In total, 86 PTC tissues and adjacent non-cancerous tissues were obtained from PTC patients receiving standard resection without radiation therapy at the First Affiliated Hospital of Zhengzhou University from 2020 to 2021. RNA (86 pairs) was separated from frozen tissues for quantitative real-time PCR (qRT-PCR) tests to assess the expression of FTO in PTC, followed by evaluation of clinicopathological parameters through univariate and multivariate Cox regression analyses. The study, carried out in accordance with the Declaration of Principles of Helsinki, won the approval of the Ethics Committee of the First Affiliated Hospital of Zhengzhou University. The license number of the ethical review for the study is 2021-KY-0202-002.

Human papillary thyroid carcinoma cell lines, like TPC-1 and KTC-1 with STR identification, were cultured in 1640 (Gibco, Detroit, MI, USA) in a humid atmosphere of 95% air and 5% CO₂ at 37°C, which was routinely supplemented with 10% fetal bovine serum (Gibco, Detroit, MI, USA), penicillin (100 units/ml, Grand Island, NY, USA), and streptomycin (100 μg/ml, Grand Island, NY, USA). The STR fingerprint authentications of TPC-1 and KTC-1 cells are available upon request.

A total RNA extraction kit (Vazyme, Nanjing, China) was used to collect total RNA from TPC-1 and KTC-1 cells. The PrimeScript RT reagent kit (Vazyme, Nanjing, China) was employed for the reverse transcription of RNA into cDNA. Then, quantitative real-time PCR was performed through the iTaq Universal SYBR Green Supermix kit (Bio-Rad) and ABI 7500 PCR system (ABI, USA), and GAPDH served as an internal control. Furthermore, all reactions were carried out based on the following cycle parameters: kept at 95°C for 10 min, then followed by 40 cycles of 95°C for 15 s and 60°C for 45 s. Specific amplification through a dissociation curve analysis was conducted. All data represent the average of triplicates under the calculation of 2^−ΔΔCt method.

Cells were harvested and dissolved in radio immunoprecipitation assay lysis buffer (Beyotime Biotechnology, Jiangsu Province, China), and the protein concentration was assessed by the bicinchoninic acid protein assay (Beyotime Biotechnology, Jiangsu, China). Then, sodium dodecyl sulfate–polyacrylamide gel electrophoresis (NCM Biotech, Jiangsu, China) was put into use for separating proteins, and these were transferred into polyvinylidene difluoride membranes (Millipore, USA) by an electro-blotting apparatus. After blocking with 5% skimmed milk, the membranes were incubated overnight in primary antibodies at 4°C based on the manufacturer’s instructions. The unbound antibody was washed away, followed by further incubation of the membranes with horseradish peroxidase (HRP)-conjugated rabbit secondary antibodies (Beyotime Biotechnology, Jiangsu, China) at room temperature for 1 h. The visualization of bands was performed with an enhanced chemiluminescence detection reagent (Beyotime Biotechnology, Jiangsu, China). All antibodies used in this study were purchased from Abcam (Cambridge, UK). These primary antibodies are known as rabbit monoclonal anti-FTO (ab126605, Abcam), rabbit monoclonal anti-SLC7A11 (ab37185, Abcam), rabbit monoclonal anti-β-actin (ab8226, Abcam), rabbit monoclonal anti-Glutathione Peroxidase 4 (ab125066, Abcam), rabbit monoclonal anti-COX2/Cyclooxygenase 2 (ab179800, Abcam), rabbit monoclonal anti-FACL4 (ab155282, Abcam), rabbit monoclonal anti-Ferritin Heavy Chain (ab183781, Abcam), and rabbit monoclonal anti-NOX1 (ab131088, Abcam).

Cell proliferation was valued by the Cell Counting Kit-8 (CCK-8) Assay Kit (Vazyme, Nanjing, China). The KTC-1 and TPC-1 cells were cultured in 96-well plates with 1.5 × 10³ cells/well; 10 μl of CCK-8 reagent was added to each well and then incubated for 3 h in a dark environment. A microplate reader was used to measure the absorbance at 450 nm. The proportion of relative cell proliferation through the average optical density (OD) of each group was calculated. The formula is as follows: (OD treatment/OD control) × 100%.

TPC-1 and KTC-1 cells were inoculated into 24-well transwell chambers with polycarbonate membrane (353097) (Corning Life Science, Acton, MA, USA) to evaluate invasion capacities and cell migration under the function of Matrigel (Corning Life Science, 354234) or without it. The upper chamber was injected with cells in a serum-free medium at a density of 4 × 10⁵ cells/well, and the outer chamber got filled with the complete-culture medium. After a 12-h culture process, the cells that adhered to the underside of the chamber for 30 min were fixed with 4% paraformaldehyde, stained with crystal violet for 15 min, and later got calculated under an optical microscope (Zeiss, Germany).

Lentiviruses expressing FTO, SLC7A11, FTO-shRNA (target sequences: gcAGCATACAACGTAACTTTG), SLC7A11-shRNA (target sequence: ccCTGGAGTTATGCAGCTAAT), scrambled shRNA, and empty vector were purchased from Genechem Company (Shanghai, China). TPC-1 and KTC-1 cells were selected to establish a stable knockdown and over-expressed cell lines. In brief, 3 × 10⁴ cells were seeded into a 6-well plate and then transfected with the specified lentivirus using HitransG P/A (Genechem, China), with reference to the manufacturer’s instructions. In addition, infected cells were selected by using 2.5 μg/ml puromycin (MCE, USA) for ≥14 days, and the transfection efficiency was verified by RT-qPCR and Western blot analysis.

Seqhealth Technology Co., Ltd. (Wuhan, China) was the main supplier of MeRIP sequencing and following data analysis. The stably transfected cell, TPC-1 cells with FTO over-expression, and controlled cells with empty vector were collected. Total RNAs were extracted by TRIzol Reagent (Invitrogen, catalog number 15596026). After RNA extraction with the help of DNaseI, DNA digestion was performed. Then, the RNA quality was evaluated by checking A260/A280 with Nanodrop TM OneC spectrophotometer (Thermo Fisher Scientific Inc.). RNA integrity was verified by 1.5% agarose gel electrophoresis. In the end, qualified RNAs were weighed by Qubit3.0 with QubitTM RNA Broad Range Assay kit (Life Technologies, Q10210). Furthermore, 50 μg total RNAs was used for polyadenylated RNA enrichment by VAHTS mRNA Capture Beads (VAHTS, catalog number N401-01/02). Moreover, 20 mM ZnCl2 was added to mRNA and incubated at 95°C for 5–10 min until the RNA fragments were evenly distributed in 100–200 nt. Then, 10% RNA fragments was saved as “input”, while the remaining parts were utilized for m6A immunoprecipitation (IP). The specific anti-m6A antibody (Synaptic Systems, 202203) was applied into m6A immunoprecipitation. TRIzol reagent (Invitrogen, catalog number 15596026) was utilized to prepare both input and IP RNA samples. In addition, referring to the manufacturer’s instructions, the stranded RNA sequencing library was established by KC-Digital TM Stranded mRNA Library Prep Kit for Illumina^® (catalog number DR08502, Wuhan Seqhealth Co., Ltd., China). The kit used a unique molecular identifier (UMI) of 8 random bases to label pre-amplified cDNA molecules, thereby removing repetitive bias in PCR and sequencing steps. The library products with 200–500 bp were enriched, quantified, and finally sequenced on Novaseq 6000 sequencer (Illumina) with the PE150 model. In a word, the raw data was aligned with human genome GRCh37/hg19 via the STAR software (version 2.5.3a). Then, m6A peaks were acquired by the ExomePeak software (version 3.8) based on the Ensembl database. Sequence motifs abundant in m6A peak regions were identified using Homer (version 4.10). Gene Ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis were both implemented by KOBAS software (version: 2.1.1) for annotated genes, and the cutoff value of the corrected p-value at 0.05 aimed at assessing the statistical significance of enrichment.

UID RNA-seq experiment, high throughput sequencing, and data analysis were performed by Seqhealth Technology Co., Ltd. (Wuhan, China). TRIzol Reagent (Invitrogen, catalog number 15596026) was used for the extraction of total RNA from TPC-1-Lv-FTO/TPC-1-vector. DNA digestion was carried out after RNA was extracted with the help of DNaseI. RNA quality was valued by studying A260/A280 with Nanodrop TM OneCspectrophotometer (Thermo Fisher Scientific, Inc.). Later, RNA integrity was tested by 1.5% agarose gel electrophoresis. The qualified RNAs were quantified by Qubit3.0 equipped with QubitTM RNA Broad Range Assay kit (Life Technologies, Q10210). Then, total RNAs weighing 2 μg were used for the library preparation of stranded RNA sequencing using the KC-DigitalTM Stranded mRNA Library Prep Kit for Illumina^® (catalog number DR08502, Wuhan Seqhealth Co., Ltd., China) based on the manufacturer’s instruction. The kit used a UMI of 8 random bases to label pre-amplified cDNA molecules, hereby eliminating repetitive bias in PCR and consequent steps. The library products with 200–500 bp were enriched, quantified, and finally sequenced on the Novaseq 6000 sequencer (Illumina) under the PE150 model. Distinctively expressed genes were used to draw heat maps and conduct the KEGG ontology enrichment analysis. As for the KEGG enrichment analysis, P-value <0.05 was used as the criteria of a significantly enriched gene set. NCBI (http://www.ncbi.nlm.nih.gov/Genbank/GenbankSearch.html) and KEGG databases (http://www.genome.jp/kegg/) got respectively used for analyzing gene ontology (GO) and signal pathways.

Overall, 2 μg of total RNA and m6A spike-in control mixture was added to 300 μl of 1IP buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% NP40, and 40 U/μl RNase inhibitor) containing 2 μg of anti-m6A rabbit polyclonal antibody (Synaptic Systems). Twenty microliters of Dynabeads™ M-280 sheep anti-rabbit IgG suspension of each sample was blocked with freshly prepared 0.5% bovine serum albumin for 2 h at 4°C, washed three times with 300 μl of 1IP buffer, and resuspended in the total RNA–antibody mixture prepared as detailed above. Binding of RNA to m6A-antibody beads was head-to-tail rotated for 2 h at 4°C, and the enriched RNA was washed with 200 μl elution buffer for 1 h at 50°C, after which the co-precipitated RNA was isolated, and the RNA was subjected to qRT-PCR.

The m6A quantification was carried out utilizing the EpiQuik™ m6A RNA Methylation Quantification Kit (Colorimetric) (P-9005, Epigentek, China). The relevant solutions were added to each well according to the manufacturer’s instructions. The solution was mixed by gentle tilting from side to side, after which m6A RNA capture was performed. It was ensured that any residual wash buffer in the wells was thoroughly removed at each wash step. Signals were detected at the end when the color of the positive control wells changed to medium blue. A 100-μl Stop Solution was then added to each well to stop the enzyme reaction. After adding the Stop Solution, the color of the compound solution changed to yellow. The absorbance was read at 450 nm on a microplate reader.

The human tissue sections need to be deparaffinized and rehydrated, and 10 mM citrate buffer should be used for antigen extraction. After treatment with 3% H₂O₂, the sections were soaked with the primary antibodies overnight at 4°C and then incubated with the HRP-conjugated secondary antibody for 1 h at room temperature. Referring to the manufacturer’s instructions, diaminobenzidine (G1212-200T) (Wuhan Servicebio, China) was used to check the immune complexes. The Image-Pro Plus software (Media Cybernetics, MD, USA) was chosen to quantify protein levels by calculating the integrated optical density of each stained area. In short, the sections were deparaffinized and boiled in Tris antigen retrieval buffer. Afterwards, the sections were incubated overnight with primary antibodies against FTO (ab126605, Abcam, 1:500 dilution), and SLC7A11 (ab37185, Abcam, 1:400 dilution) at 4°C. Immunohistochemistry images were captured by using an Axio Imager 2 (Zeiss). FTO and SLC7A11 signals in immunohistochemical staining were assigned with values by a semi-quantitative scoring system based on the product of positive percentage and signal intensity. The positive percentage formula is as follows: 0 = 0%, 1 = 1–25%, 2 = 26–50%, 3 = 51–75%, 4 = above 75%. On the other hand, the signal intensity is as follows: 0 = no staining, 1 = weak staining, 2 = medium staining, 3 = strong staining. Then, the median value of FTO and SLC7A11 scoring in all specimens got selected as the cutoff value. Those with FTO or SLC7A11 scoring below the cutoff value were labeled as the low-expression specimen, while for those scoring higher than the value, they were labeled as the high-expression specimen. Two trained pathologists performed the quantification of IHC staining.

Cells were stained by using Annexin V Alexa Fluor 647/PI Apoptosis detection Kit (4A Biotech Co., Beijing, China). Referring to the manufacturer’s recommendations, apoptosis was analyzed by the BD Accuri C6 Flow cytometer (BD Biosciences, USA). The cells, first digested with 0.25% trypsin (without EDTA), were then washed twice with phosphate-buffered saline (PBS; centrifugation at 1 × 10³ rpm for 5 min), and 1 × 10⁶ cells were collected. The re-suspension of cells was performed in 1 × 10³ µl binding buffer. As 100 μl of cell suspension in a 5-ml flow tube was taken out, 5 μl Annexin V/Alexa Fluor 647 was added then. The suspension was incubated at room temperature for 5 min in dark experimental conditions, added with 10 μl of 20 ug/ml propidium iodide solution and 400 μl PBS through the use of a flow cytometer to collect 2 ×10⁴ events. Apoptotic cells in their early and late stages were separated and then presented in a quadratic graph.

Male BALB/c nude mice aged between 5 and 6 weeks were purchased from Beijing Vital River Laboratory Animal Technology (Vital River, Beijing, China) and kept in an animal facility free from pathogen for at least 7 days before use. Then, a cohort of 15 male nude mice was assigned to the groups of TPC-1-Lv-FTO, TPC-1-vector, and TPC-1 (n = 5 per group) under random selection. A total of 1 × 107 cells kept in a mixture of 100 μl PBS and Matrigel were injected into the BALB/c nude mice subcutaneously. Tumor volumes were then monitored twice per week after the end of the first week. After 4 weeks, the tumor tissues of sacrificed mice were collected for subsequent analysis. The entire experiment was approved by the Animal Care and Use Committee of the First Affiliated Hospital of Zhengzhou University.

The dual-luciferase reporter gene vector of target gene SLC7A11 and mutants with mutations in the FTO-binding site were formulated respectively as Luc-SLC7A11-3’UTR and Luc-SLC7A11-3’UTR-MUT. Two reporter plasmids, FTO and FTO-NC plasmids, were co-transfected into 293T cells, and the cells were lysed for 36 h after transfection and centrifuged at 12,000 rpm for 1 min. Next, the supernatant was collected, and Dual-Luciferase^® Reporter Assay System (Promega) was used to detect luciferase activity. Furthermore, 100 μl of firefly luciferase working solution was added into each cell sample to detect firefly luminescence (firefly luminescence), and Renilla luciferase working solution was added into the sample to identify Renilla luciferase (renilla luciferase). It is worth noting that luciferase was employed as well. The experiment was repeated three times.

Firstly, TPC-1-Lv-FTO cells were collected and placed in 2.5% glutaraldehyde phosphate (0.1 M, pH 7.4; Servicebio, Wuhan, China) overnight at 4°C, then fixed in 2% buffered osmium tetraoxide, and finally embedded in Epon812 (Merck), followed by dehydration. Then, ultrathin sections were cut out (60 nm in thickness), and uranyl acetate and lead citrate were used for staining. Eventually, a TEM microscope was utilized for picture checking (FEI, Hillsboro, OR, USA).

Cellular glutathione (GSH)/oxidized GSH (GSSG) ratio was analyzed by using the GSH and GSSG assay kit (Beyotime Biotechnology, Jiangsu, China, S0053) under the manufacturer’s instructions. Full sample preparation, standard solution preparation, and concentration detection were carried out by referring to the manufacturer’s protocols. Fresh cells were used for the assay, and PBS was employed to wash the cells once. The cells were collected by centrifugation, and the supernatant was aspirated, with a protein removal reagent added to the cell pellet. Liquid nitrogen was used under 37°C water bath to quickly freeze, and the sample was thawed twice. The cells were placed at 4°C or ice bath for 5 min and harvested by centrifugation at 1 × 10⁴ g for 5 min at 4°C. Then, the supernatant was taken out, and it was time to determine the amount of total glutathione (GSSG + GSH) by using a plate reader to measure the absorbance at 412 nm. The use of appropriate reagents to remove GSH in the sample was first done, and then the above-mentioned reaction principle was employed to determine the content of GSSG. The content of GSSG was subtracted from the amount of total glutathione (GSSG + GSH) to calculate the content of GSH.

Based on the manufacturer’s instructions, Lipid Peroxidation MDA Assay Kit (Beyotime Biotechnology, Jiangsu, China, S0131S) was used to make a quantitative detection of malondialdehyde (MDA) in the cells by the colorimetric method to analyze the level of lipid oxidation in the sample. The cell samples were lysed using cell lysis buffer. After lysis, the samples were centrifuged at 12,000g for 10 min at 4°C for the subsequent determination of supernatant. The MDA detection working solution was added to the sample supernatant, mixed well, and heated at 100°C or under the condition of boiling water bath for 15 min. The mixtures were then cooled down in a water bath till room temperature and centrifuged at 1,000 g for 10 min at room temperature. Two hundred microliters of supernatant was later taken out and added to a 96-well plate, and then its absorbance at 532 nm was measured by a microplate reader.

Referring to the manufacturer’s instructions, the iron colorimetric assay kit (Applygen, Beijing, China, E1042) was used to quantitatively detect and analyze Fe²⁺ in the cells under the ferrozine colorimetric method. The cells cultured in the iron-containing medium were washed twice with cold PBS, then the PBS got aspirated, and the cell sample was lysed with cell lysate and placed on a shaker for 2 h. in total, 4.5% potassium permanganate was added, mixed uniformly, and incubated at 60°C for 1 h. After the sample was cooled down to room temperature, iron ion detection reagent was well added and mixed, followed by incubation at room temperature for 30 min. Next, 200 μl of supernatant was taken and added to a 96-well plate; then, its absorbance at 550 nm was measured with a microplate reader.

With reference to the manufacturer’s instructions, Cellular Glutathione Peroxidase Assay Kit with a kit for glutathione peroxidase (GPx) activity in tissues or other samples, NADPH (Beyotime Biotechnology, Jiangsu, China, S0056) got employed for cell detection. The cells were collected through a cell scraper and washed once with PBS. The cell sample, lysed with cell lysate, was then centrifuged at 12,000 g for 10 min at 4°C. The supernatant for identifying the enzyme activity was taken out. In the next stage, on a 96-well plate, the detection buffer was added to test the sample, and the GPx detection working solution in sequence was evenly mixed and incubated at room temperature for 15 min. Next, after the adding and mixing of peroxide reagent solution, attached with the appropriate enzyme label, the solution was put immediately at 25°C and room temperature. The measurement of A340 with an instrument or a trace ultraviolet spectrophotometer was conducted for 5 consecutive minutes.

Statistical analysis was performed using GraphPad Prism 8.0 (GraphPad, Inc., USA) and SPSS 22.0 (SPSS, Inc., USA) software. All statistical tests followed the two-way path, and p <0.05 was treated with statistical significance. The difference between groups was tested by one-way analysis of variance. Representative data was demonstrated as mean ± SD. In addition, a correlation analysis of gene expression was carried out with linear regression. The P-values for every result were labeled on the figures, and P <0.05 was reckoned as statistically significant (*P <0.05, **P <0.01, ***P <0.001, ****P <0.0001).

After The Cancer Genome Atlas (TCGA) data set (http://ualcan.path.uab.edu/index.html) was studied, the analysis of PTC cohort from TCGA has shown that FTO was downregulated in tumor tissues compared with the corresponding noncancerous tissues ( Figures 1A, B).

To estimate FTO expression in PTC and the probable clinical significance, 86 cases of tumor and adjacent tissue from PTC patients were analyzed ( Table 1 ). As a result, real-time qRT-PCR showed a remarkable reduction in FTO mRNA levels in PTC tissues in comparison with non-cancerous controls. Immunohistochemistry staining displayed the existence of a number of positively FTO-stained cells in normal thyroid control tissues, while negative staining was detected in PTC tissues ( Figure 1C ). GAPDH got employed as internal control, individually, in the qRT-PCR assays (*P < 0.05, **P < 0.01, and ***P < 0.001).

To evaluate the functional roles of FTO in PTC, the expression of FTO was first identified in PTC cell lines—TPC-1 cells which were chosen to establish FTO-silencing and FTO-overexpression models. Consequently, the transfection efficiency was verified by western blot ( Figure 2A ).

Indications from the CCK-8 assays showed that the upregulation of FTO weakened the proliferation capability of TPC-1 cells, while FTO silencing proved the opposite effect in KTC-1 cell ( Figure 2B ). Besides this, the FTO knockdown effects on the proliferation and migration of PTC cells were verified using a scratch-wound model. Then, these results demonstrated that FTO knockdown significantly promoted the gap closure rate of PTC cells ( Figure 2C ). Transwell assays were conducted, and the inhibition of FTO was found to facilitate both the migration and invasion of PTC cells. In addition, the over-expression of FTO can impair these phenotypes ( Figure 2D ). On the whole, these findings indicate that increased FTO expression may curb the malignancy of PTC cells. The flow cytometry experiments suggest a stronger apoptotic ability in FTO knockout cell lines, while FTO overexpression cell lines show the opposite trend ( Figure 2E ).

In order to further verify the role of FTO in PTC, in vivo experiments with subcutaneous tumor models were implemented. On top of that, when FTO was over-expressed, the volumes and weights of the xenografted tumors decreased compared with those of the control group. These results indicate that FTO would exert an inhibitory effect on PTC tumor growth in in vivo and in vitro experiments. In addition, the xenografts derived from TPC-1-Lv-FTO cells (FTO overexpression) demonstrated tumor weight and volume which were much lower than those extracted from TPC-1 vector (empty) ( Figure 2F ). The monitoring of the size and the weight of the tumor formed in the subcutaneously implanted mice model was performed every 3 days (*P < 0.05, **P < 0.01).

To further investigate the potential pathways of the FTO tumor suppressor effect in papillary thyroid carcinoma, RNA-seq sequencing on TPC-1-Lv-FTO and vector cell lines was performed and repeated 3 times. Through RNA-seq sequencing, 17 genes were found to be upregulated, while 25 genes were downregulated in the presence of TPC-1-Lv-FTO and vector cell lines [logFC <-1 or logFC >1, p <0.05, false discovery rate (FDR) <0.05] ( Figures 3A, B , Tables 2 , 3 ). This result is shown through a volcano plot ( Figure 3B ). The gene enrichment pathway of RNA-seq correlated with the FTO expression ( Figure 3C ), and the GO ( Figure 3D ) and KEGG pathways ( Figure 3E ) of the upregulated genes and downregulated genes were analyzed respectively.

After the comparative RNA-Seq analysis of FTO over-expressed TPC-1 cells (TPC-1-Lv-FTO) and control group (TPC-1-vector), a decreased pattern was detected in the Lv-FTO cells of m6A levels. Next, to map the m6A modifications, m6A Seq was carried out. The motif analysis performed by using the HOMER program has shown that “GAACA” serves as the m6A consensus motif of TPC-1 cells ( Figure 4A ). Further investigations on the distribution of the peak m6A indicated that m6A in the vehicle group and the Lv-FTO group showed a similar total distribution pattern. In order to discover the precise mechanisms in support of the observed FTO-dependent phenotypes, an integral approach combining meRIP-Seq and RNA-Seq was employed through the utilization of vector cells and stable FTO’s over-expression. In general, m6A-seq identified 5,245 and 3,899 gene numbers from 9,150 and 6,006 m6A peaks in vector (normal control) and Lv-FTO cells ( Figures 4B, C ). Through meRIP-Seq sequencing, 6 genes were found to be upregulated, while 34 genes were downregulated in the presence of TPC-1-Lv-FTO and vector cell lines (logFC <-1 or logFC >1, p <0.05, FDR <0.05) ( Tables 4 , 5 , and Figures 4D, E ). meRIP-Seq has shown m6A peaks, and the read distribution demonstrates the proportion of common m6A peaks in the indicated areas in TPC-1 cells and the distribution of reads in the peak of the gene functional area ( Figure 4F ). The demethylated genes in TPC-1-Lv-FTO (logFC >1, p <0.05) were screened through meRIP-Seq and analyzed with GO ( Figure 4G ) and KEGG ( Figure 4H ).

Upon combination analysis of meRIP-Seq and RNA-Seq and selection of SLC7A11, it is speculated that SLC7A11 may be used as a downstream target gene of FTO to regulate papillary thyroid carcinoma ( Figure 5A ).

First, we first verified whether FTO altered the m6A levels in papillary thyroid cancer cell lines, and we found that TPC-1-Lv-FTO occurred with the downregulation of m6A levels, while KTC-1-shFTO occurred with the upregulation of m6A levels ( Figure 5B ). By checking the meRIP-seq data to determine whether the m6A modification was enriched in the SLC7A11 sequence, we found that SLC7A11 clearly showed m6A level downregulation by comparing the difference in m6A peak on the SLC7A11 INPUT and IP samples ( Figure 5C ).

Next, we performed anti-m6A immunoprecipitation and use qRT-PCR to analyze the m6A-immunoprecipitated RNA. The results showed that FTO overexpression resulted in the downregulation of m6A modification of SLC7A11 mRNA. In contrast, FTO knockdown increased the m6A modification of SLC7A11 mRNA ( Figure 5D ). From the perspective of SLC7A11 mRNA, the mRNA level of SLC7A11 was down- or upregulated when FTO was over-expressed or knocked down ( Figure 5E ). These results fully illustrate the regulation of SLC7A11 expression by FTO.

In order to prove the essential role of m6A in regulating SLC7A11, a luciferase reporter, an inserted mutant (Mut), and a wild-type (WT) SLC7A11–3′UTR sequence equivalent to its mutating putative m6A sites were designed ( Figure 5F ). The mutation of m6A consensus sequences were generated by substituting adenine for thymine. Then, the relative luciferase activities of the wild type and 5 mutant SLC7A11 3′UTR reporter vectors in TPC-1-Lv-FTO (overexpression) cells were implemented and evaluated. The dual-luciferase reporter experiment of the over-expression of FTO increased the luciferase activity of the reporter carrying the wild-type SLC7A11 3′UTR rather than its 3′UTR mutant in the FTO recognition site, demonstrating that FTO could be bound to the putative m6A sites of SLC7A11 3′UTR ( Figure 5G ).

In order to examine the expression of SLC7A11 in PTC and its potential clinical significance, 86 cases of tumor and corresponding noncancerous tissue from PTC patients were evaluated. As a result, real-time qRT-PCR showed a crucial elevation in SLC7A11 mRNA levels in PTC tissues compared with those under noncancerous controls ( Figure 6A ). Immunohistochemistry staining has demonstrated masses of positively SLC7A11-stained cells in PTC tissues, while negative staining could be identified in normal thyroid control tissues ( Figure 6B ).

To test the role of SLC7A11 in PTC, SLC7A11-knockdown TPC-1 and SLC7A11’s over-expressed KTC-1 cell lines were first formulated and tested by western blot ( Figure 6C ). The CCK-8 assay demonstrated that the silence of SLC7A11 would inhibit cell growth and viability ( Figure 6D ). In addition, the loss of SLC7A11 resulted in the suppression of the migration and invasion abilities of PTC cells. Through cell scratch experiments, it was found out that the proliferation rate of SLC7A11 over-expression cells increased ( Figure 6E ). Transwell migration and invasion experiments also proved the ability of SLC7A11 overexpression to promote the migration and invasion of papillary thyroid cancer cells, while SLC7A11 knockdown cells showed the adverse effect ( Figure 6F ).

The latest research demonstrates that SLC7A11 can enhance tumor development by inhibiting ferroptosis and other ferroptosis-independent functions [31]. Ferroptosis, as an iron-dependent form of regulated cell death, is the result of the lethal accumulation of lipid-based reactive oxygen species (ROS). Then, given the strict regulation of iron homeostasis for the benefit of health, it is unsurprising that disturbances in iron balance would lead to various diseases. It is under speculation that the regulation of SLC7A11 gene expression may be a crucial target of FTO in modulating PTC cell ferroptosis and tumor progression. In our next set of experiments, the role of ferroptosis in PTC is to be investigated.

Firstly, SLC7A11 expression in PTC cells was estimated. It is to be noted that SLC7A11 expression was noticeably downregulated in TPC-1-Lv-FTO cells. However, SLC7A11 expression was significantly elevated in KTC-1-shFTO cells ( Figure 7A ). These results demonstrated that FTO might be involved in the regulation of PTC cell ferroptosis. Furthermore, with regard to the mRNA level of FTO and SLC7A11 in 86 pairs of PTC tissues assessed by Spearman rank correlation coefficient analysis, a negative correlation between FTO and SLC7A11 was found (p < 0.001; Figure 7B ).

To clarify the mechanisms in which FTO regulates PTC cell death and tumor progression, the type of cell death upon FTO overexpression was first identified. Furthermore, after the preparation of TPC-1-Lv-FTO cells, TEM was completed at ×2,000 magnification. The cell morphology was found to change significantly as the cell mitochondria became smaller, the membrane density increased, and the cristae decreased sharply or even vanished ( Figure 7C ).

Then, the protein levels of ferroptosis biomarkers—NOX4, COX2, FACL4, and FTH—were set by western blot to detect whether ferroptosis was related to PTC. The results demonstrated that the protein expression of acyl-CoA synthetase long-chain family member 4 (ACSL4) NOX4 and COX2 got markedly elevated, while the GPX4 and FTH levels were overtly reduced in TPC-1-Lv-FTO compared with that of the vector group. The results have also illustrated that the protein expression of ACSL4 NOX4 and COX2 was markedly reduced, while the level of GPX4 and FTH was notably elevated in KTC-1-shFTO in comparison with that of the vector group ( Figure 7D ).

Through monitoring of the levels of MDA ( Figure 7F ), the GSH/GSSG ratio ( Figure 7E ), GPx ( Figure 7H ), and iron ions in various cell lines, as for Fe²⁺ level ( Figure 7G ), MDA in TPC-1-Lv-FTO was found to be significantly higher than TPC-1/vector, and MDA in KTC-1-shFTO proved to be significantly lower than in KTC-1/scramble. In terms of the GSH/GSSG ratio, the GPx level in TPC-1-Lv-FTO was considerably lower than that of the TPC-1/vector, while for GSH/GSSG ratio, the GPx level in KTC-1-shFTO was much higher than that of the KTC-1/scramble. However, the Fe²⁺ level held the opposite trend as MDA in TPC-1-shSLC7A11 was significantly higher than TPC-1/scramble, and MDA in KTC-1-Lv-SLC7A11 was substantially lower than in the KTC-1/vector. When it comes to the GSH/GSSG ratio, the GPx level in TPC-1-shSLC7A11 was remarkably lower than in the TPC-1/scramble, while at the GSH/GSSG ratio, the GPx level in KTC-1-Lv-SLC7A11 was dramatically higher than in the KTC-1/vector.

Finally, whether FTO could regulate PTC cell ferroptosis by mediating the m6A demethylation of SLC7A11 was reversely verified. To validate that the observed phenotype was mediated by the dysregulation of the FTO–SLC7A11 axis, a few functional rescue assays were performed. Then, as demonstrated by the CCK-8 assay, the knockdown effect of FTO resulted in the enhanced proliferation capacity in KTC-1 cells, which could be reverted by SLC7A11-knockdown. The over-expression of SLC7A11 also substantially increased the abolished mobility induced by the over-expression of FTO ( Figure 8A ). Besides this, wound healing assays ( Figure 8B ) and transwell migration and invasion assays ( Figure 8C ) also confirmed the initial speculation. To sum up, SLC7A11 dysfunction may lead to FTO-mediated PTC cell proliferation or mobility signatures.