Transcriptomic data for thyroid cancer were obtained from The Cancer Genome Atlas (TCGA) via the Genomic Data Commons (GDC) portal, and from the Gene Expression Omnibus (GEO) database (GSE138042, n = 95). Raw count data were normalized using the transcripts per million (TPM) method. EMT scores for each sample were calculated based on the expression levels of a predefined set of epithelial and mesenchymal marker genes, as previously described [16]. Pearson correlation analysis was performed between the expression levels of all genes and the EMT scores across samples. Genes with correlation coefficients greater than 0.8 (R $>$ 0.8, p 0.05) were considered to be strongly associated with EMT. Selected genes were visualized using heatmaps generated in R (version 4.3; R Foundation for Statistical Computing, Vienna, Austria) with the heatmaps package.

The thyroid cancer cell lines used in this study (KTC-1, BCPAP, BHT-101) were purchased from the Cell Bank of the Chinese Academy of Sciences, Shanghai. TPC-1 cell line was obtained from Haixing Biosciences Co., Ltd. The human normal thyroid epithelial cell line (HTori-3) was purchasded from Shanghai Enzyme-linked Biotechnology Co., Ltd. The cancer cell and HTori-3 cells were maintained in RPMI-1640 (11875093, Gibco, Waltham, MA, USA) or DMEM (11965092, Gibco, Waltham, MA, USA) supplemented with 10% fetal bovine serum (SH30071.03, HyClone, Cytiva, Logan, UT, USA). Cultures were incubated at 37 °C in a humidified 5% CO₂ atmosphere and passaged at 70–80% confluence using standard trypsinization. The cells were routinely confirmed as mycoplasma-negative and maintained with 1% penicillin–streptomycin (15140122, Gibco, Waltham, MA, USA). Antibiotics were omitted during transfection. All cell lines were validated by STR profiling.

For shRNA-mediated knockdown, cells were seeded 24 h before transfection to reach 40–60% confluence. Plasmid-encoded PTTG1IP shRNA (OriGene, Rockville, MD, USA) was introduced with Lipofectamine 3000 (L3000008, Invitrogen, Waltham, MA, USA) prepared in Opti-MEM (31985062, Gibco, Waltham, MA, USA) according to the manufacturer’s instructions. Medium was replaced with complete growth medium after 6–8 h, and cells were collected 3–5 days later for downstream assays.

Cells were seeded into 96-well plates at 2000 cells per well in 100 µL growth medium and cultured for 5 days. For each condition, 5 technical replicates were set up per experiment, with the experiments repeated independently three times. On the day of measurement, CCK-8 reagent (CK04-01, Dojindo Laboratories, Kumamoto, Japan) was added directly to the culture medium at 10% v/v (10 µL per well). After 60 min incubation at 37 °C, absorbance at 450 nm was recorded using a microplate reader. Blank wells containing medium plus CCK-8 (no cells) were used for background subtraction. Values were normalized to vehicle controls and reported as fold changes relative to the control.

Cells were washed with ice-cold PBS (ST476, Beyotime Biotechnology, Shanghai, China) and lysed in RIPA buffer (P0013B, Beyotime Biotechnology, Shanghai, China) freshly supplemented with a protease inhibitor cocktail (P1005, Beyotime Biotechnology, Shanghai, China). Lysates were cleared (12,000 $\times$g, 10 min, 4 °C), and protein concentration was determined using a BCA kit (#23235, Thermo Fisher, Waltham, MA, USA). Equal protein amounts (30 µg) were mixed with 4$\times$ Laemmli buffer, heated at 95 °C for 5 min, and resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) (8–12% polyacrylamide gels). Proteins were transferred to polyvinylidene difluoride (PVDF) membranes (pre-activated in methanol) using wet transfer (100 V, 2 h). Membranes were blocked with 5% Bovine Serum Albumin (BSA; A7030, Sigma-Aldrich, St. Louis, MO, USA) in Tris-buffered saline with Tween-20 (TBST) (TBS + 0.1% Tween-20) for 1 h at room temperature, then incubated overnight at 4 °C in TBST + 5% BSA with the following primary antibodies and dilutions: PTTG1IP (1:500, 12575-1-AP, Proteintech, Rosemont, IL, USA), SNAIL (1:1000, 3879, Cell Signaling Technology, Danvers, MA, USA), TWIST1 (1:1000, 90445, Cell Signaling Technology, Danvers, MA, USA), Vimentin (CST #5741, 1:1000), RAD51 (1:1000, 8875, Cell Signaling Technology, Danvers, MA, USA), ATM (1:1000, 2873, Cell Signaling Technology, Danvers, MA, USA), FANCA (1:1000, 14657, Cell Signaling Technology, Danvers, MA, USA), POLD1 (1:200, sc-374025, Santa Cruz Biotechnology, Dallas, TX, USA), and GAPDH (1:10000, #60004-1-Ig, Proteintech, Rosemont, IL, USA). After three 10 min TBST washes, HRP-conjugated secondary antibodies (CST anti-rabbit #14708 or anti-mouse #14709, 1:10,000, Cell Signaling Technology, Danvers, MA, USA) were applied for 2 h at room temperature. Chemiluminescent substrates were used to reveal proteins, and the signals detected with the iBright CL1000 imaging system (Invitrogen, Waltham, MA, USA) or film-based detection.

Total RNA was extracted with the QIAGEN Total RNA Purification Kit (74104, Qiagen, Hilden, Germany) and treated on-column with DNase I (79254, Qiagen, Hilden, Germany) to remove genomic DNA. RNA (500 ng per sample) was reverse-transcribed using the Vazyme Reverse Transcription Kit (R423-01, Vazyme Biotech, Nanjing, China) according to the manufacturer’s protocol. Quantitative PCR was performed with Vazyme SYBR Green reagent (Q712-02, Vazyme Biotech, Nanjing, China) in 10–20 µL reactions on a real-time cycler, using the following primer pairs: PTTG1IP: F-5^′-GTCTGGACTACCCAGTTACAAGC-3^′, R-5^′-CGCCTCAAAGTTCACCCAA-3^′; BRCA1: F-5^′-GAAACCGTGCCAAAAGACTTC-3^′, R-5^′-CCAAGGTTAGAGAGTTGGACAC-3^′; BRCA2: F-5^′-CACCCACCCTTAGTTCTACTGT-3^′, R-5^′-CCAATGTGGTCTTTGCAGCTAT-3^′; RAD51: F-5^′-CAACCCATTTCACGGTTAGAGC-3^′, R-5^′-TTCTTTGGCGCATAGGCAACA-3^′; RAD51D: F-5^′-CCAGCACTCGGATTCTCCTG-3^′, R-5^′-TTGGCTGTCGGGAAGATTTGG-3^′; RAD51AP1: F-5^′-TGGTGGTGTTCAAGGGAAAAG-3^′, R-5^′-AGGTGCAAAGTCTGGTTCAGT-3^′; FANCA: F-5^′-GGCACACAGTATGTTCTCCCG-3^′, R-5^′-TTGTACGTGAAGATGCCACAC-3^′; FANCI: F-5^′-CCACCTTTGGTCTATCAGCTTC-3^′, R-5^′-CAACATCCAATAGCTCGTCACC-3^′; POLD1: F-5^′-ATCCAGAACTTCGACCTTCCG-3^′, R-5^′-ACGGCATTGAGCGTGTAGG-3^′; GAPDH: F-5^′-GGAGCGAGATCCCTCCAAAAT-3^′, R-5^′-GGCTGTTGTCATACTTCTCATGG-3^′. Ct values were normalized to GAPDH, and relative expression was calculated using the 2^{-Δ⁢Δ⁢Ct} method.

Cells were seeded onto glass coverslips and allowed to adhere overnight. After fixation with 4% paraformaldehyde for 20 min at room temperature, the cells were rinsed with PBS and permeabilized with 0.3% Triton X-100 in PBS for 10 min. Non-specific binding was blocked using 5% BSA for 1 h at room temperature. Samples were subsequently incubated overnight at 4 °C with primary antibodies: PTTG1IP (1:200, 12575-1-AP, Proteintech, Rosemont, IL, USA), CTTN (1:200, 11381-1-AP, Rosemont, IL, USA), and p-H2AX (1:200, 9718S, Cell Signaling Technology, Danvers, MA, USA). They were then washed with PBS and incubated with species-appropriate, fluorophore-conjugated secondary antibodies for 2 h at room temperature, followed by 4^′,6-diamidino-2-phenylindole (DAPI) counterstaining for 10 min. Coverslips were mounted in anti-fade medium before imaging. Negative controls without primary antibody were included to assess background fluorescence.

To examine PTTG1IP occupancy at selected loci, ChIP was carried out with the Transcription Factor iDeal ChIP-seq Kit (Diagenode, Denville, NJ, USA) as recommended by the supplier. Cells were fixed in 4% paraformaldehyde for 10 min at room temperature, nuclei were isolated, and chromatin was sonicated into 200–500 bp fragments. Aliquots of sheared chromatin were incubated overnight at 4 °C with anti-PTTG1IP antibody or control immunoglobulin G (IgG). After sequential washes, immune complexes were eluted and cross-links reversed with proteinase K treatment. DNA was purified and analyzed by SYBR Green qPCR using locus-specific primers. Enrichment was quantified relative to the input and IgG controls. All ChIP-qPCR reactions were run in technical replicates, and amplicon specificity was verified by melt-curve analysis.

After harvesting, cells were rinsed three times in chilled 1$\times$ PBS and centrifuged to obtain a pellet. This was incubated with 5 µM PI (Sigma, St. Louis, MO, USA) for 15 min at ambient temperature and then analyzed by flow cytometry.

Analyses were performed with GraphPad Prism v10 (GraphPad Software, San Diego, CA, USA) and R (version 4.3, R Foundation for Statistical Computing, Vienna, Austria)). Data are reported as the mean $\pm$ standard error of the mean (SEM). Two-sided tests were used throughout. Unpaired (or paired, where applicable) Student t test was used for two groups and analysis of variance (ANOVA) for $\ge$3 groups, with appropriate correction for multiple comparisons. Where multiple hypotheses were tested (e.g., transcriptome-wide correlations), p values were adjusted by the Benjamini–Hochberg false discovery rate (FDR) procedure. Significance was set at p 0.05, or q 0.05 after FDR.

We prioritized EMT-related genes by deriving an EMT score for every sample [17] in two independent datasets—The Cancer Genome Atlas–Thyroid Carcinoma (TCGA-THCA) and GEO GSE138042 (n = 95)—followed by gene-wise Pearson correlation analysis. Genes showing strong associations ($|$R$|$ $>$ 0.8) were retained and displayed as ranked heatmaps, revealing a coherent EMT signature in both datasets (Fig. 1A,B). Intersection of the two gene sets identified a compact overlap of candidates, among which PTTG1IP consistently ranked at the top by virtue of its correlation strength and cross-cohort reproducibility (Fig. 1C).

Fig. 1.

PTTG1IP emerges as a leading EMT-associated gene in thyroid cancer. (A) Heatmap of genes most strongly correlated with the EMT score in the TCGA-THCA cohort. (B) Heatmap of EMT-correlated genes in the GEO cohort GSE138042 (n = 95), using the same pipeline as in (A). (C) Venn diagram showing the overlap of EMT-associated genes between TCGA-THCA and GSE138042. PTTG1IP was among the top shared candidates. (D) Correlation map for PTTG1IP versus key EMT regulators in TCGA-THCA, including transcription factors (ZEB1/2, TWIST1/2, SNAI1/2/3), epithelial/mesenchymal markers (CDH1, CDH2, VIM), and matrix metalloproteinases (MMP2, MMP9). Pearson’s R values (and p-values, where indicated) summarize the association strength. PTTG1IP, Pituitary Tumor Transforming 1 Interacting Protein; EMT, epithelial-mesenchymal transition; TCGA, The Cancer Genome Atlas; THCA, Thyroid Carcinoma; ZEB1/2, zinc finger E-box-binding homeobox 1/2; TWIST1/2, twist family bHLH transcription factor 1/2; SNAI1/2/3, snail family transcriptional repressor 1/2; CDH, cadherin; VIM, vimentin. ns, no significant; *, p 0.05; ****, p 0.0001.

We then examined how PTTG1IP relates to the canonical EMT circuitry. PTTG1IP expression aligned positively with EMT-activating transcription factors (ZEB1, ZEB2, TWIST1, TWIST2, SNAI1, SNAI2, SNAI3) and mesenchymal markers (CDH2, VIM), while showing an inverse relationship with the epithelial determinant CDH1. It also correlated with matrix-remodeling enzymes (MMP2, MMP9) (Fig. 1D). Together, these patterns support a model in which PTTG1IP associates with mesenchymal programs and extracellular-matrix dynamics that are characteristic of EMT in thyroid cancer.

We next examined whether PTTG1IP is dysregulated in tumors. Analysis of bulk RNA-seq from TCGA-THCA showed higher PTTG1IP mRNA levels in thyroid carcinoma than in normal thyroid tissue (Fig. 2A). A patient-matched comparison yielded the same pattern: in paired samples, tumors generally exhibited increased expression relative to adjacent normal counterparts (Fig. 2B). Consistent with the transcript data, protein-level evidence from the Human Protein Atlas indicated stronger PTTG1IP immunoreactivity in thyroid cancer tissues than in normal thyroid (Fig. 2C). Together, these orthogonal readouts support the recurrent upregulation of PTTG1IP in thyroid cancer, consistent with a role in disease pathogenesis.

Fig. 2.

PTTG1IP is elevated in thyroid cancer at both RNA and protein levels. (A) Normalized PTTG1IP mRNA expression in normal thyroid versus thyroid carcinoma samples from the TCGA-THCA cohort. (B) Paired tumor–adjacent normal analysis within TCGA-THCA; each line connects samples from the same patient, illustrating the within-patient increase. (C) Representative immunohistochemistry (IHC) images from the Human Protein Atlas (HPA; https://www.proteinatlas.org/) showing stronger PTTG1IP immunostaining in thyroid cancer compared with normal thyroid tissue (scale bar = 500 μm). *, p 0.05; **, p 0.01.

We first profiled PTTG1IP across models and found higher mRNA levels in thyroid cancer cell lines (TPC-1, KTC-1, BCPAP) than in non-malignant thyroid cells (HTori-3) (Fig. 3A). To assess function, PTTG1IP was silenced using shRNA and the knockdown efficiency validated by qRT-PCR (Fig. 3B). Loss of PTTG1IP resulted in cell growth inhibition, as reflected by the CCK-8 assay result for TPC-1 and KTC-1 cells (Fig. 3C,D). Moreover, TUNEL assays revealed a significant increase in the percentage of apoptotic cells in PTTG1IP-silenced TPC-1 and KTC-1 cells compared to control cells (Fig. 3E), indicating that knockdown of PTTG1IP effectively promotes apoptosis. Taken together, these results indicate that PTTG1IP supports proliferative capacity and protects against apoptosis in thyroid cancer cells.

Fig. 3.

PTTG1IP promotes growth and cell survival in thyroid cancer cells. (A) qRT-PCR comparison of PTTG1IP mRNA in HTori-3 versus thyroid cancer lines (TPC-1, KTC-1, BCPAP) (n = 3). (B) The knockdown efficiency of PTTG1IP was demonstrated by qRT-PCR assay (n = 3). (C,D) Control or PTTG1IP-knockdown TPC-1 and KTC-1 cells were transfected and their viability assessed by CCK8 assay at Day 5 (n = 5; scale bar = 100 μm). (E) Apoptosis was quantified by TUNEL staining in control and shPTTG1IP TPC-1/KTC-1 cells (n = 3; scale bar = 50 μm). qRT-PCR, quantitative real-time reverse transcriptase PCR; HTori-3, human thyroid follicular epithelial cell line; TPC-1, KTC-1, BCPAP, human thyroid carcinoma cell lines; TUNEL, Terminal deoxynucleotidyl transferase dUTP nick end labeling. *, p 0.05; **, p 0.01; ***, p 0.001; ****, p 0.0001.

To clarify how PTTG1IP engages the EMT circuitry, we focused on its reported binding partners, CTTN (cortactin) and p53 [18, 19]. Consistent with the known roles of CTTN in actin dynamics, migration, and adhesion remodeling [20, 21], our cross-cohort screen also identified CTTN as an EMT-associated candidate (Fig. 1C). In TCGA-THCA, mRNA expression of PTTG1IP correlated positively with that of CTTN (Pearson R = 0.572, p 0.001) and TP53 (R = 0.619, p 0.001) (Fig. 4A,B). At the protein level, co-immunoprecipitation (Co-IP) confirmed an endogenous PTTG1IP–CTTN complex in thyroid cancer cells (Fig. 4C). A correlation overview further showed that PTTG1IP aligns with canonical EMT drivers (SNAIL and TWIST1) and mesenchymal markers (VIM) (Fig. 4D). Together, these data indicate that PTTG1IP couples cytoskeletal remodeling (via CTTN) with EMT-linked transcriptional programs, thereby supporting migratory and invasive phenotypes in thyroid cancer.

Fig. 4.

PTTG1IP physically associates with CTTN and aligns with EMT programs. (A) Scatter plot of PTTG1IP versus CTTN mRNA in TCGA-THCA (Pearson R = 0.572, p 0.001). (B) Scatter plot of PTTG1IP versus TP53 mRNA in TCGA-THCA (Pearson R = 0.619, p 0.001). (C) Co-immunoprecipitation demonstrating endogenous PTTG1IP–CTTN interaction, including IgG and input controls. (D) Western blot analysis of EMT markers (SNAI1, TWIST1, VIM) after PTTG1IP knockdown, with GAPDH serving as the loading control. CTTN, cortactin; IgG, immunoglobulin G; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

TP53 mutations are common in thyroid cancer and compromise the DNA damage response (DDR) to facilitate tumor progression [22, 23]. Building on prior reports that PTTG1IP can associate with p53 [14, 15, 17], as well as our evidence of a PTTG1IP–CTTN complex (Fig. 4C), we next investigated whether PTTG1IP can influence DDR programs. In TCGA-THCA, PTTG1IP expression showed positive correlations with key DDR genes, including BRCA1, BRCA2, RAD51, and FANCA (Fig. 5A). Notably, its binding partner CTTN exhibited similar correlations with the same DDR genes (Fig. 5B), suggesting the PTTG1IP–CTTN axis may align with DDR activity. Functionally, shRNA-mediated knockdown of PTTG1IP in thyroid cancer cells reduced DDR gene expression at both the mRNA level (qPCR; Fig. 5C,D) and protein level (Western blot; Fig. 5E). Mechanistically, ChIP–qPCR demonstrated occupancy of PTTG1IP at the promoter regions of BRCA1, BRCA2, RAD51, RAD51AP1 and ATM in TPC-1 and KTC-1 cells (Fig. 5F,G). Collectively, these results identify PTTG1IP as a modulator of DDR, potentially acting through its association with CTTN and direct engagement with DDR gene promoters. Such features may contribute to genomic instability and treatment resistance in thyroid cancer.

Fig. 5.

PTTG1IP modulates the DNA damage response via association with CTTN. (A) Correlation matrix showing PTTG1IP versus DDR genes in TCGA-THCA. (B) Correlation matrix showing CTTN versus the same DDR genes in TCGA-THCA. (C,D) qPCR analysis showing decreased mRNA levels for DDR genes after PTTG1IP knockdown in TPC-1 and KTC-1 cells (n = 3). (E) Western blot analyses showing reduced DDR protein levels following PTTG1IP knockdown. (F,G) ChIP–qPCR confirming PTTG1IP occupancy at the BRCA1, BRCA2, ATM, RAD51 and RAD51AP1 promoters in TPC-1 and KTC-1 cells (n = 3). RAD, radiation-sensitive; DDR, DNA damage response. ****, p 0.0001.

Given the connection between PTTG1IP and DDR pathways, we next examined whether the loss of PTTG1IP alters cellular responses to irradiation. Viability assays showed that irradiated cells with PTTG1IP knockdown displayed a more pronounced decrease in survival than irradiated controls (Fig. 6A). To determine if this effect was conserved in a more aggressive, TP53-mutant context, we repeated this experiment in BHT-101 anaplastic thyroid cancer cells and observed a similarly significant reduction in viability (Fig. 6A). Consistent with this observation, flow cytometry revealed a higher fraction of dead cells in the knockdown group after radiation exposure (Fig. 6B), a finding that was recapitulated in the TP53-mutant BHT-101 cell line (Fig. 6B). In addition, immunofluorescence for the double-strand break marker p-H2AX showed stronger DNA-damage signals in PTTG1IP-silenced cells following irradiation (Fig. 6C). Together, these findings indicate that suppression of PTTG1IP enhances radiation-induced cytotoxicity and DNA damage, suggesting this gene may be a potential target for improving the effectiveness of radiation therapy in thyroid cancer.

Fig. 6.

PTTG1IP knockdown sensitizes thyroid cancer cells to radiation. (A) Cell viability after irradiation with or without PTTG1IP knockdown. The knockdown group shows greater loss of viability (n = 5). (B) Flow-cytometric quantification of cell death following irradiation, showing higher death rates in cells with PTTG1IP knockdown (n = 3). (C) Immunofluorescence assessment of p-H2AX after irradiation, indicating increased DNA-damage signals in PTTG1IP-silenced cells (scale bar = 10 μm). *, p 0.05; **, p 0.01; ***, p 0.001; ****, p 0.0001.