All clinical tissue samples from patients with PTC and normal adjacent tissues were obtained from populations in China and were collected at the Department of Head and Neck Surgical Oncology, Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College between 2016 and 2019. The research was approved by the Ethics Committee of National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College and conducted under the guidance of the Declaration of Helsinki. Informed consent regarding the use of specimens was obtained from all patients.

TPC1 cells and BCPAP cells were cultured in DMEM/F-12 medium (Gibco, Thermo Fisher Scientific) supplemented with 2 mM glutamine, 10% fetal bovine serum (Gibco, Thermo Fisher Scientific), and 100 U/ml penicillin/streptomycin (Sigma, St. Louis, MO, United States). The cells were maintained at 37°C in a humidified atmosphere containing 5% CO₂. Rapamycin (Sirolimus) was purchased from Selleck (Shanghai, China).

The siRNAs of ENO1 and CST1 were purchased from GenePharma. siCtrl: 5′-UUCUCCGAACGUGUCACGU-3′; siENO1#1: 5′-CGUACCGCUUCCUUAGAACUU-3′; siENO1 #2:5′-GAAUGUCAUCAAGGAGAAAUA-3′; siCST1#1:5′-CC ACCAAAGAUGACUACUA-3′; siCST1#2:5′-GCUCUUUCGAG AUCUACGA-3′; siCST4#1:5′-CUUUCGAGAUCUACGAAGU UCTT-3′; siCST1#2:5′-CCAUCAGCGAGUACAACAATT-3′. According to the manufacturer’s protocol, Lipofectamine RNAiMAX transfection reagent (Invitrogen; Thermo Fisher Scientific) was used for transfection of siRNA. Lipofectamine 3000 (Thermo Fisher Scientific) was used for plasmid overexpression.

CCK-8 reagent was purchased from Sigma. Cells seeded into a 96-well plate with a density of 2000 cells per well were maintained for the indicated days, followed by 10 μl CCK-8 reagent added to each well and incubated at 37°C for 3 h. The absorbance at 450 nm was measured using a microplate reader.

TPC1 and BCPAP cells were seeded into 6-well plates at 1,000 cells/well, and fresh culture medium was changed every 3 days. Colonies formed after culturing the cells for 14 days. After washing in phosphate-buffered saline (PBS) three times, colonies were fixed in 4% paraformaldehyde for 20 min and stained with GIEMSA staining solution for 20 min. Groups of more than 50 cells were identified as colonies.

TRIzol (Thermo Fisher Scientific) was used for the isolation of total RNAs from cells. The RNAs were reverse-transcribed using M-MLV-RTase (Promega) following the manufacturer’s protocol. The qRT-PCR analysis was performed using SYBR Master Mixture (TAKARA) on the Agilent MX3000p real-time PCR system. qRT-PCR primers were as follows: ENO1 forward, 5′-AAAGCTGGTGCCGTTGAGAA-3′ and reverse, 5′-GGTTGTGGTAAACCTCTGCTC-3′; CST1 forward 5′-GCCTGCGCCAAGAGACA-3′ and reverse, 5′-CCCTGC TGAGCAACAAAGGA-3′; CST4 forward 5′-CCTCTGTGTAC CCTGCTACTC-3′ and reverse, 5′-CTTCGGTGGCCTTG TTGTACT-3′; β-actin forward, 5′-GAGCTGCGTGTGGC TCCC-3′ and reverse, 5′-CCAGAGGCGTACAGGGAT AGCA-3′.

The eBioscience^TM Annexin V-FITC apoptosis detection kit (Thermo Fisher Scientific) was used to measure apoptosis. Suspending cells were incubated with 5 μL annexin V-FITC for 10–15 min. After incubation, cells were washed in 1× binding buffer, then resuspended in 1× binding buffer. Resuspended cells were incubated with propidium iodide (20 μg/mL). Finally, samples were subjected to flow cytometry.

FxCycle PI/RNase Staining Solution (Thermo Fisher Scientific) was used to analyze the cell cycle according to the manufacturer’s protocol. Briefly, cells were trypsinized and centrifuged at 13,000 rpm for 5 min. After washing in iced D-Hanks (pH = 7.2–7.4) buffer, cells were fixed with iced 75% ethanol for at least 1 h. Then, cells were centrifuged and washed in D-Hanks, followed by incubation in 0.5 mL of FxCycle PI/RNase staining solution for 15–30 min at room temperature. Finally, without washing, the analysis was performed using 488-nm excitation and 585/42-nm emission or the equivalent on the Guava easyCyte HT system (Millipore).

Cells were lysed with RIPA buffer with proteinase inhibitors. Protein concentrations were measured using the Bradford reagent (Sigma; Merck). Subsequently, 20 μg of total protein was subjected to 10% SDS-PAGE. Then proteins were transferred to nitrocellulose membranes and blocked with 5% non-fat milk at room temperature for 1 h. Membranes were incubated with antibodies against ENO1 (ABclonal, 1:2,000), HIF1a (Sigma, 1:2,000), β-actin (Sigma A5316, 1:5,000), and GAPDH (Proteintech, 1:5,000) at 4°C overnight, followed by secondary antibody (Thermo Fisher Scientific) at room temperature for 60 min.

We seeded 3.0 × 10⁴ cells/well into the upper chamber of 24-well Corning^® FluoroBlok TM Cell Culture Inserts (NY, United States) to measure cell migration. The lower chambers were filled with DMEM/F-12 medium supplemented with 10% fetal bovine serum to serve as a chemoattractant. Finally, cells that migrated to the other side of the filter were stained with 0.5% crystal violet and counted under an inverted fluorescence microscope.

Cells were plated in 96-well plate and cultured overnight. Wounds were made in confluent monolayer cells using 96 Wounding Replicator (VP Scientific) and cells were cultured in medium supplemented with 1% FBS. Wound healing was detected at 0, 24 h or 0, 48 h for overexpression or knockdown, respectively. The representative fields at different time points were photographed.

The TargetScan database¹ was used to predict potential targets of HIF1a (Jaiswal et al., 2019). The 3′ untranslated region (UTR) of the ENO1 sequence was amplified and cloned into a pGL4.10-report vector (Promega Corporation, Madison, WI, United States). Equal quantities of the pGL4.10-3UTR-ENO1 and Renilla expression vector pRL-TK (Promega Corporation) were co-transfected into 293T cells. The luciferase activity was measured at 48 h post-transfection using the Dual-Glo Luciferase Assay system (Promega Corporation).

We subcutaneously injected 2 × 10⁶ transfected TPC1 cells into the forelegs of immunodeficient nude mice (male BALB/c, 4-week-old). At the indicated times, thyroid tumors were isolated and weighed following international criteria. Animal studies were conducted in compliance with the regulations on management of animal welfare set by the Ethics Committee of National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College. The protocol for animal experiments was approved by the Ethics Committee of National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College.

Total RNA was extracted from TPC1 cells incubated with exosomes using TRIzol reagent (Invitrogen, Carlsbad, CA, United States). Agarose electrophoresis was used to determine the total RNA’s integrity, and a NanoDrop (Thermo Scientific NanoDrop 2000 Microvolume Spectrophotometer, RRID:SCR_018042) was used for quality control and quantification. A sequencing library was constructed using an RNA library construction kit (NEB, United States). Their sequences were analyzed using Illumina HiSeq^TM 2000 (Wistar Genomics Facility, RRID:SCR_010205; Illumina Inc., San Diego, CA, United States).

TPC1 cells were fixed in 1% formaldehyde for 10 min at 4°C. Cells were then sonicated to shear genomic DNA, followed by incubating overnight with 5 μg of Rb IgG or HIF1α antibody (Sigma, 1:2,000). The resulting complexes were precipitated using Fastflow G-Sepharose beads (GE Healthcare, Little Chalfont, United Kingdom), eluted, purified, and analyzed using PCR.

According to their P-values, a GSEA algorithm was used to identify the enrichment of specific functions in the list of genes pre-ranked to test differences in expression between cell lines. The statistical significance of the enrichment score was calculated by permuting the genes 1,000 times as performed by the GSEA software. To collapse each probe set on the array to a single gene, we selected the probe with the highest variance among multiple probes corresponding to the same gene. Gene sets were derived from the REACTOME pathway database.

The StarBase V3.0 project² was used to analyze expression levels of ENO1 and CST1. Correlation between ENO1 and CST1 was obtained from 510 samples in thyroid carcinoma.

The analysis was performed using GraphPad Prism 6.0 software. Differences were analyzed using the Student’s t-test or one-way ANOVA. Differences were statistically significant when P < 0.05.

To study the function of ENO1 in thyroid carcinoma, we first analyzed the THCA database to determine the relevance of ENO1 in thyroid carcinoma. We found upregulation of ENO1 in thyroid carcinoma compared with normal samples (Figure 1A). And we also found that the higher level of methylation of ENO1’s promoter in normal group than tumor group (Figure 1B). IHC staining showed that ENO1 was upregulated in thyroid carcinoma samples as compared with normal samples (Figure 1C and Table 1). The expression ENO1 was not associated with age, gender, T classification and lymph node metastasis (Table 2). qRT-PCR and western blotting were also performed to confirm these results; there were higher ENO1 mRNA and protein levels in thyroid carcinoma samples (Figures 1D,E). These findings suggest that ENO1 might be an oncogene.

To elucidate the role of ENO1 in thyroid carcinoma progression, small interfering RNAs were identified to downregulate ENO1 expression in TPC1 and BCPAP cells. Western blotting showed high knockdown efficiency of siENO1 in both cell lines (Figure 2A). Then, a CCK8 assay was performed to measure cell proliferation. Depletion of ENO1 suppressed TPC1 and BCPAP thyroid carcinoma cells’ proliferation to nearly half that of control cells (Figure 2B). The colony formation assay demonstrated that fewer colonies were formed in siENO1 cells (Figures 2C,D).

Regarding migration ability, transwell results showed that the percentage of migration cells was decreased (greater than 50% in both cells) with ENO1 depletion compared to control cells (Figures 2E,F). In addition, wound healing assays demonstrated that ENO1 knockdown suppressed the migration capacity of TPC1 and BCPAP cells (Supplementary Figure 1A). To explore the function of ENO1 in vivo, a tumor formation assay was performed. We used ENO1 shRNA to downregulate the expression of ENO1, and the tumor was isolated for measurements at indicated times. Isolated tumors only weighed 0.1 g in shENO1 samples compared with 0.4 g in the control samples (Figures 2G–I). These findings further suggest an oncogenic role for ENO1.

Because the cell cycle is frequently disturbed by oncogenes, we performed flow cytometry. The depletion of ENO1 decreased the number of cells in the S phase, while the G0/G1 cells’ percentage increased (Figure 3A). Apoptotic cells significantly increased with ENO1 depletion. Compared with control cells, the percentage of apoptosis nearly doubled in siENO1 cells in bPC1 and BCPAP cells (Figure 3B). These findings suggest that cell cycle progression was severely impaired by ENO1 depletion.

To confirm the knockdown results, we performed overexpression experiments. We performed qRT-PCR and western blotting to demonstrate high overexpression levels of ENO1 in both TPC1 and BCPAP cells (Figures 4A–C). Next, the CCK8 assay demonstrated that ENO1 overexpression significantly improved cell viability compared with control cells from 1 to 4 days (Figures 4D,E). The ability of colony formation (Figures 4F,G) and Transwell migration (Figures 4H,I) in ENO1 overexpressing cells were consistent with the knockdown results (Figures 2D–F). Wound healing results also showed that ENO1 overexpression enhanced the migration capacity of TPC1 and BCPAP cells (Supplementary Figure 1B). We also performed flow cytometry analysis. Apoptosis was inhibited with ENO1 overexpression in both TPC1 and BCPCP cells (Figures 4J,K). Never the less, ENO1 accelerated cell cycle progression, with S stage cells increasing and G0/G1 stage cells decreasing (Figures 4L,M). These findings suggest that ENO1 regulates the cell cycle and apoptosis to promote cancer cell progression, functioning as an oncogene.

Next, we elucidated the mechanism of ENO1’s oncogenic role. ENO1 is a glycolysis-associated enzyme that participates in various cancers (Nakajima et al., 1986; Tu et al., 2010; Song et al., 2014; Fu et al., 2015; Liu et al., 2015; Principe et al., 2015; Zhu et al., 2015, 2018; Niccolai et al., 2016; Qian et al., 2017; Zhan et al., 2017); therefore, we focused on glycolysis-associated pathways; the mTOR/HIF1α pathway is also involved in glycolysis (Majumder et al., 2004; Semenza, 2011; Chi, 2012; Cheng et al., 2014; Courtnay et al., 2015; Yang et al., 2015; Xiao et al., 2017). Therefore, we examined ENO1 expression in TSC1 knockout MEF cells and found that TSC1 depletion aberrantly activated the mTOR pathway (Cao et al., 2020). Western blotting showed that knockout of TSC1 upregulated ENO1 expression and the phosphorylation of S6, the downstream of mTOR pathway (Figure 5A), as did HIF1α, another downstream molecule of the mTOR/HIF1α pathway. To confirm these results, we used rapamycin to inhibit the mTOR pathway (Edwards and Wandless, 2007). ENO1 was downregulated upon rapamycin addition, more significantly in TSC1-depleted cells (Figure 5B).

To explore the relationship between ENO1 and HIF1α, an RNA interference assay was performed. With HIF1α depletion, ENO1 was downregulated in both TPC1 and BCPAP cells (Figure 5C). The addition of rapamycin further reduced ENO1 expression. Overexpression of HIF1α rescued ENO1 expression both at the protein and mRNA levels (Figure 5D). We also performed a luciferase assay. The luciferase activity increased in HIF1α-transfected cells compared with control cells (Figure 5E). The ChIP-qPCR assay demonstrated that the HIF1α targeted the promoter region of ENO1 (Figures 5F,G). These findings suggest that ENO1 is located downstream of and is regulated by the mTOR/HIF1α pathway.

To further explore the mechanism, we performed differential expression analysis between ENO1-depleted cells and control cells. Variations of mRNA expression between the two groups were plotted, and CST1 and CST4 are the most decreased genes by ENO1 knockdown (Figure 6A). GESA analysis showed that ENO1 depletion altered downregulated including G2M checkpoints and glycolysis signaling pathways (Figure 6B). We chose the genes with a maximum difference for further analysis (Figure 6C). Correlation analysis of the THCA database showed that CST1 and CST4 expression were positively correlated with ENO1 expression (Figures 6D,E). qRT-PCR and immunoblotting confirmed that depletion of ENO1 downregulated CST1 and CST4 expression at the mRNA and protein levels (Figure 6F).

CST1 and CST4 were previously reported to be oncogenes (Jiang et al., 2018; Cui et al., 2019; Liu et al., 2019; Wang et al., 2020). Next, we studied the role of CST1 and CST4 in thyroid carcinoma. Western blotting confirmed the high knockdown efficiency of siCST1 in both TPC1 and BCPAP cells (Figure 7A). Cell viability analysis with CCK8 assay showed that depletion of CST1 significantly reduced cell viability (Figure 7B). However, we found that knockdown of CST4 could not affect the proliferation of THCA cells (Supplementary Figure 2). Colony numbers also decreased in siCST1 cells (Figures 7C,D). Invasion ability was also inhibited with CST1 depletion, decreasing to nearly half of the control cell levels (Figures 7E,F). These findings suggest that CST1 is an oncogene in thyroid carcinoma.

To further clarify the relationship between ENO1 and CST1 in thyroid carcinoma, we combined overexpression and knockdown assays in TPC1 cells. Western blotting showed that ENO1 overexpression upregulated CST1 expression and reduced it after shCST1 treatment (Figure 8A). Next, cell viability and colony formation assays were performed. Knockdown of CST1 significantly reduced cell viability enhanced with ENO1 overexpression to control level (Figures 8B,C). Invasion ability was also attenuated by CST1 depletion (Figure 8D). Finally, in vivo tumorigenicity was measured. Following the previous result (Figure 2G), overexpression of ENO1 resulted in larger tumor sizes (Figures 8E–G), suggesting that ENO1 was an oncogene. As predicted, tumor size enlargement was reduced to control levels with shCST1 treatment (Figures 8E–G). These findings suggest a synergistic role between ENO1 and CST1 in thyroid carcinoma progression.