutation Data Analysis of Thyroid Cancer

Single‐nucleotide variation data for the TCGA‐Thyroid Cancer (Rectal Adenocarcinoma) dataset were downloaded from the TCGA website. ‘Masked Somatic Mutation’ data were selected for analysis. The mutation annotation format (MAF) files were analysed and visualised using the ‘maftools’ package in R.

2.33. Statistical Analysis

Data were derived from at least three independent experiments and are presented as mean ± standard deviation (Mean ± SD). Comparisons between the two groups were performed using an independent‐sample t‐test. Repeated measures ANOVA was used to compare data at different time points. For comparisons among three or more groups, one‐way ANOVA was employed, and if significant differences were detected, Tukey's HSD post hoc test was conducted to compare differences between groups. For non‐normally distributed data or those with unequal variances, the Mann‐Whitney U test or Kruskal‐Wallis H test was used. All statistical analyses were performed using GraphPad Prism 9 (GraphPad Software, Inc.) and R. A two‐sided p‐value less than 0.05 was considered statistically significant, with *p < 0.05, *p < 0.01, and **p < 0.001 indicating increasing levels of significance.

3. RESULTS

3.1. BRAF as a Key Mediator of ATC Resistance

To investigate the gene mutation landscape in thyroid cancer, we first downloaded somatic mutation profiles of 496 thyroid cancer patients from TCGA and visualised the mutation data using the ‘maftools’ package. As shown in Figure S1A, the mutation information for each gene in each sample is displayed as a waterfall plot. The results indicated that somatic mutations were detected in 377 out of 496 thyroid cancer samples (377/496, 76.01%), with BRAF mutations accounting for 59% of the total mutations. Based on the cBioPortal database, we found that BRAF‐mutant samples exhibited higher mRNA levels of BRAF (Figure S1B). ATC is the most aggressive subtype of thyroid cancer, characterised by resistance and non‐responsiveness to all treatments and surgical interventions. Research on BRAF‐mutant thyroid cancer plays a crucial role in tumour recurrence, resistance, and metastasis. Additionally, the V600E mutation in BRAF showed the highest efficiency (Figure S1C). Although selective BRAF inhibitors (V600E) have been developed, they have shown poor responses in clinical trials for ATC (Díaz‐Gago et al. 2024).

We detected BRAF protein levels in normal human thyroid cells (Nthy‐ori3‐1), BRAF‐mutant ATC cell lines (8505c (BRAF^V600E) and BHT‐101 (BRAF^V600E)), and wild‐type ATC cell lines (CAL‐62 (BRAF^WT) and KMH‐2 (BRAF^WT)). As shown in Figure S2A, compared to the Nthy‐ori3‐1 group, BRAF protein levels were significantly elevated in BRAF‐mutant ATC cell lines (8505c (BRAF^V600E) and BHT‐101 (BRAF^V600E)) and wild‐type ATC cell lines (CAL‐62 (BRAF^WT) and KMH‐2 (BRAF^WT)), with 8505c (BRAFV600E) having the highest expression level, an increase of 179.48%.

Among these, the highest expression level was observed in 8505c (BRAF^V600E). This indicates the importance of BRAF mutations in the drug resistance mechanisms of ATC. We treated 8505c (BRAF^V600E), BHT‐101 (BRAF^V600E), CAL‐62 (BRAF^WT), and KMH‐2 (BRAF^WT) cell lines with varying concentrations of the B‐Raf kinase inhibitor Vemurafenib (PLX4032). The results, shown in Figure 1A, indicate that BRAF‐mutant ATC cell lines (8505c (BRAF^V600E) and BHT‐101 (BRAF^V600E)) exhibit stronger resistance to Vemurafenib treatment compared to wild‐type ATC cell lines. Furthermore, we compared BRAF expression in 8505c and BHT‐101 cell lines with and without Vemurafenib treatment. The results, shown in Figure S2B, demonstrate that Vemurafenib treatment significantly reduced BRAF expression in 8505c and BHT‐101 cell lines compared to the control group.

FIGURE 1.

Identification of BRAF as a key mediator of ATC resistance using CRISPR‐Cas9. (A) Cell viability of BRAF wild‐type and mutant cells treated with various concentrations of Vemurafenib, measured by CCK‐8 assay; (B) CCK‐8 assays were also used to compare viability in BHT‐101 and 8505C cells transfected with BRAF‐KO sgRNAs versus control sgRNAs; (C) Colony formation assays were performed to evaluate the clonogenic ability of BHT‐101 and 8505C cells after BRAF knockout using different sgRNAs; (D)–(E) Transwell assays assessed the migration and invasion of BHT‐101 and 8505C cells after BRAF knockout. Statistical analyses were conducted using one‐way ANOVA for comparisons among three or more groups. *** denotes p < 0.001.

In recent years, the CRISPR/Cas genome editing system has been widely applied for its ability to produce targeted genome edits (Huang et al. 2018). In this study, we used CRISPR/Cas9 gene editing technology to generate stable BRAF knockout subclones in BRAF‐mutant ATC cell lines 8505c (BRAF^V600E) and BHT‐101 (BRAF^V600E) (Figure S2C) and analysed their sensitivity to Vemurafenib. The results showed that compared to the sgNC group, the sgBRAF‐2 and sgBRAF‐3 groups exhibited significantly decreased cell viability, colony formation ability, migration, and invasion, both with and without Vemurafenib treatment (Figure 1B–E). These findings indicate that CRISPR/Cas9‐mediated BRAF knockout can inhibit the resistance response in BRAF‐mutant ATC cells.

3.2. BRAF Deletion Contributes to the Resistance of BRAF‐Mutant ATC Cells to Drug‐Induced Ferroptosis

To further investigate the mechanism by which BRAF deletion inhibits drug resistance in BRAF‐mutant ATC cells, we performed KEGG enrichment analysis on DEGs following BRAF deletion based on transcriptome sequencing. The analysis revealed enrichment of ferroptosis‐related genes after BRAF knockout (Figure S3A,B). Recent studies have shown that Vemurafenib induces potent ferroptosis, which is a key component of its antitumour effects (Tang et al. 2020). In fact, we observed that Vemurafenib induces lipid peroxidation and ROS generation (Figure S4A,B). TEM demonstrated that cancer cells treated with Vemurafenib exhibited morphological characteristics of ferroptosis, such as mitochondrial shrinkage and increased membrane density (Figure S4C). Next, we investigated the potential role of ferroptosis in Vemurafenib‐induced cell death. To this end, we analysed the effects of the ROS scavenger NAC, apoptosis inhibitor ZVAD‐fmk, necroptosis inhibitor necrostatin‐1, and ferroptosis inhibitor ferrostatin‐1 on the survival of BRAF‐mutant ATC cells treated with Vemurafenib. We observed that exposure to Vemurafenib impaired clonal survival in the tested BRAF‐mutant ATC cells. However, this impairment was partially reversed by treatment with ferrostatin‐1 or NAC (Figure S4D). Studies have shown that GPX4 utilises glutathione (GSH) as a substrate to reduce toxic lipid hydroperoxides to non‐toxic lipid alcohols, thereby preventing the propagation of lipid peroxidation. Inhibition or downregulation of GPX4 leads to the accumulation of lipid peroxides and ultimately triggers ferroptosis. Similarly, SLC7A11 transports extracellular cystine into the cell, where it is reduced to cysteine, a rate‐limiting precursor for GSH synthesis. Since GSH is essential for GPX4 activity, SLC7A11 indirectly regulates ferroptosis through GSH synthesis. Suppression of SLC7A11 reduces cystine uptake, lowers GSH levels, and decreases GPX4 activity, thereby sensitising cells to ferroptosis (Li et al. 2021c). ACSL4 preferentially catalyses polyunsaturated fatty acids such as arachidonic and adrenic acid into phospholipids, which are substrates for lipid peroxidation. Overexpression of ACSL4 increases these phospholipids and enhances ferroptosis, whereas ACSL4 deficiency reduces ferroptosis sensitivity (Doll et al. 2016). We demonstrated that Vemurafenib significantly inhibited the expression of GPX4 and SLC7A11 while increasing the expression of ACSL4 (Figure 2A). Erastin, a ferroptosis inducer, primarily functions by inhibiting SLC7A11, blocking cystine uptake, leading to reduced GSH and impaired GPX4 activity, which triggers ferroptosis (Zhao et al. 2020). RSL3 is another ferroptosis inducer that directly targets and degrades GPX4 via the ubiquitin–proteasome pathway, resulting in increased lipid peroxidation and ferroptotic cell death (Li et al. 2021c). Additionally, ferroptosis‐resistant BHT‐101 and 8505c cells were obtained by continuous exposure to erastin and RSL3. Notably, BRAF‐mutant ATC cells resistant to RSL3 and erastin also exhibited resistance to Vemurafenib (Figure S4E,F). Our data strongly suggest that Vemurafenib induces ferroptosis in BRAF‐mutant ATC cells. Furthermore, our findings indicate that ferroptosis plays a crucial role as a component of the cell death response triggered by Vemurafenib.

FIGURE 2.

Regulation of Ferroptosis in BRAF‐Mutant ATC cells. (A) Western blot detection of SLC7A11, ACSL4, and GPX4 in BHT‐101 and 8505C cells treated with 5 µM Vemurafenib. Two‐group comparisons used unpaired t‐tests; (B) Live/dead staining of cells treated with Vemurafenib, sgBRAF, or both. Scale bar = 25 µm; (C) TEM images showing ferroptotic morphology in sgBRAF‐treated BHT‐101 and 8505C cells. Scale = 1 µm/500 nm; (D) Fluorescence microscopy of lipid peroxidation after different treatments. Scale = 25 µm; (E)–(F) C11‐BODIPY and flow cytometry were used to detect lipid ROS accumulation; (G) GSH level detection; (H) MDA level detection; (I) FerroOrange detection of intracellular Fe²⁺. Scale = 25 µm. p < 0.05 indicates significance. All experiments were repeated at least three times. Two‐group comparisons used unpaired t‐tests; for ≥3 groups, one‐way ANOVA was applied. *p < 0.05, ***p < 0.001.

We further investigated whether BRAF regulates the sensitivity of BRAF‐mutant ATC cells to Vemurafenib by modulating ferroptosis. We confirmed that BRAF deletion enhanced the efficacy of Vemurafenib in erastin‐ and RSL3‐resistant BRAF‐mutant ATC cells (Figure S4G,H). Calcein‐AM/PI staining was used to assess cell viability, and the results showed that Vemurafenib treatment significantly increased cell death, an effect that was further enhanced by BRAF deletion. Co‐treatment with Vemurafenib and sgBRAF raised the cell death rate to around 40% (Figure 2B). TEM revealed that cells treated with Vemurafenib exhibited typical ferroptosis morphological features, such as mitochondrial shrinkage and increased membrane density. These ferroptosis characteristics were more pronounced in BRAF‐deleted cells, further confirming the role of BRAF in regulating ferroptosis (Figure 2C). Changes in lipid peroxidation in ATC cells were also observed using a fluorescence microscope (Figure 2D). C11‐BODIPY staining indicated that lipid peroxidation increased Vemurafenib, and was further elevated by BRAF deletion (Figure 2E). ROS levels detected by ROS‐Violet 610 also increased significantly with Vemurafenib and even more so in combination with sgBRAF (Figure 2F). The GSH/GSSG ratio decreased significantly following Vemurafenib treatment and was further reduced by sgBRAF (Figure 2G). MDA levels rose following treatment and were further increased by BRAF knockout (Figure 2H). Using FerroOrange probes, intracellular Fe²⁺ concentration was found to increase Vemurafenib and further elevated with sgBRAF (Figure 2I).

To further verify the role of BRAF in ferroptosis, we used the ferroptosis inhibitor ferrostatin‐1 (fer). We found that BRAF deletion significantly increased Vemurafenib‐induced cell death (Figure S5A), reduced glutathione (GSH) levels (Figure S5B), and increased lipid peroxidation products (MDA) levels (Figure S5C). However, in cells treated with ferrostatin‐1, BRAF deletion had no additional effect on cell death, GSH levels, or MDA levels. Additionally, ferrostatin‐1 treatment restored the decreased clonogenic capacity of BRAF‐mutant ATC cells caused by BRAF deletion (Figure S5D). Ferrostatin‐1 also significantly enhanced the migration and invasion abilities of BRAF‐deleted BRAF‐mutant ATC cells (Figure S5E,F). These results collectively indicate that BRAF plays a crucial role in regulating ferroptosis and Vemurafenib resistance. The combined use of Vemurafenib and sgBRAF can significantly improve the therapeutic efficacy in BRAF‐mutant ATC cells.

3.3. TMTP1‐sgBRAF‐EVs as an Effective Protective Delivery Method for In Vivo Gene Editing

Although CRISPR‐Cas9 represents a significant advancement in gene editing technology, effectively delivering it to tumour sites using existing delivery methods remains challenging. A critical gap in the therapeutic translation of CRISPR editing is the development of an efficient and safe tumour delivery system. To overcome the limitations of Cas9/sgRNA complex delivery, we developed a method to load Cas9/sgRNA into EVs for the treatment of various diseases. Previous studies have shown that tumour cells produce and secrete more nucleic acids, proteins, and lipids than healthy cells. Some of these molecules are transported in the blood or encapsulated in EVs, such as sEVs, and released into the extracellular environment. Once released, these sEVs become part of a communication network used by other tumour cells and organs (Sun et al. 2018; Mirzaei et al. 2017; Qiu et al. 2019). Previous studies have shown that EVs exhibit organ‐ or cell‐specific homing behaviour (Petroni et al. 2023). To investigate whether ATC tumour EVs can be home to parental cells in vitro, we isolated EVs from different BRAF‐mutant ATC cell lines, 8505c and BHT‐101. Western blot analysis showed that known EV markers Alix and TSG101 were enriched in EV particles, whereas Calnexin, an endoplasmic reticulum marker, was not detected in the EVs (Figure S6A). We then co‐cultured 8505c cells with DiR‐labelled 8505c EVs or DiR‐labelled BHT‐101 EVs for 12 h and quantified EV uptake by 8505c cells using fluorescence microscopy. We observed that the uptake of 8505c EVs by 8505c cells was higher compared to the uptake of BHT‐101 EVs (Figure 3A,B). To ensure this finding was not specific to 8505c cells, we tested whether BHT‐101 EVs also exhibited a preference for their parental cells. We co‐cultured 8505c and BHT‐101 cells with red (DiR)‐labelled BHT‐101 EVs and green (DiO)‐labelled 8505c EVs (Figure S6B). Although both types of EVs were internalised by BHT‐101 and 8505c cells, the uptake of BHT‐101‐EVs was significantly higher in BHT‐101 cells, and the uptake of 8505c‐EVs was significantly higher in 8505c cells (Figure S6C,D). We then reversed the dye labelling of the EVs to ensure that the uptake differences were not due to labelling discrepancies (Figure S6E). Consequently, the cell‐specific localisation of EVs matched the origin of the vesicle‐producing cells, indicating that parental cancer cells more effectively uptake EVs from their own origin compared to those from other sources.

FIGURE 3.

Preparation and characterisation of tumour‐derived EVs Encapsulating Cas9 Protein and sgRNA. (A) Schematic design of the in vitro study evaluating the homing ability of cancer‐derived EVs; (B) Representative microscopy images showing the uptake of DiR‐labelled 8505c‐EVs and BHT‐101‐EVs by 8505c cells, with quantitative analysis of EV uptake. Endocytosed EVs (red) are visible around the cell nucleus, scale bar = 25 µm; (C) Representative ex vivo fluorescence imaging of major organs and tumour tissues 12 h post‐EV injection, along with statistical analysis of in vivo EV biodistribution. p/s: photons/s; (D) Schematic showing the engineering of TMTP1‐conjugated EVs with tumour‐penetrating capability; (E‐F) Western blot analyses detecting EV markers and Cas9 protein loading efficiency in TMTP1‐modified EVs; (G) Proteinase K and RNase protection assay: TMTP1‐sgBRAF‐EVs treated with/without 0.25% Triton X‐100 and with or without proteinase K (0.5 µg/mL) for 30 min, followed by Western blotting to detect SpCas9 and CD9; (H) RNase A protection assay confirming the encapsulation of sgRNA within EVs. TMTP1‐sgBRAF‐EVs were treated with RNase A alone or in combination with Triton X‐100, and the relative sgRNA levels were quantified by RT‐qPCR. p < 0.05 was considered significant. Animal experiments were conducted with n = 3. ***p < 0.001.

To confirm this finding in vivo, we established an 8505c tumour model in immunodeficient nude mice and intravenously injected DiR‐labelled 8505c or BHT‐101 EVs (Figure S6F). The biodistribution of the EVs was detected using the IVIS imaging system. The 8505c EVs are preferentially localised to the tumour tissue. Compared to BHT‐101 EVs, the expression of the fluorescent signal from 8505c EVs was higher (Figure 3C). In summary, our data suggest that cancer‐derived EVs preferentially return to their parental cancer cells, indicating that tumour cells are suitable for use in tumour‐targeted therapy.

Although 8505c‐EVs exhibit stronger homing abilities, more EVs accumulate in the liver than in the tumour. Therefore, we further modified 8505c‐derived tumour cells with the tumour‐targeting peptide TMTP1 (Jiang et al. 2021; Wei et al. 2019) to enhance their regulation of tumour cells. To achieve stable expression of TMTP1‐Lamp2b‐HA protein, 8505c cells were infected with lentivirus particles (Figure 3D). Western blot showed significant expression of TMTP1‐Lamp2b‐HA in transfected 8505c cells (Figure 3E). In preliminary experiments, we attempted to load Cas9 protein/sgRNA complexes into purified EVs using electroporation, freeze‐thaw cycles, and ultrasonication. Electroporation induces transient pores in EV membranes, allowing efficient RNP loading under optimised conditions. This process is likely driven by a combination of electrostatic interactions and osmotic pressure differentials. Western blot demonstrated that Cas9 protein was effectively loaded into isolated EVs by electroporation (Figure S7A). The morphology of unmodified EVs and TMTP1‐sgBRAF‐EVs was evaluated using TEM (Figure S7B). Specifically, unmodified EVs appeared round with an average diameter of 103 nm. After TMTP1 conjugation, the average diameter of the EVs increased to 132 nm (Figure S7C). Next, NTA was used to assess the size distribution of both unmodified and modified EVs, showing peaks around 100 nm for both types (Figure S7D). To evaluate the stability of TMTP1‐sgBRAF‐EVs over time, they were stored in phosphate‐buffered saline (PBS, 1×, pH 7.4) at 4°C for 7 days. To simulate their eventual in vivo environment, TMTP1‐sgBRAF‐EVs were placed in fresh serum at 37°C for 24 h. We used NanoSight to monitor the stability of TMTP1‐sgBRAF‐EVs in both environments. The results showed no significant aggregation of TMTP1‐sgBRAF‐EVs in either PBS buffer or serum (Figure S7E,F), indicating good stability. Western blot analysis revealed that the loading efficiency of Cas9 protein in TMTP1‐sgBRAF‐EVs was nearly 40% (Figure 3F). Based on Figure 3F, 160 ng of Cas9 was detected in TMTP1‐sgBRAF‐EVs after loading with 400 ng input protein, indicating 40% encapsulation efficiency. Electroporation is a key method for loading Cas9‐gRNA RNPs. During electroporation, a strong transient electric field disrupts the EV phospholipid bilayer to create temporary pores, facilitating the entry of Cas9‐gRNA RNPs. Electrostatic interactions between EV surface charges and the RNPs may also assist in targeting and internalisation.

To determine whether the SpCas9/sgRNA complex was encapsulated within extracellular vesicles (EVs) or merely attached to their surface, we performed protease protection and RNase A assays. As shown in Figure 3G, treatment of TMTP1‐sgBRAF‐EVs with proteinase K in the absence of Triton X‐100 did not abolish the SpCas9 band, indicating that the protein was protected from degradation by the EV membrane. In contrast, disruption of the EV membrane with Triton X‐100 rendered SpCas9 susceptible to proteinase K digestion, whereas the membrane protein CD9 remained stable under all treatment conditions, further supporting that SpCas9 is enclosed inside the vesicles. Moreover, to assess whether sgRNA was similarly encapsulated, we carried out RNase A protection experiments (Figure 3H). sgRNA levels remained largely unchanged following RNase A treatment alone; however, a marked reduction was observed when RNase A was combined with Triton X‐100, indicating that sgRNA is likewise shielded within the EV lumen and protected by the intact membrane.

These results indicate that TMTP1‐sgBRAF‐EVs may serve as a suitable protective delivery system for in vivo gene editing.

3.4. Enhanced Vemurafenib Sensitivity via TMTP1‐sgBRAF‐EVs Promoting Ferroptosis

To evaluate the transport capacity of peptide‐modified EVs, we co‐cultured TMTP1‐sgBRAF‐EVs with 8505c cells. Fluorescence microscopy imaging revealed that TMTP1‐sgBRAF‐EVs exhibited the highest targeting efficiency compared to unmodified EVs (Figure 4A). Furthermore, the subcellular analysis showed co‐localisation of eSpCas9‐GFP and sgBRAF‐cy3 in 8505c cells (Figure 4A). To investigate whether TMTP1‐sgBRAF‐EVs could modulate the sensitivity of BRAF‐mutant ATC cells to Vemurafenib, we conducted colony formation, migration, and invasion assays. The results demonstrated that the combination of Vemurafenib with either sgBRAF or TMTP1‐sgBRAF‐EVs significantly inhibited colony formation (Figure S8A), migration (Figure S8B), and invasion (Figure S8C) of 8505c cells compared to Vemurafenib alone. Moreover, Western blot analysis revealed that treatment with TMTP1‐sgBRAF‐EVs significantly reduced the expression of proliferation marker PCNA, Vim, and N‐cadherin (N‐cad) (Figure S8D).

FIGURE 4.

TMTP1‐sgBRAF‐EVs induce ferroptosis in BRAF^mut ATC cells. (A) Immunofluorescence images showed cellular uptake of unmod‐EVs and TMTP1‐sgBRAF‐EVs by 8505C cells. Scale bar = 25 µm; (B) Viability analysis (live/dead staining) following EV uptake. Scale = 25 µm; (C) TEM revealed classical ferroptotic morphological features in BRAF‐deficient 8505C cells after uptake; (D‐E) C11‐BODIPY and fluorescence microscopy detected increased lipid peroxidation after treatment with unmod‐EVs, sgBRAF, and TMTP1‐sgBRAF‐EVs. Scale = 25 µm; (F) Flow cytometry showed increased membrane lipid ROS accumulation; (G‐H) GSH and MDA levels were measured after treatment; (I) FerroOrange probe detected increased intracellular Fe²⁺. Scale = 25 µm. Animal experiments were conducted with n = 3. One‐way ANOVA was used for multi‐group comparisons. ***p < 0.001.

Using Calcein AM/Propidium Iodide (PI) staining, we detected live and dead cells in 8505c cells. The results showed a significant increase in PI‐positive (PI⁺) cells following treatment with a combination of Vemurafenib and either sgBRAF or TMTP1‐sgBRAF‐EVs compared to Vemurafenib alone (Figure 4B). TEM revealed enhanced ferroptosis morphological features in cells treated with both TMTP1‐sgBRAF‐EVs and Vemurafenib (Figure 4C). Staining with the C11‐BODIPY probe demonstrated that the combination treatments significantly increased lipid peroxidation levels compared to Vemurafenib alone (Figure 4D). Fluorescence microscopy further confirmed that TMTP1‐sgBRAF‐EVs significantly elevated lipid peroxidation in 8505c cells (Figure 4E). Flow cytometry analysis showed that treatment with a combination of Vemurafenib and either sgBRAF or TMTP1‐sgBRAF‐EVs significantly increased lipid ROS accumulation on the cell membrane compared to Vemurafenib alone (Figure 4F). In both BHT‐101 and 8505c cells, the combination treatments significantly reduced GSH levels (Figure 4G) and elevated levels of the lipid peroxidation product MDA (Figure 4H) compared to Vemurafenib alone. Using the FerroOrange probe to detect intracellular Fe²⁺ concentrations, we observed a significant increase in the labile iron pool following combination treatments compared to Vemurafenib alone (Figure 4I), confirming the role of BRAF in iron homeostasis. TMTP1‐sgBRAF‐EVs (IV) induced significantly higher levels of PI‐positive cells, lipid peroxidation, ROS, MDA, and intracellular Fe²⁺, and lower GSH levels compared to sgBRAF alone, indicating enhanced ferroptosis. These results suggest that TMTP1 modification not only improved the delivery efficiency of sgBRAF but also promoted ferroptosis in target cells by increasing oxidative stress and iron accumulation. This enhanced ferroptosis induction likely contributes to the increased sensitivity of BRAFV600E‐mutant ATC cells to Vemurafenib.

In summary, TMTP1‐sgBRAF‐EVs enhance the sensitivity of BRAF‐mutant ATC cells to Vemurafenib by promoting ferroptosis. These findings provide new evidence supporting TMTP1‐sgBRAF‐EVs as a potential therapeutic approach to improve the response of BRAF‐mutant ATC cells to Vemurafenib by targeting the ferroptosis pathway.

3.5. TMTP1‐sgBRAF‐EVs Significantly Enhance Vemurafenib Efficacy by Promoting Ferroptosis

We further investigated the biodistribution and tumour‐targeting capability of TMTP1‐sgBRAF‐EVs by constructing a subcutaneous xenograft model using 8505c cells in nude mice. After intravenous injection of saline or EVs (TMTP1‐sgBRAF‐EVs and unmodified EVs) followed by Vemurafenib administration, fluorescence images were captured at different time points using the IVIS spectrum system (Figure 5A). The results showed that the tumour fluorescence signal was stronger in the TMTP1‐sgBRAF‐EVs group compared to the unmodified EVs group (Figure 5B). Post‐imaging, the organs and tumours of the mice were excised for ex vivo imaging. The luminescence channel indicated the tumour location and displayed the distribution of EVs in various organs, showing a stronger fluorescence signal at the tumour site in the TMTP1‐sgBRAF‐EVs group compared to the unmodified EVs group (Figure 5C). Although TMTP1‐modified EVs improved tumour‐targeting specificity, unmodified EVs exhibited broader distribution and overall higher uptake. Despite lower total fluorescence, TMTP1‐sgBRAF‐EVs showed more targeted accumulation in tumours. Tumour tissues from each group were then cryosectioned and observed under a fluorescence microscope. Cas9 was labelled with GFP, sgBRAF with Cy3, and EVs with DiR. In the TMTP1‐sgBRAF‐EVs treatment group, tumour tissues exhibited more GFP, Cy3 and DiR fluorescence, indicating superior penetration and accumulation of TMTP1‐sgBRAF‐EVs within the tumour tissue compared to the other groups. Notably, TMTP1‐sgBRAF‐EVs were predominantly distributed within the tumour boundary, demonstrating their enhanced tumour‐penetrating ability (Figure 5D). Compared to unmodified EVs, TMTP1‐sgBRAF‐EVs were more effective in targeting the tumour core. Subsequent RT‐qPCR revealed that BRAF mRNA levels in tumours decreased by more than 50% after EV injection (Figure 5E), consistent with Western blot analysis (Figure 5E). TMTP1‐sgBRAF‐EVs were more effective in penetrating the tumour area than unmodified EVs. In addition, since the CRISPR‐Cas9 system can induce non‐homologous end joining (NHEJ) repair events, resulting in insertions or deletions of nucleotides within the target gene sequence, we performed a T7 Endonuclease I (T7E1) assay to detect the frequency of NHEJ events at the BRAF locus in each group. The results showed that treatment with TMTP1‐sgBRAF‐EVs generated approximately 50% NHEJ events, which was significantly higher than the ∼20% observed with sgBRAF alone, while other groups showed minimal NHEJ activity (Figure 5F). Further Sanger sequencing validation revealed clear overlapping peaks at the sgBRAF target site in tumour tissues treated with TMTP1‐sgBRAF‐EVs, indicating the occurrence of insertion/deletion (indel) editing events and further confirming that sgBRAF delivered by EVs exerted functional activity in vivo (Figure 5G).

FIGURE 5.

Biodistribution of TMTP1‐sgBRAF‐EVs and their impact on the efficacy of vemurafenib treatment. (A) Schematic representation of the experimental design. A subcutaneous tumour model was established in nude mice by injecting 8505C cells. Mice received tail vein injections of EVs (1 mg/mL, 200 µL TMTP1‐sgBRAF‐EVs or unmodified EVs) and 5 µM Vemurafenib. After seven treatment cycles, bioluminescence imaging was performed; (B) IVIS Spectrum system captured fluorescent images of mice at different time points following injection of EVs (1 mg/mL, 200 µL TMTP1‐sgBRAF‐EVs or unmod‐EVs) and 5 µM Vemurafenib; (C) Ex vivo imaging of organs and tumours revealed the distribution of EVs in different tissues after administration; (D) Fluorescence microscopy of frozen tumour sections showed EV localisation. Scale bars = 50 µm; (E) RT‐qPCR and Western blot analysis of BRAF mRNA and protein expression levels in each group; (F) T7E1 assay detected NHEJ events at the BRAF locus; black arrows indicate cleaved fragments; (G) Sanger sequencing results showing indel‐associated overlapping peaks at the BRAF target site in the TMTP1‐sgBRAF‐EVs treatment group; (H) tumour volume measurements of treated nude mice; (I) GSH levels measured post‐treatment; (J) MDA levels, a marker of lipid peroxidation, measured post‐treatment; (K) Immunohistochemistry revealed reduced BRAF protein and increased 4‐HNE (a lipid peroxidation biomarker) following combined treatment with Vemurafenib and TMTP1‐sgBRAF‐EVs. Scale bar = 50 µm. Animal experiments used n = 5 per group. “Only” refers to treatment with Vemurafenib alone. One‐way ANOVA was applied for comparisons among three or more groups. **p < 0.01, ***p < 0.001.

We next assessed whether TMTP1‐sgBRAF‐EVs could enhance the sensitivity of Vemurafenib in an in situ 8505c model. Cas9 protein and sgBRAF gRNA were co‐administered at precise doses. Systemic administration of TMTP1‐sgBRAF‐EVs (administered weekly via tail vein) improved the therapeutic efficacy of Vemurafenib. The combination of Vemurafenib and TMTP1‐sgBRAF‐EVs was more effective than the combination of Vemurafenib and sgBRAF alone, resulting in a statistically significant reduction in tumour volume during the combination treatment (Figure 5H). Notably, the combination of Vemurafenib and TMTP1‐sgBRAF‐EVs led to a sharp decrease in GSH levels and an increase in lipid peroxidation products (Figure 5I,J). Consistent with the observed tumour growth inhibition, immunohistochemical staining showed that the combination treatment resulted in reduced BRAF protein levels. 4‐HNE, a biomarker of lipid peroxidation and ferroptosis, was significantly increased in the combination group. The combination therapy significantly increased the intensity of 4‐HNE staining (Figure 5K). In summary, our findings suggest that Vemurafenib may benefit from TMTP1‐sgBRAF‐EVs‐induced ferroptosis, enhancing its antitumour efficacy.

To evaluate the immune response and toxicity induced by TMTP1‐sgBRAF‐EVs, healthy BALB/c nude mice were intravenously injected with TMTP1‐sgBRAF‐EVs three times every other day. Throughout the treatment, mice receiving TMTP1‐sgBRAF‐EVs exhibited blood parameters and biochemical characteristics similar to those of mice given saline (Figure S9A,B). The mice also maintained their body weight (Figure S9C), indicating minimal impact on haematological parameters or liver and kidney functions. A comparison of plasma levels of various cytokines, including IL‐10, IL‐1β, TNF‐α and IFN‐γ, revealed no significant changes (Figure S9D). These results suggest that systemic administration of TMTP1‐sgBRAF‐EVs at therapeutically relevant doses is neither toxic nor immunogenic.

3.6. BRAF Deletion Promotes Ferroptosis by Inhibiting Autophagy‐Mediated TFRC Degradation

To further elucidate the specific mechanisms of ferroptosis, we analysed changes in ferroptosis‐related genes in BRAF‐deficient 8505c cells. In 8505c‐sgBRAF cells, we observed an increase in glutamate‐cysteine ligase catalytic subunit (GCLC), solute carrier family 39 member 14 (SLC39A14/Zip14), ferritin heavy chain 1 (FTH1), heme oxygenase 1 (HMOX1), and TFRC, while solute carrier family 40 member 1 (SLC40A1) and solute carrier family 11 member 2 (SLC11A2/DMT1) were decreased (Figure 6A). TFRC is a crucial mediator that transports iron(III) bound to transferrin (TF) into cells (Feng et al. 2020). Once inside the cell, Fe³⁺ is reduced to Fe²⁺ by the metalloreductase STEAP3 within endosomes and then transported into the cytoplasm via SLC11A2. SLC40A1 serves as the primary channel for exporting Fe²⁺ from the cell membrane (Montalbetti et al. 2013), while SLC39A14 is involved in the uptake of non‐TF‐bound iron by cells (Zhao et al. 2017; Pinilla‐Tenas et al. 2011). To validate the changes in gene expression, we conducted RT‐qPCR assays in BHT‐101 and 8505c cells. The results showed that BRAF deletion significantly upregulated the mRNA levels of TFRC, FTH1, and SLC39A14 while downregulating the mRNA levels of SLC40A1 and SLC11A2 (Figure 6B). Next, we examined the expression levels of TFRC, an essential protein for cellular iron uptake. Western blot analysis confirmed that TFRC protein expression remained significantly upregulated in BRAF‐deficient cells even after 48 h of treatment with ferrostatin‐1 (1 µM) or deferoxamine (DFO, 200 µM) (Figure 6C). We also used CHX to inhibit protein synthesis and evaluate the half‐life of the TFRC protein. The results indicated that BRAF deletion significantly prolonged the half‐life of the TFRC protein (Figure 6D). Therefore, we hypothesise that the elevated expression of TFRC might be responsible for the ferroptosis observed in sgBRAF cells, as BRAF deletion appears to inhibit the degradation of TFRC protein in ATC cells.

FIGURE 6.

Impact of BRAF deletion on the expression of Ferroptosis‐related genes and proteins in 8505c and BHT‐101 cells. (A) Heatmap showing the changes in ferroptosis‐related gene expression in 8505C cells after BRAF deletion; (B) RT‐qPCR detection of mRNA levels of ferroptosis‐related genes in 8505c and BHT‐101 cells following BRAF deletion; (C) Western blot analysis of the effect of BRAF deletion on TFRC and BRAF protein expression after treatment with ferrostatin‐1 (1 µM) or DFO (200 µM) for 48 h; (D) Measurement of the half‐life of TFRC protein in sgBRAF and sgNC cells after inhibition of intracellular protein synthesis with CHX. Repeated measures ANOVA was used to compare data over time; (E) Western blot detection of LC3B and SQSTM1 protein levels; (F) Immunofluorescence analysis showing changes in colocalisation of TFRC with lysosomal marker LAMP1. Scale bar = 25 µm. (G) Western blot showing increased TFRC protein accumulation in BRAF KO cells treated with chloroquine or bafilomycin A1. All cell experiments were repeated at least three times. Two‐group comparisons used unpaired t‐tests. One‐way ANOVA was applied for comparisons among three or more groups. “ns” indicates no significance. ***p < 0.001.

Research indicates that TFRC degradation can occur through autophagy (Xiong et al. 2021; Shan et al. 2023). Based on this, we hypothesised that BRAF deletion might prevent TFRC degradation by inhibiting autophagy‐mediated pathways. To test this hypothesis, we examined the levels of autophagy‐related protein LC3B and the autophagy receptor SQSTM1. The results showed that LC3B and SQSTM1 protein expression levels were significantly higher in BRAF‐deficient cells compared to control cells. However, treatment with the autophagy inducer Rap or the autophagy inhibitor CQ did not change the expression levels of LC3B and SQSTM1 in BRAF‐deficient cells compared to the control group (Figure 6E). Immunofluorescence analysis revealed reduced co‐localisation of TFRC with the lysosomal marker LAMP1 in BRAF‐deficient cells (Figure 6F), further supporting that BRAF deletion increases TFRC stability by inhibiting its autophagy‐mediated degradation. We further treated BRAF knockout (KO) cells with chloroquine or bafilomycin A1 (both of which inhibit autophagic degradation), followed by Western blot analysis to assess TFRC levels. The results showed that TFRC accumulation increased in BRAF‐deficient cells treated with these inhibitors (Figure 6G). These findings suggest that the changes in ferroptosis‐related gene expression induced by BRAF deletion, particularly the inhibition of autophagy‐mediated TFRC degradation, lead to increased intracellular iron accumulation and promote ferroptosis.

3.7. TFRC Knockdown Reduces the Therapeutic Effect of TMTP1‐sgBRAF‐EVs on ATC Cells by Inhibiting Ferroptosis

To verify the role of TFRC in TMTP1‐sgBRAF‐EVs‐induced ferroptosis, we knocked down TFRC (shTFRC) in 8505c cells and assessed cell death, ROS levels, lipid peroxidation, GSH levels, MDA levels, and intracellular Fe²⁺ concentration. We used shRNA to knock down TFRC in 8505c cells (shTFRC‐1 and shTFRC‐2) and analysed the effects on BRAF and TFRC protein expression by Western blot. The results showed that TFRC knockdown significantly reduced TFRC expression, and we selected the shRNA with the highest knockdown efficiency for subsequent experiments (Figure S10A). Next, we investigated the combined effect of TFRC and BRAF knockdown on BRAF and TFRC protein expression in 8505c cells. The results indicated that TFRC knockdown (shTFRC) significantly reduced TFRC expression, while BRAF knockdown (sgBRAF) significantly reduced BRAF expression. However, when both TFRC and BRAF were knocked down (shTFRC + sgBRAF), TFRC protein expression significantly increased compared to the shTFRC group and decreased compared to the sgBRAF group (Figure S10B). Calcein‐AM/PI staining showed that treatment with TMTP1‐sgBRAF‐EVs significantly increased cell death to around 60%, while TFRC knockdown notably attenuated this effect (Figure 7A). Using the ROS‐Violet610 probe, we detected that TMTP1‐sgBRAF‐EVs treatment significantly elevated ROS levels, which were substantially reduced by TFRC knockdown (Figure 7B). Lipid peroxidation was assessed using the oxidised BODIPY‐FITC probe, revealing that TMTP1‐sgBRAF‐EVs treatment significantly increased lipid peroxidation levels and this increase was significantly mitigated by TFRC knockdown (Figure 7C). Measurement of GSH levels showed that TMTP1‐sgBRAF‐EVs treatment significantly reduced the GSH/GSSG ratio, while TFRC knockdown significantly alleviated this reduction (Figure 7D). MDA levels, indicative of lipid peroxidation, were significantly increased by TMTP1‐sgBRAF‐EVs treatment, and this increase was notably reduced by TFRC knockdown (Figure 7E). Using the FerroOrange probe to detect intracellular Fe²⁺ concentration, we found that TMTP1‐sgBRAF‐EVs treatment significantly increased Fe²⁺ concentration, which was significantly diminished by TFRC knockdown (Figure 7F). These findings demonstrate that TFRC plays a crucial role in TMTP1‐sgBRAF‐EVs‐mediated ferroptosis in ATC cells, affecting cell death, ROS levels, lipid peroxidation, GSH levels, MDA levels, and intracellular Fe²⁺ concentration.

FIGURE 7.

Role of TFRC in TMTP1‐sgBRAF‐EVs induced Ferroptosis. (A) Live/dead staining with Calcein‐AM and Propidium Iodide (PI) was used to assess cell viability in 8505c cells treated with TFRC knockdown, TMTP1‐sgBRAF‐EVs, or the combination of TMTP1‐sgBRAF‐EVs + shTFRC. Scale bar = 25 µm. (B) ROS levels were measured using the ROS‐Violet610 probe in 8505c cells under the same treatments. (C) Lipid peroxidation levels were evaluated using oxidised BODIPY‐FITC probes. (D) Glutathione levels were analysed by measuring the GSH/GSSG ratio. (E) Malondialdehyde (MDA) levels were assessed to detect lipid peroxidation products. (F) FerroOrange probe was used to measure intracellular ferrous iron (Fe²⁺) concentrations in the treated 8505c cells. Scale bar = 25 µm. (G) Colony formation assay evaluated the clonogenic ability of 8505c cells. (H) Migration ability was assessed via transwell migration assay. Scale bar = 50 µm. (I) Invasion ability was determined via transwell invasion assay. Scale bar = 50 µm. All cell experiments were repeated at least three times. Two‐group comparisons used unpaired t‐tests. One‐way ANOVA was applied for comparisons among three or more groups. *p < 0.05, ***p < 0.001.

Further clonogenic assays revealed that TMTP1‐sgBRAF‐EVs treatment significantly inhibited the clonogenic potential of 8505c cells, while TFRC knockdown markedly attenuated this inhibitory effect (Figure 7G). Migration and invasion assays demonstrated that TMTP1‐sgBRAF‐EVs treatment significantly suppressed the migration and invasion capabilities of 8505c cells, and TFRC knockdown significantly reduced these suppressive effects (Figure 7H,I). These results collectively indicate that TFRC plays a critical role in TMTP1‐sgBRAF‐EVs‐induced ferroptosis and the enhancement of Vemurafenib's therapeutic efficacy.

4. Discussion

ATC is a highly aggressive and lethal subtype of thyroid cancer, characterised by rapid progression and high resistance to current therapeutic strategies (Greenberg et al. 2022; Tang et al. 2022). Although the B‐Raf kinase inhibitor Vemurafenib has shown remarkable efficacy in treating BRAF V600E‐mutant melanoma, its therapeutic effect in ATC is limited due to strong intrinsic resistance in BRAF‐mutant ATC cells (Lu et al. 2023; Lee et al. 2022). This study aims to investigate the mechanisms by which CRISPR/Cas9 gene editing can reverse Vemurafenib resistance in BRAF‐mutant ATC and to develop a novel extracellular vesicle (EV)‐based delivery system modified with the tumour‐targeting peptide TMTP1.

In recent years, CRISPR/Cas9 gene editing technology has been widely applied in cancer research due to its high efficiency, strong specificity, and ease of use, earning it the moniker ‘gene scissors’ (Wang et al. 2022c; Ding et al. 2023; Liang et al. 2023). By utilising CRISPR/Cas9 for gene knockout or modification, researchers can precisely investigate the roles of specific genes in tumour initiation and progression (Xie et al. 2024; Kim et al. 2022b; Issagholian et al. 2023). Additionally, we developed a TMTP1‐modified EV delivery system to improve the efficiency and specificity of CRISPR/Cas9 in vivo. Ferroptosis, a regulated cell death process dependent on iron and ROS, has gained increasing attention in cancer studies (Chen et al. 2022; Wang et al. 2023; Bi et al. 2024). It is characterised by lipid peroxidation and elevated ROS, tightly associated with intracellular iron accumulation (Chen et al. 2022; Akyuz et al. 2021).

In this study, we utilised CRISPR/Cas9 technology to knock out the BRAF gene in BRAF‐mutant ATC cell lines and observed a significant reduction in resistance to Vemurafenib. Further mechanistic investigations revealed that BRAF knockout markedly enhanced Vemurafenib‐induced ferroptosis, primarily through increased lipid peroxidation and ROS accumulation. Notably, we demonstrated for the first time that BRAF depletion promotes ferroptosis by inhibiting autophagy‐mediated degradation of transferrin receptor (TFRC), thereby increasing intracellular iron levels. These findings offer new insights into the mechanisms underlying Vemurafenib resistance in BRAF‐mutant ATC and propose ferroptosis modulation as a promising therapeutic strategy.

To achieve efficient and tumour‐specific delivery of BRAF sgRNA, we developed a novel TMTP1‐modified extracellular vesicle system (TMTP1‐sgBRAF‐EVs). The incorporation of the TMTP1 tumour‐homing peptide enabled targeted delivery and significantly improved intracellular transfection efficiency. Cas9 RNPs were loaded into EVs via electroporation, where transient membrane pores allowed efficient encapsulation, assisted by electrostatic interactions and osmotic gradients. Compared to traditional delivery systems, TMTP1‐sgBRAF‐EVs demonstrated enhanced in vivo stability, targeting specificity, and reduced off‐target effects and immunogenicity. This innovative platform not only improved Vemurafenib efficacy but also minimised its side effects, offering a new therapeutic avenue for BRAF‐mutant ATC (Montgomery and Worswick 2022; Khaddour et al. 2025, Szklener et al. 2022).

Interestingly, although T7E1 and sequencing analyses revealed indel frequencies of approximately 20% in the sgBRAF group and 50% in the TMTP1‐sgBRAF‐EVs group, Western blot results showed an almost complete depletion of BRAF protein. Such a discrepancy between moderate editing efficiency and near‐complete protein loss has also been observed in other CRISPR studies and can be attributed to multiple mechanisms: (i) nonsense‐mediated mRNA decay (NMD) triggered by frameshift‐inducing indels, resulting in rapid degradation of mutant transcripts (Nakagawa and Rathinam 2019); (ii) production of truncated proteins that are unstable or non‐functional (Batista et al. 2018); (iii) enhanced protein instability and proteasomal degradation following CRISPR‐induced mutations (Kowenz‐Leutz et al. 2010); and (iv) the nonlinear sensitivity of Western blot detection, which tends to underestimate residual low‐level protein. Consistently, a recent Nature Communications report showed that even with moderate editing efficiencies (∼14%–25%), target gene mRNA and protein expression were almost completely lost due to NMD or protein instability (Wan et al. 2024). These findings suggest that TMTP1‐sgBRAF‐EVs can achieve highly efficient functional silencing of BRAF expression, even at moderate indel levels, consistent with our qRT‐PCR results showing marked BRAF mRNA reduction (Figure 5E).

By integrating CRISPR/Cas9 gene editing with a tumour‐targeted EV delivery system, our study provides a comprehensive exploration of the mechanisms by which BRAF contributes to Vemurafenib resistance in ATC. We demonstrated that TMTP1‐sgBRAF‐EVs enhanced Vemurafenib efficacy by promoting ferroptosis (Figure 8) and established that disrupting BRAF not only sensitised ATC cells to treatment but also triggered ferroptotic cell death via intracellular iron accumulation. These findings not only offer mechanistic insights into drug resistance in ATC but may also apply to other BRAF‐mutant cancers. Furthermore, our engineered EV system highlights a new path for clinical application of gene editing technologies, improving therapeutic precision and efficacy.

FIGURE 8.

Schematic illustration of TMTP1‐sgBRAF‐EVs reversing resistance mechanisms in BRAF‐mutant ATC (created by BioRender).

Despite the significant progress made in this study, several limitations remain. First, the research was primarily conducted in vitro using cell lines and in vivo using mouse models; the safety and efficacy of these findings have yet to be validated in clinical trials. Second, while the TMTP1‐modified EV system demonstrated promising results in delivering CRISPR/Cas9 tools, its long‐term stability and potential immune responses in humans require further investigation. As safety assessment is critical for therapeutic translation, extended biodistribution studies are warranted to elucidate EV clearance rates and to confirm whether they preferentially accumulate in tumours rather than in off‐target organs such as the liver. Such data will be essential to ensure that EVs do not persist in non‐target tissues. Additionally, this study focused on BRAF‐mutant ATC, and further research is needed to explore the resistance mechanisms in other types of thyroid cancer. Therefore, future studies should expand the sample size and conduct larger‐scale clinical trials to validate the generalizability and application potential of these findings.

Future research should focus on optimising the TMTP1‐modified EV delivery system to enhance its stability and delivery efficiency in vivo while minimiColocalisation potential immune responses. Larger‐scale clinical trials are essential to validate the safety and efficacy of this system in clinical applications. Additionally, applying this delivery system to other types of cancer could help explore its broad applicability and effectiveness in different tumour treatments. By further research and optimisation, TMTP1‐sgBRAF‐EVs could become a versatile tool for tumour therapy, offering new treatment hopes for patients with BRAF‐mutant ATC and paving the way for a new era in cancer treatment.

Author Contributions

Shuo Zhang: data curation(equal), software(equal), visualization(equal), writing – original draft(equal), writing – review and editing(equal). Zhenrong Ji: data curation(equal), formal analysis(equal), validation(equal), writing – review and editing(equal), Xiaoyu Cheng: formal analysis(equal), software(equal), visualization(equal), writing – original draft(equal), writing – review and editing(equal). Yue Ma: investigation(equal), validation(equal), writing – original draft(equal), writing – review and editing(equal). Mingliang Feng: conceptualization(equal), data curation(equal), methodology(equal), software(equal), validation(equal), visualization(equal), writing – review and editing(equal). Dasheng Cai: conceptualization(equal), data curation(equal), investigation(equal), methodology(equal), resources(equal), software(equal), validation(equal), writing – review and editing(equal). Tao Bai: conceptualization (equal); data curation (equal); funding acquisition (equal); project administration (equal); writing ‐ original draft (equal); writing – review and editing (equal).

Ethics Statement

All animal experiments were approved by the Animal Ethics Committee of the First Hospital of China Medical University.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Note: (A) Mutation landscape of 496 thyroid cancer samples from the TCGA database. Each column represents a tumour sample, and each row represents a gene. Different colours indicate various types of mutations, as shown in the legend. The bar graph at the top displays the tumour mutation burden (TMB) for each sample, while the bar graph on the right shows the number of samples with mutations in each gene; (B) mRNA expression levels of BRAF in different thyroid cancer samples from the cBioPortal database; (C) Schematic representation of the BRAF protein, showing the somatic mutation rates and positions. The V600E mutation is a known driver mutation in thyroid cancer.

Supporting figure:2 Protein Expression in BRAF‐Mutant ATC Cells.

Note: (A) BRAF expression levels were analysed by Western blot in Nthy‐ori3‐1 and ATC cell lines. Quantification showed significantly higher BRAF expression in BRAFV600E‐mutant ATC cells, with the highest in 8505C; (B) Treatment with 5 µM Vemurafenib reduced BRAF protein levels in BHT‐101 and 8505C cells; (C) BRAF protein levels in BHT‐101 and 8505C were examined following knockout using different sgRNAs. All experiments were repeated at least three times. One‐way ANOVA was used for multiple‐group comparisons. ***p < 0.001.

Supporting figure:3 Transcriptome Sequencing of ATC Cells Treated with sgBRAF.

Note: (A) Volcano plot of DEGs in 8505c cells treated with sgNC and sgBRAF, with upregulated genes in yellow and downregulated genes in green (n = 3); (B) KEGG analysis of DEGs.

Supporting figure:4 Vemurafenib‐Induced Ferroptosis in BRAF‐Mutant ATC Cells.

Note: (A) Lipid peroxidation in BHT‐101 and 8505C cells after 5 µM Vemurafenib using BODIPY‐FITC and flow cytometry; (B) Intracellular ROS levels measured with ROS probes; (C) Ferroptotic morphology under TEM; (D) Cell viability under Vemurafenib and inhibitors: NAC, ZVAD‐fmk, necrostatin‐1, ferrostatin‐1; (E‐F) Detection of live and dead cells in RSL3‐resistant and erastin‐resistant BRAF‐mutant ATC cells, scale bar = 25 µm; (G‐H) Detection of live and dead cells in sgBRAF‐treated RSL3‐resistant and erastin‐resistant BRAF‐mutant ATC cells, scale bar = 25 µm. Cell experiments were repeated at least three times. All experiments were repeated at least three times. Two‐group comparisons used unpaired t‐tests; for ≥3 groups, one‐way ANOVA was applied. ***p < 0.001.

Supporting figure:5 Effect of BRAF Deletion on Vemurafenib‐Induced Ferroptosis.

Note: (A) The effect of 100 nM ferroptosis inhibitor ferrostatin‐1 on Vemurafenib‐induced cell death in BRAF‐deficient BHT‐101 and 8505C cells. Scale bar = 25 µm; (B) Measurement of GSH levels in BRAF‐deficient BHT‐101 and 8505C cells treated with 100 nM ferrostatin‐1; (C) Detection of malondialdehyde (MDA) levels in BRAF‐deficient BHT‐101 and 8505C cells following ferrostatin‐1 treatment; (D) Colony formation assay showing the effect of 100 nM ferrostatin‐1 on the clonogenic ability of BRAF‐deficient BHT‐101 and 8505C cells; (E‐F) Migration and invasion assays, respectively, in BRAF‐deficient cells after treatment with ferrostatin‐1. Scale bar = 50 µm. All cell experiments were repeated at least three times. One‐way ANOVA was used for comparisons among three or more groups. *p < 0.05, **p < 0.01, ***p < 0.001.

Supporting figure:6 Cancer Cell‐Derived EVs Preferentially Fuse with Their Parental Cells.

Note: (A) Western blot analysis of Alix, TSG101, and Calnexin isolated from cells and EVs. The supernatant obtained from ultracentrifugation during EV isolation serves as a negative control; (B) Schematic design of the in vitro study to evaluate the ability of cancer‐derived EVs to regulate their parental cells; (C‐D) Representative microscopy images showing the uptake of DiR‐labelled BHT‐101‐EVs (red) and DiO‐labelled 8505c‐EVs (green) by BHT‐101 and 8505c cells, with quantitative analysis of EV uptake in BHT‐101 and 8505c cells, scale bar = 25 µm; (E) Representative microscopy images showing the uptake of DiO‐labelled BHT‐101‐EVs (green) and DiR‐labelled 8505c‐EVs (red) by BHT‐101 and 8505c cells, with quantitative analysis of EV uptake in BHT‐101 and 8505c cells, scale bar = 25 µm; (F) Schematic design of the in vivo study to evaluate the homing ability of 8505c tumour EVs in a subcutaneous tumour model in nude mice. Cell experiments were repeated at least three times. Experimental animals n = 3. Unpaired t‐tests were used for two‐group comparisons. For comparisons among three or more groups, one‐way ANOVA was performed. ***p < 0.001.

Supporting figure:7 Preparation and characterisation of Tumour‐Derived EVs Encapsulating Cas9 Protein and sgRNA.

Note: (A) Loading capacity of purified EVs with Cas9 protein/sgRNA complexes using electroporation, freeze‐thaw, and sonication methods; (B) TEM images of unmodified EVs and TMTP1‐sgBRAF‐EVs, scale bar = 100 nm; (C) AFM images of unmodified EVs and TMTP1‐sgBRAF‐EVs. Unpaired t‐tests were used for two‐group comparisons, *p < 0.05; (D) Size distribution of unmodified EVs and TMTP1‐sgBRAF‐EVs as detected by NTA; (E‐F) Size stability of TMTP1‐sgBRAF‐EVs stored in PBS and serum at 37°C for 24 h and at 4°C for 7 days, respectively. Repeated measures ANOVA was used to analyse differences across time points. All cell experiments were repeated three times.

Supporting figure:8 Study on the Sensitivity of BRAF^mut ATC Cells to Vemurafenib with TMTP1‐sgBRAF‐EVs.

Note: (A) Colony formation assay showing the effect of BRAF deletion and treatment with TMTP1‐modified extracellular vesicles (TMTP1‐sgBRAF‐EVs) on the proliferation of 8505c cells; (B) Migration assay showing the effect of BRAF deletion and TMTP1‐sgBRAF‐EVs treatment on the migration ability of 8505c cells, scale bar = 50 µm; (C) Invasion assay showing the effect of BRAF deletion and TMTP1‐sgBRAF‐EVs treatment on the invasion ability of 8505c cells, scale bar = 50 µm; (D) Western blot analysis showing the effect of BRAF deletion and TMTP1‐sgBRAF‐EVs treatment on the expression of proliferation marker PCNA, Vim, and N‐cadherin (N‐cad) in 8505c cells. Cell experiments were repeated at least three times. One‐way ANOVA was used for multi‐group comparisons. **p < 0.01, ***p < 0.001.

Supporting figure:9 Toxicity and Immune Response Evaluation of TMTP1‐sgBRAF‐EVs.

Note: (A) Blood parameters and biochemical characteristics of healthy BALB/c nude mice; (B) H&E staining of tissue sections from different organs, scale bar = 50 µm; (C) Monitoring of body weight changes in nude mice; (D) Measurement of cytokine levels (IL‐10, IL‐1β, TNF‐α, IFN‐γ) in plasma. Experimental animals n = 3.

Supporting figure:10 Effect of TFRC Knockdown on BRAF and TFRC Protein Expression.

Note: (A) Western blot analysis showing the effect of TFRC knockdown (shTFRC‐1 and shTFRC‐2) on BRAF protein expression in 8505c cells. The bar graph on the right quantifies the relative expression levels of BRAF and TFRC proteins; (B) Western blot analysis showing the effect of TFRC knockdown and BRAF deletion (sgBRAF) on BRAF and TFRC protein expression in 8505c cells. The bar graph on the right quantifies the relative expression levels of BRAF and TFRC proteins. Cell experiments were repeated at least three times. One‐way ANOVA was applied for comparisons among three or more groups. ***p < 0.001.

Supporting Table: jev270170‐sup‐0002‐TableS1.docx

Zhang, S. , Ji Z., Cheng X., et al. 2025. “TMTP1‐Modified Small Extracellular Vesicles Target BRAF Mutation in Anaplastic Thyroid Cancer Reversing Vemurafenib Resistance With CRISPR/Cas9 Delivery.” Journal of Extracellular Vesicles 14, no. 9: e70170. 10.1002/jev2.70170

Data Availability Statement

The datasets used or analysed during the current study are available from the corresponding author on reasonable request.