EpCAM activates the ERK-EGR1 signaling axis and promotes TNF-α-induced the progression of anaplastic thyroid cancer

Abstract

Background

Epithelial cell adhesion molecule (EpCAM) is a pleotropic transmembrane glycoprotein comprising an extracellular domain (EpEX), a single transmembrane domain, and an intracellular domain (EpICD). For several types of cancer, high EpCAM expression is associated with tumor progression, metastasis, immune evasion, and overall poor prognosis. A rare but aggressive form of thyroid cancer, anaplastic thyroid cancer (ATC), has a mean survival time of only 3–6 months after diagnosis.

Methods

The levels of EpEX and EpICD cleavage in clinical patient samples was evaluated by using tissue microarray. Capture enzyme-linked immunosorbent assay (ELISA) was conducted to explore the of shedding EpEX and TNF-α. Western blotting and qRT-PCR was taken to determine the expression of proteins in ATC cell lines. RNA-seq was conducted, and differential gene expression and gene set enrichment analyses were subsequently performed to identify EpCAM knockout–associated molecular alterations and affected biological pathways. The functions of EpCAM or EpEX in ATC were investigated by proliferation, invasion, and stemness analysis. The combined therapy was validated via a tail vein injection method for the metastasis ATC models.

Results

In this study, we observed that ATC samples typically exhibited a marked loss of membrane EpEX along with increased nuclear and cytoplasm accumulation of EpICD, as compared to non-ATC samples. Furthermore, we found that EpEX induced phosphorylation of EGFR, HGFR and Wnt receptors in ATC cells to promote cell growth, invasion and stemness activity. EpCAM signaling also increases TNF-α expression and induces TNF-α cleavage, and clarify the crosstalk between EpCAM and TNF-α to regulate ERK-EGR1 axis signaling. The inhibition of EpCAM signaling suppressed regulated intramembrane proteolysis (RIP) of EpCAM and shedding of the EpEX and EpICD in ATC cells. Combined treatment of EpAb2-6 and BRAF inhibitor dabrafenib coordinately induced apoptosis, while also inhibiting invasion, stemness and lung metastasis, and prolong survival in an animal model of metastatic ATC.

Conclusions

This study uncovers the molecular mechanisms underlying promotion of anaplastic thyroid cancer (ATC) tumor progression by EpCAM signaling via the ERK-EGR1-TNF-α axis. Our findings suggest that the EpCAM neutralizing antibody may improve the therapeutic efficacy of BRAF inhibitor, dabrafenib, in ATC.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-025-07575-z.

Introduction

Epithelial cell adhesion molecule (EpCAM; also known as cluster of differentiation 326, CD326) is a transmembrane glycoprotein that is expressed at low levels on the basolateral membrane of normal epithelial cells and functions as an intercellular adhesion molecule [1–3]. The EpCAM protein comprises three functional domains, including an extracellular domain (EpEX), a single-transmembrane domain, and an intracellular domain (EpICD). EpEX and EpICD are cleaved from the full-length protein and released via a process called regulated intramembrane proteolysis (RIP); EpEX is released after EpCAM is cleaved by a disintegrin and metalloprotease 17 (ADAM17; also called tumor necrosis factor-α converting enzyme, TACE), while EpICD is generated after cleavage by the multi-subunit protease complex, γ-secretase [4]. In cancer cells, RIP is often regarded as the initial event in EpCAM signaling, which may be triggered by binding of free EpEX to cell surface-expressed EpCAM or by intercellular interaction between EpCAM molecules on adjacent cells [5–7]. Moreover, the soluble EpEX that is shed after RIP can act as a ligand for both EGFR and HGFR to induce AKT and ERK1/2 signaling [8–12].

Numerous reports have demonstrated that EpCAM signaling can promote cancer cell stemness, proliferation, migration, invasion, epithelial-mesenchymal transition (EMT) and metastasis in different types of cancer. These functions appear to be largely mediated by activation of key oncogenic signaling cascades, such as EGFR, HGFR, and Wnt pathways, as well as suppression of the antitumor T cell response in the tumor microenvironment via stimulation of the PD-1/PD-L1 pathway [8–15].

Thyroid carcinoma is recognized as the most prevalent malignant endocrine neoplasm, and its incidence has shown a notable upward trend in recent years, and it has a high 5-year overall survival rate. However, undifferentiated tumors of the thyroid follicular epithelium comprise a rare but aggressive form of thyroid cancer called anaplastic thyroid cancer (ATC). In total, ATC accounts for less than 2% of all thyroid cancer cases, but it has a disease-specific mortality of 98%, which is in sharp contrast to that of differentiated subtypes. Given the extremely rapid course of disease progression and poor treatment outcomes for ATC, end-of-life issues and plans for comfort care measures are an integral part of initial disease management. Early recognition of the disease is essential to allow for prompt initiation of therapeutic interventions [16, 17]. ATC commonly presents as a rapidly increasing neck mass that spreads locally, compresses adjacent structures, and has a tendency to disseminate to regional lymph nodes and distant sites [18]. Most commonly, ATC patients are diagnosed with invasive disease, and metastasis is identified in more than 50% of cases. These features render surgical interventions almost entirely ineffective, and the patients have historically been treated mainly with palliative approaches. The choices for treatment of ATC are highly limited, and the median survival time is only 3 to 6 months after diagnosis [19]. Since ATC has a very poor prognosis even with therapies such as surgery, radiation and systemic chemotherapy, novel treatments are urgently needed. To identify suitable therapeutic targets for the disease, it is important to understand the molecular mechanisms of ATC pathogenesis [20–22].

In this study, we report a role for EpCAM and EpEX in promoting ATC cell proliferation and invasion. We show that in ATC cells, EpCAM signaling activates EGFR and HGFR signal transduction, including downstream AKT and ERK pathways. EpEX also induces RIP activity of both TACE and γ-secretase. Importantly, we further found that an EpCAM-neutralizing antibody, EpAb2-6, attenuates EpEX shedding, RIP activity and EpCAM-induced signaling. Moreover, the combination of dabrafenib and EpAb2-6 cooperatively induces apoptosis, and inhibits invasion and stemness in ATC cells. Treatment with humanized EpAb2-6 (hEpAb2-6) markedly prolongs survival in a mouse model of metastatic ATC, and the combination of dabrafenib with hEpAb2-6 shows enhanced therapeutic efficacy compared to each agent alone.

Materials and methods

Chemicals and antibodies

Antibodies against human EpCAM (#93790), total ERK (#4695), Thr202/Tyr204-phosphorylated ERK (#4370), total AKT (#9272), Ser473-phosphorylated AKT (#4060), total HGFR (#8198), Tyr1234/1235-phosphorylated HGFR (#3077), total EGFR (#4267), Tyr845-phosphorylated EGFR (#6963), LRP6 (#3395), Ser1490-phosphorylated LRP6 (#2568), OCT4a (#2840), E-cadherin (#3195), Vimentin (#5741), Snail (#3879), Slug (#9585), TNF-α (#11948), and EGR1 (#4154) were obtained from Cell Signaling Technology. Antibodies against human α-tubulin (T5168) and β-tubulin (T8328) were obtained from Sigma Aldrich. Anti-human EpICD rabbit monoclonal antibody (E144) (ab32392), anti-human SOX2 (ab97959), and NANOG (ab109250) antibodies and Human TNF alpha ELISA Kit (ab181421) were obtained from Abcam. Anti-human EpEX mouse monoclonal antibody (MOC-31)(#248 M-10) was obtained from Cell Marque. IgG isotype control antibody (12371-MI) was obtained from Invitrogen. The SensoLyte 520 TACE (α–Secretase) Activity Assay Kit (AS-72085) was obtained from AnaSpec. Dabrafenib (HY-14660) was obtained from Med Chem Express (MCE). TAPI (metalloproteinase inhibitor) (S2215) and DAPT (γ-secretase inhibitor) (E7213) were obtained from Selleckchem.

Cell lines and culture

ATC cell lines MDA-T178 (P178), MDA-T254 (P254), MDA-T264 (P264), MDA-T265 (P265) were generated at MD Anderson Cancer Center and passaged as previously described [23, 24]. Each of the ATC cell lines has a different mutational profile: MDA-T178 (HRAS^Q61R and EIF1AX^A133fs), MDA-T254 (none identified), MDA-T264 (BRAF^D594N and NRAS^Q61R), and MDA-T265 (BRAF^V600E and PI3K^E542K). All tumor specimens were obtained from patients who underwent a surgical intervention at MD Anderson Cancer Center and provided written informed consent. The ATC cells were cultivated in RPMI1640 supplemented with 100 units/ml penicillin-streptomycin (P/S; Gibco), 10% fetal bovine serum (FBS; Gibco), nonessential amino acid mixture (Cambrex BioScience #MT25025CI), and 1 mM sodium pyruvate (Cambrex BioScience MT25000CI) at 37 °C in a humidified incubator with 5% CO₂.

EpCAM gene knockout

For the EpCAM knockout experiments, CRISPR/Cas9 gRNA constructs were purchased from Genescript. To produce the lentivirus, 293T cells were transiently transfected with CRISPR/Cas9 gRNA plasmids, including the EpCAM gRNA (target sequence: GTG CAC CAA CTG AAG TAC AC), packaging plasmid (pCMV-ΔR8.91) and an envelope expression plasmid (pMD.G). MDA-T264 cells were cultured with lentivirus-containing medium and selected with 2 μg/ml puromycin. Single-cell clones were isolated from the selected pool, and the expression of EpCAM was examined with qRT-PCR, Western blotting, and flow cytometry.

Production and purification of EpEX‑His recombinant protein

The methods for production and purification of EpEX recombinant protein were described previously [9]. In brief, recombinant protein was expressed and purified using the Expi293 expression system. Cells were grown in Expi293 expression medium, and protein expression was induced by the addition of an enhancer reagent. The supernatant was harvested by centrifugation. After centrifugation at 8000 g for 20 min at 4 °C, the supernatant was incubated with nickel-chelated affinity resin (Ni-NTA, Qiagen) for 2 h at 4 °C. The resin was washed with wash buffer containing 50 mM Tris-HCl (pH 8.0), 500 mM NaCl, and 20 mM imidazole, and the proteins were eluted with elution buffer containing 50 mM Tris-HCl (pH 8.0), 500 mM NaCl, and 250 mM imidazole.

Generation of monoclonal antibodies and purification of IgG

EpAb2-6 and EpAb3-5 are anti-EpCAM antibodies, which target EpEX and were developed by our lab at Academia Sinica, as described previously [25]. The experimental protocol was approved by the Committee on the Ethics of Animal Experiments of Academia Sinica (AS IACUC: 11–04-166). The humanized anti- EpCAM antibody EpAb2-6 (hEpAb2-6) was generated by our lab, the construction and expression of hEpAb2-6 as described previously [11, 12, 25].

Tissue specimens

A tissue microarray (TMA) of thyroid cancer (TH208a) was purchased from TissueArray.Com LLC and used for immunohistochemical (IHC) staining with anti-EpICD and EpEX antibodies. The thyroid cancer TMA included 26 cases of papillary thyroid carcinoma (PTC), 22 cases of follicular thyroid carcinoma (FTC), 12 cases of ATC, and 9 normal thyroid tissues. Clinical information for the tissue samples is summarized in Table 1.

Table 1.

Clinic pathological characteristics of thyroid carcinoma tissue microarray

Patient characteristics	Number
Sex
Female	42
Male	18
Histological subtype
Papillary thyroid carcinoma (PTC)	26
Follicular thyroid carcinoma (FTC)	22
Anaplastic thyroid carcinoma (ATC)	12
TNM stage
I	38
II	5
III	4
IV	13

Immunohistochemistry analysis

A formalin-fixed paraffin-embedded thyroid cancer TMA was used for EpICD and EpEX immunostaining. First, the TMA was deparaffinized and rehydrated. Then, antigen retrieval was carried out using 0.01 M citrate buffer, pH 3.0 and endogenous peroxidase activity was blocked by incubating the tissue sections with hydrogen peroxide (0.3%, v/v) for 20 min. The TMA samples were then incubated with either anti-EpICD rabbit monoclonal antibody E144 (dilution 1:100) for 120 min or mouse monoclonal antibody MOC-31 (dilution 1:200) for 60 min, followed by incubation with secondary antibody for 16 min. The IHC signal was generated using an OptiView DAB Detection Kit (Roche). Tissue sections were then counterstained with hematoxylin. The staining procedure was performed using a Roche Ventana BenchMark XT system.

Evaluation of immunohistochemical staining

A TMA sample was scored as positive if the tumor cells showed immunopositivity in the membrane, cytoplasm or nucleus when observed by pathologist who was blinded to the clinical information. Scoring of immunohistochemical staining was performed according to criteria from a previous report [19]. The proportions of positive cells were scored as follows: 0, < 10% cells; 1, 10–30% cells; 2, 30–50% cells; 3, 50–70% cells; and 4, > 70% cells showed immunoreactivity. TMA samples were also scored semi-quantitatively on the basis of staining intensity: 0, none; 1, mild; 2, moderate; and 3, intense. Finally, a total score (ranging from 0 to 7) was obtained by adding the scores for proportion of positive cells and intensity for each of the tissue sections. Each tissue section was further scored for cytoplasmic/nuclear EpICD and for membrane EpEX according to the same scoring scheme.

Capture ELISA

A 96-well ELISA plate was coated with capture antibody in coating buffer (0.1 M sodium bicarbonate, 0.08 M sodium carbonate, pH 9.5) at 4 °C for 6 h, and then blocked with blocking buffer (1% BSA in PBS) at 4 °C overnight. Serially diluted samples were added at a volume of 50 μl per well in blocking buffer and incubated for 2 h at room temperature. Then, biotinylated detection antibody was incubated for 2 h at room temperature. Horseradish peroxidase (HRP)-conjugated streptavidin was then reacted for 20 min at room temperature. After a thorough washing, the plates were incubated with substrate solution 3,3’ ,5,5’-tetramethylbenzidine (TMB) (BioFX) for 20 min. The reaction was stopped by adding 3 N HCL, and signal was detected by absorption at 450 nm. PBS was used to wash the wells between each step. The capture antibody was anti-hEpCAM-IgG-1 (produced in-house), and the detection antibody was anti-EpEX antibody EpAb3-5 (produced in-house).

Protein extraction and Western blotting

Whole-cell extracts were prepared with RIPA buffer (50 mM Tris-HCl pH7.4, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 2 mM EDTA, 50 mM NaF). Protein concentrations of the cell lysates were determined by the BCA assay. The lysates were separated on a 10% polyacrylamide gel and then transferred to PVDF membrane. The membrane was blocked for 1 h with 3% BSA in PBST, then incubated overnight with primary antibodies. Horseradish peroxidase-conjugated secondary antibodies (Millipore) were applied as appropriate, and the membranes were incubated at room temperature for 1 h. The protein bands were subsequently visualized with chemiluminescence reagents (Millipore) and detected on a BioSpectrum 600 Imaging system (UVP). The protein level was quantified from band intensity using Gel-Pro analyzer 3.1 (Media Cybernetics).

Immunofluorescence assay

For immunofluorescence experiments, 2 × 10⁴ cells were seeded on 12-mm coverslips in 24-well culture plate. After treatment, the cells were fixed with 2% paraformaldehyde and then incubated with 0.1% Triton X-100 to increase the permeability of cell membrane. Then, the samples were blocked with 3% BSA for 1 h at room temperature and stained with primary antibody at 4 °C overnight. Slides were washed and stained with FITC- or Alexa 568-conjugated secondary antibody and DAPI for 40 min. Coverslips were then washed with PBS and mounted in Mounting Solution (Dako). The slides were examined with a confocal microscope Zeiss LSM 880 with Airyscan (Inverted).

Cell viability assay

Cell viability was assayed by measuring mitochondrial dehydrogenase activity with the WST-1 (4-[3-(4-lodophenyl)-2-(4-nitrophenyl)-2 H-5-tetrazolio-1, 3-benzene disulfonate) assay. Cells were seeded in 96-well plates at a density of 10⁴ cells/well and cultured for indicated times. At the end of the treatment period, 10 μl of WST-1 proliferation reagent (5 μg/ml) was added to each well and incubated for 1 h at 37 °C. Following the incubation, the absorbance of each well was detected at 450 nm using a spectrophotometric microplate reader.

Transwell invasion assay

Cell invasion activity was assayed with 8-μm pore size Transwell migration chambers (Millicell) with 10% Matrigel. Cells (1 × 10⁵) were added to the upper chamber in 500 μl serum-free RPMI. Then, 700 μl ATC complete medium was added to the lower chamber to serve as a chemoattractant. Migration and invasion were allowed to proceed for 16 h at 37 °C in a standard cell culture incubator. After the incubation, cells were removed from the upper surface of the membrane with a cotton swab, and the cells that had migrated to the lower surface were stained with 0.05% (w/v) crystal violet in 4% paraformaldehyde (in 1× PBS) for 15 min and washed with water. Membranes were dried for 15–20 min before at least four random fields on the membrane were examined at high power. The cells in each field were counted.

γ-Secretase activity assay

The activity of γ-secretase was measured using a previously described protocol [26]. In brief, cells were transiently transfected with control plasmid or tetracycline-inducible γ-secretase plasmid harboring luciferase reporters. Cells were then treated with 1 μg/ml tetracycline overnight, followed by experimental treatments as indicated. Cells were lysed using passive lysis buffer and then subjected to a luciferase assay.

TOP/FOP Wnt activity assay

Cells (2 × 10⁴ cells per well) were seeded in 24-well plates. Plasmids were transfected using Lipofectamine 3000 (Invitrogen). A luciferase reporter (TOPflash or FOPflash) was co-transfected with the pGL4.74 hRluc-TK construct (Renilla luciferase). Signals were generated using a dual luciferase reporter assay kit (Promega). Transcriptional activity was determined as the expression of firefly luciferase normalized to the Renilla luciferase levels at 48 h post-transfection.

Apoptosis assays

Cells were seeded, and antibodies and/or dabrafenib were treated for 24 h. Apoptotic cells were detected by Annexin V-FITC and propidium iodide (PI) staining analyzed with flow cytometry (Thermo Attune NxT). Early apoptosis was measured according to the Annexin V-FITC Apoptosis Detection kit II (BD Pharmingen). Late apoptotic nuclei were detected by PI staining.

RNA extraction, cDNA synthesis, quantitative reverse transcription polymerase chain reaction (qRT‑PCR)

Total RNA extraction, first-strand cDNA synthesis, and SYBR-green-based real-time PCR were performed according to the manufacturer’s instructions for each reagent. To extract total RNA, cells were lysed using TRIzol reagent (Invitrogen), and proteins and phenol were removed from TRIzol using chloroform. After centrifugation, the top colorless layer was collected and mixed with isopropanol to precipitate the RNA pellet. The RNA pellet was then washed with 70% ethanol, air-dried at room temperature, and dissolved in RNase-free water. For first strand cDNA synthesis, 5 μg of total RNA was used for reverse transcription with oligo(dT) primer and SuperScriptIII reverse transcriptase (Invitrogen) at 50 °C for 60 min. Target gene levels were evaluated by quantitative PCR (qPCR) using LightCycler 480 SYBR Green I Master Mix (Roche) and a LightCycler480 System (Roche). ACTB mRNA expression served as an endogenous housekeeping control and was used to normalize all qPCR reactions. The qPCR was run for 95 °C for 5 min, followed by 40 cycles of denaturation at 95 °C for 10 s, annealing at 60 °C for 10 s and extension at 72 °C for 30 s. The final results were calculated from three independent experiments. Primer sequences used for qPCR were as shows in Supplementary Table S1.

RNA sequencing

RNA purity and quantities were measured using a SimpliNano™-Biochrom Spectrophotometer (Biochrom, MA, USA). RNA degradation and integrity were monitored by Qsep 100 DNA/RNA Analyzer (BiOptic Inc., Taiwan).

A total of 1 μg total RNA per sample was prepared for sequencing. Sequencing libraries were generated using KAPA mRNA HyperPrep Kit (KAPA Biosystems, Roche, Basel, Switzerland) following the manufacturer’s recommendations; index codes were added to identify sequences from each sample. Briefly, mRNA was purified from total RNA using magnetic oligo-dT beads. Captured mRNA was fragmented by incubating at high temperature in the presence of magnesium in a solution containing KAPA Fragment, Prime and Elute Buffer (1×). Random hexamer primers were used for first strand cDNA synthesis. Combined second strand synthesis and A-tailing was performed to convert cDNA:RNA hybrids to double-stranded cDNA (dscDNA). dUTP was incorporated into the second cDNA strand, and dAMP was added to the 3’ ends of the resulting dscDNA. Then, dsDNA adapters with 3’dTMP overhangs were ligated to library insert fragments in order to generate library fragments carrying the adapters. To select cDNA fragments of 300–400 bp, the library fragments were purified with KAPA Pure Beads system (KAPA Biosystems, Roche, Basel, Switzerland). The library carrying appropriate adapter sequences at both ends was amplified using KAPA HiFi HotStart ReadyMix (KAPA Biosystems, Roche, Basel, Switzerland) along with library amplification primers. The strand containing dUTPs was not amplified, which allowed for strand-specific sequencing. PCR products were purified using KAPA Pure Beads system, and the library quality was assessed on the Qsep 100 DNA/RNA Analyzer (BiOptic Inc., Taiwan). The library quality was assessed on the Qubit ® 2.0 Fluorometer (Thermo Scientific) and Agilent Bioanalyzer 2100 system. Finally, the library was sequenced on an Illumina NovaSeq6000 platform, and 150 bp paired-end reads were generated.

Cross-sample normalization was performed on the read count matrix across samples, and differentially expressed genes (DEGs) were identified through pairwise comparisons between conditions (wild type vs. EpCAM gene knockout) using the R package “DESeq2” [27]. Genes with a log2 fold change in expression greater than 1 and an adjusted p-value less than 0.05 were considered differentially expressed. Over-representation analysis of DEGs was conducted using the org.Hs.eg.db [28] database, including gene ontology (GO) and KEGG pathways information, and the R package “clusterProfiler” [29]. In addition, the gene set enrichment analysis (GSEA) was performed on the Molecular Signatures Database (MSigDB) [30]. The heatmap of selected DEGs was generated using the R package “pheatmap” [31] and the GSEA plot was visualized with the R package “enrichplot” [32].

EGR1 promoter-driven luciferase reporter construction

The genomic DNA of anaplastic thyroid carcinoma (ATC) cells was extracted using DNeasy blood & tissue kits (Qiagen), and used as template to amplify the EGR1 promoter by PCR. The primers 5′-AAAGGTACCGCTCAGTTCGTGCTCACTGC-3′ (EGR1-1358 forward primer) and 5′-AAACTCGAGGGGAACACTGAGAAGCGTG C-3′ (EGR1+120 reverse primer) were used to amplify the EGR1 promoter from −1358 to + 120 of the transcription start site (TSS) which was cloned into pGL4.18 [33]. The genomic DNA was extracted using a Genomic DNA Isolation Kit (NovelGene, TW) according to the manufacturer’s recommendations.

Sphere formation assay

Sphere formation by cancer cells was enriched as described previously [34]. Cells were seeded in ultra-low adherent plates and maintained in DMEM/F-12 supplemented with B27. The cells were treated as indicated in the figure legends. The entire culture medium (including the treatment) was changed every other day for treatments directly added to the medium. Cells were cultured for 7 days, and on the seventh day, spheres were counted and photographed under light microscopy (Olympus).

Extreme limiting dilution assay (ELDA)

An ELDA was performed under in vitro conditions as previously described [35]. Briefly, cells were serially diluted and then seeded (from 10³ to 10 cells per well) with appropriate medium into a sterile round-bottom non-tissue culture 96-well plate. The cells were then allowed to grow for 2 to 4 weeks (depending on the efficiency of sphere formation), and then the wells with spheres were assessed. Similarly, for experiments involving treatments, cells were seeded overnight and treatments were added for 15 days, as indicated in the figure legends. The sphere-containing wells were confirmed by color change of the medium and observation under light microscopy. Data were analyzed using ELDA software (http://bioinf.wehi.edu.au/software/elda/).

ATC animal models

MDA-T264 cells suspended in PBS were subcutaneously injected into the dorsolateral flank of 6-week-old female NOD/SCID mice. When tumors reached approximately 50–100 mm³, mice were randomized and treated with control IgG or hEpab2-6 (15 mg/kg) via tail vein injection twice weekly for two weeks. Tumors were then harvested for analysis.

The protocol for combination therapy in metastatic animal models was performed as described previously [12] with modifications. MDA-T264 cells in PBS were injected into 6 to 8 weeks old female NOD/SCID mice through the tail vein (4 × 10⁵ cells/mouse). Mice were then semi-randomly assigned to different treatment groups, keeping the body weights of each group similar. After 3 days, antibodies were administered through tail vein injection twice a week for 4 consecutive weeks. In addition, dabrafenib was administered daily by oral gavage for 5 days per week (treatment for 4 weeks). For the therapeutic study, tumor-bearing mice were treated with isotype control IgG1 (15 mg/kg), dabrafenib (20 mg/kg), hEpAb2-6 (15 mg/kg), or dabrafenib (20 mg/kg) combined with hEpAb2-6 (15 mg/kg). Animal care was carried out in accordance with the guidelines of Academia Sinica. The protocol was approved by the Committee on the Ethics of Animal Experiments of Academia Sinica.

Statistical analysis

Statistical analyses were performed using SPSS and GraphPad Prism (GraphPad Software). All data are presented as mean ± SD for the indicated number of experiments. Datasets of more than two groups were tested using one-way ANOVA, and unpaired Student’s t-test (indicated in the Figure legend) was used to compare experimental and control groups as appropriate. A two-tailed Student’s t-test was performed for comparisons of two groups that passed the normality test. P-values less than 0.05 were considered statistically significant.

Results

Immunohistochemical analysis of EpEX and EpICD expression in patient thyroid carcinomas and normal tissue samples

To investigate the clinical significance of EpEX and EpICD in thyroid cancer, tumor tissues from FTC, PTC and ATC patients were analyzed by IHC and compared to normal thyroid tissues. After IHC staining, photomicrographs showing EpEX and EpICD expression in thyroid cancer subtypes were captured (Fig. 1A) and scored. The percentages of cells positive for nuclear EpICD, cytoplasmic EpICD, membrane EpEX and lost membrane EpEX were assessed (Fig. 1Band Table 2). Among the set of 60 thyroid cancer tissues we examined, intense nuclear EpICD immunostaining was routinely observed in ATC tissues. In contrast, the FTC and PTC tissues showed mild EpICD nuclear staining. The FTC and PTC tissues showed intense staining for membrane EpEX, but little to no staining for membrane EpEX was observed in ATC tissues. The normal thyroid tissues showed abundant membrane EpEX staining and no apparent EpICD nuclear staining. Cytoplasmic EpICD staining was observed in all thyroid cancer and normal tissues. These different staining patterns suggest that nuclear accumulation of EpICD may be more prevalent in some pathological classifications of thyroid cancer, with especially high levels observed in ATC cases.

Fig. 1.

Immunohistochemical staining of EpEX and EpICD in patient thyroid carcinoma and normal tissue samples. (A) Representative photomicrographs show immunostaining of EpEX and EpICD in paraffin-embedded papillary thyroid carcinoma (PTC), follicular thyroid carcinoma (FTC), anaplastic thyroid carcinoma (ATC) and normal thyroid tissues. (B) comparison of membrane EpEX, EpICD and ESLI staining scores in different subsets of the thyroid cancer tissue microarray. Bar graphs show the IHC scores determined from immunohistochemical analysis of tissue sections in samples of normal thyroid tissue (n = 9), PTC (n = 26), FTC (n = 22), and ATC (n = 12). ESLI: loss of cell membrane EpEX + EpICDnuc + EpICDcyt. Statistical differences were determined by one-way ANOVA. N = 3 independent. All data are presented as mean ± sd. ***p < 0.001

Table 2.

Scatter plot analysis of membrane EpEX, EpICD and ESLI expression in thyroid cancer tissue microarray

Patient characteristics	PTC (mean ± SD)	FTC (mean ± SD)	ATC (mean ± SD)	Thyroid tissue (mean ± SD)
EpEX membrane	6.10 ± 0.45	6.37 ± 0.58	2.08 ± 2.90	6.19 ± 0.73
Loss of membrane EpEX	0.90 ± 0.45	0.64 ± 0.58	4.91 ± 2.90	0.81 ± 0.73
EpICD nucleus	0.23 ± 0.59	0.11 ± 0.36	3.81 ± 3.27	0.00 ± 0.00
EpICD cytoplasm	6.65 ± 0.45	6.53 ± 0.60	5.94 ± 1.43	6.37 ± 1.32
ESLI (loss of membrane EpEX + EpICDnuc + EpICDcyt)	7.78 ± 0.95	7.27 ± 0.86	14.67 ± 5.42	7.19 ± 1.32

Analysis of EpCAM expression in ATC cells

EpCAM gene and protein expression levels were evaluated in four ATC cell lines using qRT-PCR and Western blotting (Figure S1A and B). The results revealed a high level of EpCAM expression in the MDA-T264 cell line, whereas the other three ATC cell lines (MDA-T178, MDA-T245, and MDA-T265) showed very low levels of endogenous EpCAM expression. In addition, the surface expression of EpCAM was higher in MDA-T264 cells than in the other ATC cell lines, as evaluated by flow cytometry analysis (Figure S1C). Next, we analyzed the concentration of EpEX released by ATC cells into conditioned medium using capture ELISA. We found that only MDA-T264 cells released a high level EpEX into the conditioned medium (Figure S1D). Subsequently, we measured the EpICD nuclear translocation by immunofluorescence staining. While the result showed apparent nuclear translocation of EpICD in all ATC cells, the most robust nuclear translocation of EpICD was observed in MDA-T264 cells (Figure S1E). Taken together, these findings show that MDA-T264 cells have the highest EpCAM expression and EpCAM cleavage among the four ATC cell lines.

EpCAM signaling promotes ATC cell viability and invasion

To investigate the biological function of EpCAM in ATC cells, we constitutively overexpressed EpCAM in MDA-T178 and MDA-T245 cells, and we knocked out EpCAM in MDA-T264 cells. For overexpression experiments, the MDA-T178 and MDA-T245 cells were infected with vector control or EpCAM-flag-encoding lentivirus. Meanwhile, EpCAM knockout was performed in MDA-T264 cells by infecting the cells with lentivirus harboring CRISPR/Cas9 gRNA that targets distinct coding sequences of EpCAM. Successful EpCAM overexpression and knockout were confirmed by both Western blotting (Fig. 2A) and flow cytometry (Fig. 2B). The levels of EpEX released into conditioned media were measured by capture ELISA (Fig. 2C). Overexpression of EpCAM significantly increased cell viability, while EpCAM knockout significantly decreased cell viability according to the WST-1 assay. Both full-length EpCAM and EpEX promoted cell viability in a dose-dependent manner (Figure S2), and EpEX further rescued the reduced cell viability caused by EpCAM knockout (Fig. 2D). Furthermore, an increased number of invading cells was observed upon overexpression of EpCAM cells, as compared with controls. Knockout of EpCAM significantly decreased cell invasion activity, and EpEX recovers the downregulation effect of EpCAM knockout (Fig. 2E). We then investigated the effects of EpEX on cell viability and invasion, finding that cell viability and invasion activity were both upregulated in ATC cells after EpEX treatment (Fig. 2F and G). Taken together, these data demonstrate that EpCAM plays a critical role in promoting cell growth and invasion activity in ATC cells.

Fig. 2.

EpCAM signaling promotes ATC cell viability and invasion. The efficiencies of EpCAM overexpression (OE-EpCAM) in MDA-T178 and MDA-T245 cells and efficiency of endogenous EpCAM knockout (KO) in MDA-T264 cells were assessed by (A) Western blot and (B) flow cytometry. (C) concentrations of EpEX shed from indicated ATC cell lines in conditioned media were assayed by capture ELISA. (D) viability of indicated ATC cell lines (MDA-T264 KO cells were treated with 50 nM of EpEX) was assayed using WST-1 over a time course. (E) the invasion activities of indicated ATC cell lines (MDA-T264 KO cells were treated with 50 nM of EpEX) were assayed by transwell migration assays with matrigel. (F) indicated ATC cell lines were treated with EpEX (50 nM), and the cell viability was assayed using WST-1 over the indicated time course. (G) indicated ATC cell lines were treated with EpEX (50 nM) for 24 h, and cell invasion activities were assayed by transwell migration assays with matrigel. Statistically significant differences were identified by (c, d, e, F) two‑tailed Student’s t-test or (g) one-way ANOVA. N = 3 independent experiments. All data are presented as mean ± sd. *p < 0.05; **p < 0.01; ***p < 0.001

EpCAM activates EGFR, HGFR and Wnt signaling pathways in ATC cells

Previously, we found that EpCAM signaling can promote colorectal cancer progression through EGFR and HGFR pathways [8, 9, 11, 12]. Therefore, we tested how EpCAM affected EGFR and HGFR signaling activity levels in ATC cells. In particular, we compared EpCAM-overexpressing MDA-T245 and EpCAM-knockout MDA-T264 cells with parental cell lines. The parental MDA-T245 cells express low levels of EpCAM, while parental MDA-T264 cells express high levels of EpCAM. Increases in the levels of phosphorylated EGFR, HGFR, AKT and ERK were observed upon EpCAM overexpression in MDA-T245 cells, but the total amounts of these proteins were unchanged. Conversely, knockout of EpCAM in MDA-T264 cells decreased the levels of phosphorylated EGFR, HGFR, and ERK (Fig. 3A). Since we saw a clear effect of EpCAM expression level on oncogenic signaling, we further determined whether overexpression or knockout of EpCAM might modulate EMT-related markers in the ATC cell lines. Western blot analysis revealed that mesenchymal markers, Vimentin, Snail, and Slug, were significantly upregulated, and the epithelial marker, E-cadherin, was downregulated in EpCAM-overexpressing MDA-T245 cells, as compared to the parental cell line. The opposite results (i.e., downregulation of mesenchymal markers and upregulation of the epithelial marker) were obtained when EpCAM was knocked out in MDA-T264 cells (Figure S3A).

Fig. 3.

EpCAM signaling activates EGFR, HGFR, and Wnt signaling pathway in ATC cells. (A) the levels of phosphorylated EGFR, HGFR, AKT and ERK were assayed by Western blotting in control-vector (ctrl) or EpCAM-overexpressing (OE-EpCAM) MDA-T245 cells, and wild type (WT) or EpCAM-knockout (KO) MDA-T264 cells. (B) indicated ATC cell lines were treated with EpEX (50 nM) for 30 min before phosphorylated EGFR, HGFR, AKT and ERK were assayed by Western blotting. (C) relative luciferase activity in wild type (WT) or EpCAM knockout (KO) MDA-T264 cells transfected with TOP-flash and FOP-flash vectors. The ratio of signals from these two reporters (each normalized to Renilla luciferase as an internal control) reflects β-catenin-specific canonical Wnt signaling. At 72 h, the TOP/FOP ratio was measured. TOP/FOP ratio was also measured in MDA-T264 cells treated with EpEX for 8 h. (D) Western blot analysis shows total and nuclear protein β-catenin levels in WT or KO MDA-T264 cells. Total and nuclear protein β-catenin levels in MDA-T264 cells treated for 8 h with 50 nM EpEX were assayed by Western blotting. (E) Western blot analysis shows the levels of phosphorylated LRP5 and LRP6 in WT or KO MDA-T264 cells. MDA-T264 cells were treated with EpEX (50 nM) for 30 min before levels of phosphorylated LRP5 and LRP6 were assayed by Western blotting. (F) MDA-T264 cells were treated with EpEX-Fc (50 nM) for the indicated times culture supernatants were pulled down with protein G to remove EpEX-Fc contamination. Then, the samples were pulled down with EpEX antibody (EpAb3-5) to isolate released EpEX. Antibody against human Fc was used to detect remaining EpEX-Fc contamination. EpEX-Fc was used as a positive control. (sup.: supernatant). The cell lysates were analyzed as described above. (G) MDA-T264 cells were treated with EpEX for 4 h, and EpICD nuclear translocation was assayed by immunofluorescence staining. (H) quantification of the proportion of cells with nuclear EpICD signal, normalized to the control value. (i) MDA-T264 cells were treated with EpEX for 1 h then TACE activity was measured with the SensoLyte 520 TACE (α–secretase) activity assay Kit (AnaSpec) and (j) γ-secretase activity was also measured. Statistically significant differences were identified by two‑tailed Student’s t-test. N = 3 independent experiments. All data are presented as mean ± sd. *p < 0.05; **p < 0.01

Since EpEX is known to act as a ligand for EGFR and HGFR [9, 12], we next investigated the effects of exogenous EpEX on EGFR and HGFR signaling. Treatment of MDA-T245 and MDA-T264 cells with EpEX induced phosphorylation of EGFR, HGFR, AKT and ERK (Fig. 3B). The expression of mesenchymal markers was also upregulated after treating the cells with EpEX (Figure S3B). To determine whether EpEX directly acts as an inducer of cell signaling, we examined whether EpEX could rescue the effects of EpICD blockade following DAPT treatment. As expected, DAPT inhibited EpICD nuclear translocation in MDA-T264 cells (Figure S4A) but had no effect on the phosphorylation of HGFR, EGFR, or their downstream effectors AKT and ERK. Meanwhile, EpEX induced phosphorylation of HGFR, EGFR, and downstream AKT/ERK in the absence of EpICD nuclear translocation (Figure S4B). EpEX also restored the DAPT-suppressed cell invasion ability (Figure S4C).

Interactions between Wnt ligands and their receptors serve to recruit the β-catenin destruction complex to the cell membrane. This recruitment leads to dephosphorylation of β-catenin at various activation sites, which allows the protein to translocate to the nucleus [36]. Interestingly, we found that EpCAM or EpEX increased Wnt signaling activity, as measured by the TOP/FOP ratio (Fig. 3C). Western blotting also showed EpCAM or EpEX promoted nuclear translocation of β-catenin and phosphorylation of LRP5 and LRP6 Wnt receptors (Fig. 3D and E). Previous studies have shown that DAPT blocks EpICD nuclear translocation and suppresses the transcriptional activity of Wnt receptors [15]. In the current study, we found that EpEX was able to restore Wnt activity in cells despite DAPT treatment (Figure S4D).

RIP of EpCAM is known to release EpEX, which may stimulate further RIP and cleavage of EpICD, a major transducer of EpCAM signaling [9, 37]. Thus, we next investigated whether exogenous EpEX treatment might modulate TACE and γ-secretase activities. As predicted, our experimental results indicated that EpEX increased the release of endogenous EpEX from EpCAM and promoted the cleavage of EpICD nuclear translocation (Fig. 3F–H). We also found that EpEX treatment increased the activities of TACE and γ-secretase in MDA-T264 cells (Fig. 3I and J).

EpCAM induces signaling receptor activator activity and TNF signaling pathway

We next wanted to determine the signaling pathways that mediate the effects of EpCAM, so we performed gene ontology (GO) pathway analysis on the RNAseq data from the EpCAM knockout and wild-type groups. We observed ‘signaling receptor activator activity pathway’ was among the top identified functional terms in the data from MDA-T264 cells (Fig. 4A). GSEA indicated that genes related to signaling receptor activator activity were enriched in EpCAM knockout cells (Fig. 4B). We also generated heat maps of gene expression changes for several of the genes under the functional term signaling receptor activator activity pathway in EpCAM knockout cells based on the results of the RNAseq analysis (Fig. 4C). KEGG analysis further showed that pathways associated with signaling receptor activity and metastasis were significantly downregulated in EpCAM knockout cells; these pathways included TNF signaling, focal adhesion, and PI3K-Akt signaling pathways (Fig. 4D).

Fig. 4.

EpCAM induces signaling receptor activator activity and TNF signaling pathway. MDA-T264 wild type (WT) and EpCAM knockout (KO) cells were subjected to RNA-seq analysis. (A) gene ontology (go) pathway analysis showed the most enriched pathways. (B) GSEA revealed enrichment of the ‘signaling receptor regulator activity’-related gene signature in KO cells. (C) Kyoto encyclopedia of genes and genomes (KEGG) pathway analysis of KO-downregulated genes in RNA-seq, p < 0.05. (D) heatmap of significant EpCAM knockout-regulated genes associated with signaling receptor regulator activity, p < 0.05. The levels of TNF-α, CSF2, CCL20, CXCL3, and CXCL5 (E) gene and (F) protein expression was assayed by qRT-PCR and Western blotting in wild type (WT) or EpCAM-knockout (KO) MDA-T264 cells. MDA-T264 cells were treated with EpAb2-6 (20 µg/ml) for 24 h. The levels of TNF-α, CSF2, CCL20, CXCL3 and CXCL5 (G) gene and (H) protein expression were assayed by qRT-PCR and Western blotting. MDA-T264 cells were treated with EpEX (50 nM) for 8 h. The levels of TNF-α, CSF2, CCL20, CXCL3, and CXCL5 (I) gene and (J) protein expression were assayed by qRT-PCR and Western blotting. (K) concentrations of released TNF-α in conditioned media of wild-type or EpCAM knockout MDA-T264 cells were assayed with a human TNF alpha ELISA Kit (ab181421). MDA-T264 cells were treated with indicated concentrations of (L) EpAb2-6 or (M) EpEX for 24 h. Concentrations of released TNF-α in conditioned media were assayed by capture ELISA. MDA-T264 cells treated with 20 ng/ml TNF-α for different times. (N) concentrations of released EpEX in conditioned media were assayed by capture ELISA. (O) MDA-T264 cells treated with 20 ng/ml TNF-α for 4 h, and EpICD nuclear translocation was assayed by immunofluorescence staining (scale bar: 20 μm). (P) proportion of cells with nuclear EpICD signal, normalized to the control value. (Q) TACE activity levels in cell lysates were assayed by SensoLyte 520 TACE (α–secretase) activity assay Kit (AnaSpec); (R) γ-secretase activity was also measured. Statistically significant differences were identified by two‑tailed Student’s t-test. N = 3 independent experiments. All data are presented as mean ± sd. *p < 0.05; **p < 0.01

Following these analyses, we asked how EpCAM affects the expression of signaling receptor activator activity-associated genes in ATC cells. To answer this question, we compared the data from EpCAM-knockout MDA-T264 cells with parental cells. The parental MDA-T264 cells expressed high levels of CSF2 and TNF-α at both mRNA and protein levels (Fig. 4E and F). In contrast, CSF2 and TNF-α mRNA and protein expression levels were downregulated by EpAb2-6 in MDA-T264 cells (Fig. 4G and H). Meanwhile, the mRNA and protein levels of CSF2 and TNF-α were increased upon EpEX treatment in MDA-T264 cells (Fig. 4I and J). Cancer cells often express TNF-α as a transmembrane protein. This protein can be enzymatically cleaved to produce a soluble active form (TNF-α) by the metalloprotease TNF-converting enzyme (TACE) [38]. Shedding of soluble TNF-α can allow its binding to TNF receptor 1 (TNFR1), and activated TNFR1 signaling results in the assembly of complex I, which promotes inflammation, host defense, cell survival and proliferation [39]. Thus, we next investigated whether EpCAM signaling might modulate the levels of released TNF-α in conditioned media, according to a TNF-α ELISA assay. The results indicated that EpCAM knockout or EpAb2-6 treatment significantly decreased the levels of released TNF-α (Fig. 4Kand L), while EpEX increased the levels of shed TNF-α without or with DAPT treatment (Fig. 4M; Figure S4E). We also found that shedding of EpEX (Fig. 4N) and nuclear accumulation of EpICD (Figs. 4O and 4P) were elevated after TNF-α treatment in MDA-T264 cells. In addition to this TNF-α crosstalk with EpCAM signaling, TNF-α strongly induced ERK and NF-κB p65 phosphorylation, consistent with MAPK-NF-κB pathway activation and tumorigenic outcomes in ATC cells (Figure S5).

Since EpCAM cleavage by active TACE releases soluble EpEX, the results suggest that TNF-α treatment induces TACE activity (Fig. 4Q). This observation was in line with our further finding that TNF-α treatment increased γ-secretase activity (Fig. 4R). These results suggest that soluble TNF-α may induce RIP and EpCAM cleavage.

To investigate whether EpCAM signaling and TNF activation may cooperatively regulate cancer cell invasion, we analyzed EpCAM knockout cells or EpEX-treated cells with or without TNF-α treatment. We found that TNF-α-induced cell invasion activity was reduced in EpCAM knockout cells (Figure S6A). Incubation of MDA-T264 cells with EpEX in combination with TNF-α cooperatively upregulated the levels of cell invasion as compared to EpEX or TNF-α treatments alone (Figure S6B).

EpCAM signaling promotes EMT and invasion by inducing TNF-mediated EGR1 expression

EGR1 is a member of the zinc-finger transcription factor family that is induced by TNF-α [40, 41]. We therefore asked whether EpCAM might regulate TNF-α mediated EGR1 expression. Subsequently, we tested how the presence of EpCAM signaling affected EGR1 gene expression level in ATC cells. Either knockout of EpCAM or EpAb2-6 treatment in MDA-T264 cells led to decreased EGR1 gene expression (Fig. 5Aand B). Conversely, EGR1 gene expression was increased upon EpEX or TNF-α treatment in MDA-T264 cells (Fig. 5Cand D). Furthermore, dual-luciferase reporter gene was constructed to detect the EGR1 promoter activity. The results showed that in the WT group, the relative luciferase activity was upregulated with TNF-α treatment but downregulated in the KO group (Fig. 5E). EpAb2-6 treatment in MDA-T264 cells led to decreased EGR1 promoter activity (Fig. 5F). Conversely, EGR1 promoter activity was increased upon EpEX treatment in MDA-T264 cells (Fig. 5G).

Fig. 5.

EpCAM signaling promotes EMT and invasion by inducing TNF-α-mediated EGR1 expression. (A) the level of EGR1 gene expression was assayed by qRT-PCR in wild type WT or KO MDA-T264 cells. (B) MDA-T264 cells were treated with EpAb2-6 (20 µg/ml) for 24 h, (C) EpEX (50 nM) or (D) TNF-α (1 nM) for 8 h. The level of EGR1 gene expression was assayed by qRT-PCR. (E) EGR1 promoter activity in WT or KO MDA-T264 cells after treated with or without TNF-α (1 nM). MDA-T264 cells were treated with (F) EpAb2-6 (20 µg/ml) or (G) EpEX (50 nM) then EGR1 promoter activity was assayed. (H) WT or KO MDA-T264 cells were treated with TNF-α (1 nM) for 2 h. The levels of EGR1 protein expression were assayed by Western blotting. (I) MDA-T264 cells were treated with the indicated concentrations of EpEX or TNF-α for 15 min, and the expression levels of erk and phosphorylated erk were analyzed by Western blotting. For the analysis of EGR1 protein expression, cells were treated for 2 h and subsequently analyzed by Western blotting. (J) MDA-T264 cells were treated with EpEX (50 nM) or/and TNF-α (1 nM) for 15 min, and the expression levels of erk and phosphorylated erk were analyzed by Western blotting. For the analysis of EGR1 protein expression, cells were treated for 2 h and subsequently analyzed by Western blotting. (K) EGR1 was determined by Western blotting in MDA-T264 cells pretreated with SPD304, U1026 or wortmannin for 1 h, followed by treatment with EpEX or TNF-α for 2 h. (L) correlation of mRNA levels for EGR1 and EpCAM (EPCAM), Snail (SNAI1), Slug (SNAI2), or Vimentin (VIM) in data from the TCGA database by GEPIA. (M) MDA-T264 cells after shLuc or shEGR1 treatment were treated with EpEX (50 nM) or TNF-α (50 nM) in 2% FBS medium for 24 h. EGR1 and EMT‑related protein (E‑cadherin, Vimentin, Snail, and Slug) expression levels were examined by Western blotting. (N) cell invasion activity was examined using a transwell invasion assay with matrigel. (o) quantification of cell invasion is shown. Statistically significant differences were identified by one-way ANOVA. N = 3 independent experiments. All data are presented as mean ± sd. *p < 0.05; **p < 0.01

To investigate whether EpCAM and TNF activation cooperatively regulate EGR1 expression, we next wanted to examine the protein expression levels of EGR1 in EpCAM knockout cells with or without TNF-α treatment. Western blotting results showed that the protein level of EGR1 was significantly decreased in TNF-α-treated EpCAM knockout cells compared to TNF-α-treated wild-type cells (Fig. 5H). To further investigate the effects of EpEX and TNF-α on phosphorylated ERK and EGR1 in ATC cells, we analyzed the levels of EGR1 and phosphorylated ERK protein expression. The Western blotting results showed that EpEX or TNF-α enhanced the levels of phosphorylated ERK and EGR1 protein in MDA-T264 cells (Fig. 5I). We also found that the combined treatment of MDA-T264 cells with EpEX and TNF-α cooperatively upregulated the levels of phosphorylated ERK and EGR1, as compared to EpEX or TNF-α treatments alone (Fig. 5J). EpEX restored the suppressed protein expression of TNFα and EGR1 in cells following DAPT treatment (Figure S4F), indicating that EpEX can activate TNFα-EGR1 signaling independent of EpICD nuclear translocation. We next investigated which TNF downstream signaling pathway might be involved in EpEX- and TNF-α-mediated upregulation of EGR1. Treatment with SPD304 abrogated both EpEX- and TNF-α-induced upregulation of EGR1 protein level. Importantly, treatment with the MEK inhibitor U0126 but not an AKT inhibitor, abolished EpEX-mediated upregulation of EGR1 (Fig. 5K). Since the EGR1 expression level is known to influence EMT signaling activity, we next analyzed datasets from the TCGA database to test whether a correlation exists between expression levels of EGR1 and common EMT genes in ATC, such as EPCAM, VIM and SNAI1. The data revealed significant positive correlations between EGR1 and EPCAM, VIM and SNAI1 (Fig. 5L). Furthermore, we found that the induction of EMT-related protein expressions and cell invasion by EpEX or TNF-α could be prevented by EGR1 knockdown (Fig. 5M–O). Taken together, these data suggest that EpCAM signaling plays an important role in regulating TNF-mediated EGR1 expression, thereby promoting EMT and invasion activity in ATC cells.

EpAb2-6 antibody blocks EpCAM cleavage and inhibits RIP activity

We next tested how an EpCAM-neutralizing antibody (EpAb2-6) and a BRAF inhibitor, dabrafenib, could affect EpEX release into conditioned medium of MDA-T264 cells. Intriguingly, we found that EpAb2-6 could inhibit EpEX release, while dabrafenib showed no effect on the level in conditioned medium (Fig. 6A). We also used another anti-EpEX antibody to detect the expression of full-length EpCAM in cell lysates by Western blotting and capture ELISA. The amount of detected EpCAM protein was increased in the lysates of cells treated with EpAb2-6 (Fig. 6Band C), but EpAb2-6 treatment had no effects on EPCAM gene expression (Fig. 6D). These data suggest that EpAb2-6 likely inhibited EpEX cleavage and release to the medium. We also found that EpAb2-6 treatment reduced the nuclear localization of EpICD in MDA-T264 cells, according to immunofluorescence staining (Fig. 6Eand F). Furthermore, blocking EpEX shedding with EpAb2-6 could inhibit the activities of TACE (Figure S7A; Fig. 6G) and γ-secretase (Figure S7B; Fig. 6H) in MDA-T264 cells. Meanwhile, dabrafenib had no effects on the activities of these enzymes. These results suggest that EpAb2-6 might inhibit EpEX shedding to suppress further RIP activity and EpCAM signaling.

Fig. 6.

EpAb2-6 antibody blocks EpCAM cleavage and inhibits RIP activity. Cells were treated with 20 μg/ml control IgG (NMIgG) or EpAb2-6 and BRAF inhibitor dabrafenib (1 μM) for 24 h. (a) EpEX was detected in conditioned medium by capture ELISA. (b) EpCAM was measured in cell lysates by Western blotting. (c) EpCAM was detected in cell lysate by capture ELISA. (d) EPCAM gene expression was assayed by qRT-PCR. (e) EpICD nuclear translocation was assayed by immunofluorescence staining. (f) quantification of nuclear EpICD (50 cells per group) is shown. (g) TACE and (h) γ-secretase activities were measured. Combination of dabrafenib and EpAb2-6 cooperatively induces apoptosis and inhibits invasion in ATC cells. (i) for combination treatments, cells were exposed to 20 μg/ml control IgG (NMIgG) or EpAb2-6 and the BRAF inhibitor dabrafenib (0.1 and 1 μM) as indicated for 24 h. Proportions of apoptotic cells were quantified by fluorescein annexin V-FITC/PI double labeling. (j) quantification of apoptosis is shown. (k) cell invasion activity was examined by a transwell invasion assay with matrigel. (l) quantification of invasion cells is shown. (m) the levels of phosphorylated EGFR, HGFR, AKT and ERK was assayed by Western blotting. (n) concentrations of shedding TNF-α in conditioned media were assayed by capture ELISA. (o) the expression levels of EGR1 and EMT regulators was assayed by Western blotting. The statistically significant differences were identified by one-way ANOVA. N = 3 independent experiments. All data are presented as mean ± sd. *p < 0.05; **p < 0.01; ***p < 0.001

Combination of dabrafenib and EpAb2-6 cooperatively induces apoptosis and inhibits invasion in ATC cells

Our data to this point suggested that EpCAM signaling promotes cell viability and invasion activity of ATC cells, so we wanted to further test the potential anti-tumor effects of simultaneously blocking EpCAM signaling and the BRAF pathway. The experimental results showed that EpAb2-6 alone induced apoptosis and inhibited invasion activity in MDA-T264 cells, as compared to control IgG treatment (Figure S8A and B). We also found that treatment with the humanized antibody, hEpAb2-6, could induce apoptosis (Figure S9A) and inhibit EMT-associated changes (Figure S9B) in a dose-dependent manner. The Wnt signaling activity assay showed that EpAb2-6 treatment significantly decreased the TOP/FOP ratio (Figure S10A) and the phosphorylation of LRP5 and LRP6 (Figure S10B).

Notably, we found that BRAF inhibitor dabrafenib enhanced the apoptotic effects of EpAb2-6 (Fig. 6I and J), and it also enhanced the inhibitory effects of EpAb2-6 on invasion activity in ATC cells (Fig. 6K and L). We next tested whether dabrafenib and EpAb2-6 might cooperatively regulate EGFR and HGFR signaling. Incubation of MDA-T264 cells with dabrafenib in combination with EpAb2-6 downregulated EGFR and HGFR downstream signaling components phosphorylation of AKT and ERK, as compared to dabrafenib or EpAb2-6 treatments alone. However, cooperative effects of dabrafenib and EpAb2-6 on phosphorylation of EGFR and HGFR were not observed (Fig. 6M). Furthermore, we found that dabrafenib and EpAb2-6 cooperatively enhanced the inhibitory effects of TNF-α cleavage in ATC cells (Fig. 6N). Western blotting also demonstrated that dabrafenib and EpAb2-6 cooperatively downregulated EGR1 and mesenchymal markers Snail and Slug, while upregulating the expression of epithelial marker E-cadherin in MDA-T264 cells (Fig. 6O). Together, these results indicate that dabrafenib and EpAb2-6 cooperatively suppress signaling pathways, driving cancer progression and invasion in ATC cells.

Combination of EpAb2-6 and dabrafenib cooperatively inhibits cancer stemness and Wnt activity in ATC cells

EpCAM is abundantly expressed in various cancer stem cells (CSC) [42], and several lines of evidence suggest that EpEX may be closely linked to EGFR activity [43]. Thus, we reasoned that EpEX might plausibly influence EGFR or hypoxia-induced signaling to promote ATC stemness. To explore this idea, we performed EpCAM knockout in an ATC cell line (MDA-T264 cells) and conducted tumor sphere formation assays. We found that spheroids formed by knockout cells lost tumorigenic potential and exhibited significantly smaller sizes and numbers than those formed by wild-type cells (Fig. 7A). We also performed an ELDA on the knockout and wild-type cells, finding that EpCAM knockout cells displayed significantly decreased cancer stemness potential compared to wild-type counterparts (Fig. 7B). Knockout of EpCAM in MDA-T264 cells also decreased the protein expression levels of stemness markers (Fig. 7C). The results further showed that treatment of MDA-T264 cells with EpEX augmented cancer stemness potential (Fig. 7D), and EpEX could rescue the suppression of tumor-sphere formation in cells by DAPT treatment (Figure S4G). Furthermore, we found that the effects of EpEX or TNF-α on tumor sphere formation could be prevented by EGR1 knockdown (Fig. 7E). These results support the idea that EpEX and TNF augment cancer stemness in ATC cells via effects on EGR1 expression.

Fig. 7.

Combination of EpAb2-6 and dabrafenib cooperatively inhibits cancer stemness and Wnt activity in ATC cells. (A) tumor sphere formation and cell counts for wild type (WT) or EpCAM knockout (KO) MDA-T264 cells. (B) ELDA showing stemness frequency in WT and EpCAM KO MDA-T264 cells. A log-fraction plot was fitted to the data shown in the table. (C) the levels of stemness marker protein expression were assayed by Western blotting. (D) ELDA showing stemness frequency for MDA-T264 cells treated with EpEX (50 μM) for 15 days (treated on alternate days). A log-fraction plot was fitted to the data shown in the table. (E) tumor-sphere formation and cell counts for MDA-T264 cells after shLuc or shEGR1 treatment were treated with EpEX (50 nM) or TNF-α (50 nM) for 7 days. (F) tumor sphere formation and cell counts for MDA-T264 cells treated with either NMIgG, dabrafenib (1 μM), and/or EpAb2-6 (20 µg/ml) for 7 days (treated on alternate days). (G) ELDA showing stemness frequency for MDA-T264 cells treated with either NMIgG, dabrafenib (1 μM), and/or EpAb2-6 (20 µg/ml) for 15 days (treated on alternate days). A log-fraction plot was fitted to the data shown in the table. The levels of (H) mRNA and (I) protein expression of stemness markers (SOX2, OCT4 and NANOG) were assayed by qRT-PCR and Western blot analysis. (J) MDA-T264 cells were treated with NMIgG, dabrafenib (1 μM) and/or EpAb2-6 (20 µg/ml) for 4 h. Phosphorylated LRP5 and LRP6 were assayed by Western blotting. MDA-T264 cells were treated with 20 μg/ml control IgG (NMIgG) or EpAb2-6 and dabrafenib (1 μM) for 24 h. (K) β-catenin nuclear translocation was assayed by immunofluorescence staining. (L) quantification of nuclear β-catenin (50 cells per group) from immunofluorescence staining is shown. Statistically significant differences were identified by one-way ANOVA. N = 3 independent experiments. All data are presented as mean ± sd. *p < 0.05; **p < 0.01; ***p < 0.001

We therefore sought to understand if the combinatorial treatment of EpAb2-6 and dabrafenib could induce similar results to those seen with EpCAM knockout cells in direct cancer stemness assays. Indeed, treatment with either dabrafenib or EpAb2-6 inhibited sphere formation, while the combination almost completely ablated the formation of tumor spheres (Fig. 7F). The ELDA results revealed that either dabrafenib or EpAb2-6 treatment could attenuate stemness in comparison to NMIgG treatment, while the combination almost completely abolished stemness potential in ATC cells (Fig. 7G). We also found incubation of MDA-T264 cells with dabrafenib and EpAb2-6 in the sphere formation assay led to cooperative downregulation of stemness markers, such as SOX2, OCT4 and NANOG at gene expression and protein levels, as compared to NMIgG-treated controls (Fig. 7H and I). Since the Wnt pathway is thought to play a critical role in CSC signaling [36], we probed markers of Wnt pathway activity by Western blotting. Our results showed that the combination treatment of dabrafenib and EpAb2-6 significantly inhibited the phosphorylation of Wnt receptors LRP5 and LRP6 in MDA-T264 cells (Fig. 7J). We also found that dabrafenib and EpAb2-6 combinational treatment reduced the nuclear localization of β-catenin in MDA-T264 cells, as assessed by immunofluorescence staining (Fig. 7K and L).

Combination therapy with humanized EpAb2-6 and dabrafenib inhibits metastasis and prolongs survival rate in ATC metastatic animal models

Proteolysis of EpCAM is known to produce EpEX, which may stimulate further RIP and release of the EpICD molecule to transduce EpCAM signaling [5]. To evaluate whether hEpAb2-6 inhibits RIP of EpCAM in vivo, xenograft tumor models were uitlized. We found hEpAb2-6 inhibits EpEX release (Fig. 8A), RIP (phosphorylation of ADAM17 and Presenilin 2) and EpICD cleavage (Fig. 8B), suggesting that the antibody inhibits RIP of EpCAM in vivo. IHC for EpEX and EpICD showed abundant membrane EpEX and EpICD staining after hEpAb2-6 treatment (Fig. 8C).

Fig. 8.

Combination therapy of EpAb2-6 and dabrafenib inhibits metastasis and prolongs survival in a metastasis animal model of ATC. MDA-T264 cells suspended in PBS were subcutaneously injected into the dorsolateral flank of 6-week-old female NOD/SCID mice. When tumors reached approximately 50–100 mm³, mice were randomized and treated with control IgG or hEpab2-6 (15 mg/kg) via tail vein injection twice weekly for two weeks. Tumors were then harvested for analysis. (A) soluble EpEX released into tumor interstitial fluid was quantified by capture ELISA. (B) RIP activity, indicated by phosphorylation of ADAM17 and presenilin 2, as well as EpICD cleavage, was assessed by Western blotting. (C) immunohistochemistry (IHC) revealed the distribution of membrane-bound EpEX and EpICD in tumor tissues. (D) experimental timeline to evaluate the effects of humanized EpAb2‑6 (hEpab2-6) (15 mg/kg) and/or dabrafenib (20 mg/kg) in the metastasis animal model. NOD/SCID mice were intravenously injected with 4 × 10⁵ MDA-T264 cells, followed by treatment with control IgG, hEpab2-6 and/or dabrafenib as indicated. (E) Representative pictures are shown, and (F) weights were taken of lungs harvested from the treated animals. (G) H&E stains show more lung colonization and larger tumor masses in the lungs of control IgG-treated mice compared to those receiving hEpab2-6 and/or dabrafenib (upper panel). Percentage (%) of metastatic tumor nodules per lung is shown (lower panel). (H) the percentage of initial body weight is shown over time for treated animals. (I) the survival curve, median survival and overall survival are shown with time represented in days. (J) summary illustration of the cell signaling events mediating EpCAM tumorigenic effects. In brief, excess cleavage of EpEX and nuclear translocation of EpICD support activation of EGFR, HGFR, and Wnt signaling, which promote cell proliferation and metastasis. In addition, the stimulation of EGFR and HGFR signaling increases TACE and γ-secretase activities, which enhance RIP of EpCAM to generate EpEX and EpICD. In addition, EpCAM signaling increases TNF-α expression and induces TNF-α cleavage, and clarify the crosstalk between EpCAM and TNF-α to regulate ERK-EGR1 axis signaling. The EpCAM neutralizing antibody EpAb2-6 can block EpCAM signaling, which comprises a positive feedback loop to promote EpCAM cleavage and subsequently increase EpEX and EpICD production. Statistically significant differences were identified by one-way ANOVA. For lung weight and tumor nodules: N = 4 independent experiments. All data are presented as mean ± sd. *p < 0.05; **p < 0.01; ***p < 0.001. For median survival: N = 6 independent experiments. All data are presented as mean ± sd. *p < 0.05; **p < 0.01

To test the effects of combined dabrafenib and humanized EpAb2-6 (hEpAb2-6) in vivo, we injected mice with MDA-T264 ATC cells to create a model of ATC metastasis. As illustrated in Fig. 8D, NOD/SCID mice were injected intravenously with MDA-T264 cells and then co-treated with dabrafenib and hEpAb2-6, or equivalent volumes of control IgG and vehicle, starting at 3 days after cell injection. After receiving the treatments for 30 days, the mice were euthanized, and lungs were collected to analyze metastatic events. We found that lung metastasis (Fig. 8E), tumor weight (Fig. 8F), and metastatic tumor nodules (Fig. 8G) in mice receiving hEpAb2-6 alone or the combination of hEpAb2-6 and dabrafenib were significantly decreased compared to the control IgG and dabrafenib-alone groups. Similarly, the animal body weight (Fig. 8H), overall survival rate and median survival (Fig. 8I) of animals receiving the combination therapy were increased compared to the control IgG-treated group. Based on these findings, we concluded that hEpAb2-6 can improve the anti-metastatic action of dabrafenib and prolong survival in the ATC metastasis mouse model in vivo.

Discussion

EpCAM is widely thought to promote cancer progression, as high levels of the protein are associated with high rates of recurrence, metastasis and drug resistance for many cancer types [44, 45]. In this study, we identified loss of membrane EpEX along with increased nuclear accumulation of EpICD as a common feature of ATC tumor tissues. Similar enhancement of EpICD nuclear translocation in tumor cells was previously found to be predictive poor prognosis in ATC patients [19, 46]. Our complementary findings that clinical ATC tissues showed loss of membrane EpEX with increased EpICD accumulation in nucleus of tumor cells (Fig. 1) suggest that EpICD may act as an oncogenic signal transducer to promote rapid tumor growth and contribute to poor prognosis. Because the antibodies used in our IHC specifically recognize either EpEX or EpICD, quantification of total EpCAM expression in matched ATC samples is not feasible. Instead, the loss of EpEX staining with concomitant nuclear EpICD accumulation likely reflects RIP, with EpICD serving as a more relevant surrogate of EpCAM activation status in ATC. These data showing expression and subcellular localization patterns of EpEX and EpICD suggest that the proteins have potential utility as biomarkers for predicting the clinical course and aggressiveness of thyroid cancer.

Importantly, our data suggest that EpEX treatment can restore the impaired activation of cell signaling cascades and cancer-associated phenotypes of cells deficient in EpCAM or EpICD nuclear translocation. These results indicate that EpEX itself can act as a functional inducer of tumorigenic signaling. Although the effects of EpEX are not solely dependent on EpICD nuclear translocation, additional mechanisms of paracrine or autocrine signaling may mediate the outcomes.

We have previously reported that EpEX induces EGFR and HGFR signaling, which subsequently activate TACE and γ-secretase to stimulate further shedding of EpEX and EpICD. As such, soluble EpEX-mediated activation of EGFR and HGFR stimulates a positive feedback loop in which RIP of EpCAM is enhanced to increase further release of active EpCAM domains [9, 11, 12]. We report here that either EpCAM overexpression or treatment with exogenous EpEX can induce EGFR and HGFR signaling in ATC cells as well as proliferation and invasion activity, suggesting the signal may be amplified in either an autocrine or paracrine manner (Figs. 3 and 4. Our group has produced an EpCAM-neutralizing antibody called EpAb2-6, which can block the function of EpEX. EpAb2-6 treatment is known to disrupt EpCAM cleavage and subsequently attenuate EpEX and EpICD production [9, 25]. In this study we found EpAb2-6 can inhibit EpEX shedding in order to inhibit RIP activity and suppress EpCAM signaling (Fig. 6A–H).

Metastasis-associated genes include not only drivers of the metastatic phenotype, but also genes involved in secondary events, adaptive mechanisms, and counter-balancing changes that occur in the cell [47]. Thus, we asked whether EpCAM might play a decisive role in metastatic diffusion by examining genes that are concordantly dysregulated across metastasis events for multiple types of cancer. EMT is broadly associated with cancer progression and metastasis [48], and the process involves a complex series of reversible events that lead to the loss of epithelial cell adhesion and the induction of a mesenchymal phenotype in cancer cells. Ultimately, EMT affords tumor cells with stem cell-like characteristics and mesenchymal features, which allow the tumor cells to disseminate and become more invasive [49]. In many different cancer types, EMT also promotes resistance to various therapeutic drugs [50, 51]. Blocking EMT for therapeutic purposes may be accomplished by targeting essential components of the tumor microenvironment that contribute to the activation of the EMT program in cancer cells [52]. In this study, we further demonstrated that EpCAM signaling plays a critical role in promoting EMT and cell invasion activity in ATC cells (Fig. 2E and G; Figure S3A and B).

Accumulating evidence also indicates that EpCAM actively promotes EMT and invasion in other contexts [12, 13, 53]. Proteolytic cleavage of EpCAM generates the soluble extracellular fragment EpEX and the intracellular domain EpICD, both of which have been shown to activate transcriptional programs that enhance EMT and metastasis [5, 13]. Because EpCAM cleavage, rather than transcriptional downregulation, reduces surface EpCAM detection, diminished membrane staining may coexist with increased soluble extracellular fragment EpEX- and the intracellular domain EpICD-mediated signaling activity that drives EMT progression [9, 12, 13, 15]. Consistent with these studies, our data show that EpCAM signaling enhances EMT marker expression and invasive behavior in ATC cells. Notably, different effects of EpCAM may be due to different cells of origin for different cancers, heterogeneity in EpCAM regulation, and the distinct biological roles of EpCAM. In this study, the soluble EpEX that is shed after RIP can act as a ligand for both EGFR and HGFR to induce AKT and ERK1/2 signaling, and the EpICD nuclear translocation can induce Wnt-β-catenin signaling and EMT. Together, our findings support a model in which EpCAM signaling pathways promote EMT and invasion in ATC.

The EGFR pathway is known to be associated with progression and metastasis in thyroid cancer [54]. Moreover, EGFR activation is considered the most likely mechanism to cause onset of resistance to BRAF inhibitors [55]. The inhibition of EGFR signaling acts synergistically with BRAF inhibition, as resistance to BRAF inhibitors can be caused by feedback activation of EGFR [56]. Previous reports have shown that soluble EpEX promotes EpCAM signaling. Furthermore, the stimulation of ERK signaling via an EpEX-EGFR axis can increase TACE activity to enhance EpEX cleavage in colon and lung cancer cells [9–11]. In this study, we show that EpEX can activate EGFR and HGFR signaling pathways in ATC cells (Fig. 3A and B). Nuclear translocation of β-catenin is a hallmark of Wnt signaling. This event is stimulated when Wnt ligands that have been activated via palmitoylation by porcupine bind to Wnt receptors, such as LRP5/6 [57]. A previous report suggested that Wnt receptors are upregulated via the transcriptional co‑factor activity of EpICD [15], and another demonstrated that Wnt and β-catenin signaling is a therapeutic target in ATC [58]. Our data further show that EpCAM signaling could increase nuclear translocation of β-catenin and induce LRP5 and LRP6 Wnt receptor activity (Fig. 3C–E).

TNF-α is a pivotal cytokine in the tumor microenvironment, exhibiting multifunctional roles that can be either pro-tumorigenic or anti-tumorigenic. The full-length protein is primarily expressed on the cell membrane, where it can be cleaved by TACE, resulting in the release of its soluble form. Following shedding, TNF-α binds to TNF receptor 1 (TNFR1), forming the TNF-TNFR1 complex I, which activates downstream signaling pathways, including NF-κB and MAPK. This activation subsequently regulates key cellular processes such as inflammation, proliferation, survival, metastasis, and various other signaling cascades [41, 59]. TNF-α also promotes invasive growth through the upregulation of HGF and HGFR expression [40]. Our results showed that EpCAM signaling increases TNF-α expression and induces TNF-α cleavage to promote TNF signaling (Fig. 4A–M).

TNF-α may induce RIP and EpCAM and pro-tumor properties depending on the cellular context [60]. While it was originally described as a factor capable of inducing apoptosis in tumor cells, accumulating evidence indicates that TNF-α frequently promotes tumor progression through activation of pro-survival and pro-inflammatory pathways. TNF can form cytosolic Complex II (RIPK1-FADD-caspase-8), resulting in either apoptosis or necroptosis [61]. In our ATC models, TNF-α treatment robustly induced ERK phosphorylation, consistent with the activation of MAPK signaling and a tumorigenic outcome. Although TNF-induced Complex II formation may contribute to anti-tumor effects in other systems, our findings suggest that TNF-α preferentially promotes survival and cancer progression signaling in ATC cells.

EGR1 is a member of the EGR family, which can be activated by cytokines, growth factors, or stress conditions. Additionally, EGR1 belongs to the zinc-finger transcription factor family and regulates the expression of downstream target genes, either promoting or inhibiting their expression. Previous studies have demonstrated that EGR1 enhances tumor metastasis and chemokine expression in various cancer cell types [62, 63], and its transcription factor activity can be induced by TNF-α [40, 41]. In line with these findings, our data showed crosstalk between EpCAM and TNF-α in terms of EGR1 expression. Moreover, EpEX induced EMT and invasion phenotypes via EGR1 in ATC cells (Fig. 5J–L).

ATC is a rare and highly aggressive malignancy with poor prognosis. The disease is typically characterized by a highly invasive phenotype, metastasis and rapid local progression. Current treatments mainly involve surgery, radiation therapy and chemotherapy. In recent years, targeted therapies have been utilized for ATC treatment, including a combination of dabrafenib and trametinib for patients with BRAF-mutated disease [64]. Moreover, immune checkpoint inhibitors can be used alone or in combination with targeted therapies [65, 66]. Despite the introduction of these new therapeutics, the median survival time for ATC patients is still less than one year, often owing to the development of drug resistance and tumor recurrence [64, 67]. Thus, a more precise understanding of the molecular mechanisms underlying ATC tumor progression will be crucial for development of new strategies to improve patient outcomes.

Dabrafenib is used to treat patients with BRAF-mutated cancers. The molecule selectively targets BRAF kinase to interfere with mitogen-activated protein kinase (MAPK) signaling pathway, which enhances cancer cell proliferation and survival [20]. Dabrafenib was approved by the U.S. Food and Drug Administration (FDA) for use in combination with the MEK inhibitor trametinib to treat BRAF-mutated ATC [68, 69]. Notably, BRAF mutations are found in 10–50% of ATC cases and may be associated with poor prognosis [70]. Additionally, alterations to late-event MAPK pathway signaling, such as loss of p53 (50–80% of ATC cases), are common in ATC and have been implicated in tumor cell dedifferentiation [71]. We have previously demonstrated that our EpCAM-neutralizing antibody, EpAb2-6, can bind to EpEX and directly induce cancer cell apoptosis in vitro and tumor growth in vivo [12, 25]. In this study, we found that either EpAb2-6 or dabrafenib treatment alone could partially induce apoptosis and inhibit invasion-related functions in ATC cells, whereas the combination treatment had enhanced inhibitory effects (Fig. 6I–L). Meanwhile, EpAb2-6 inhibited phosphorylation of EGFR and HGFR, as well as downstream ERK and AKT signaling. Importantly, the combination of EpAb2-6 with dabrafenib further inhibited EGFR and HGFR signaling in ATC cells (Fig. 6M).

CSCs are closely associated with cancer initiation and key malignant features, including tumor progression, metastasis, and immune evasion [72]. Furthermore, disease relapse is often attributable to CSCs that have developed drug resistance, potentially via multiple mechanisms [36]. It is well established that Wnt signaling plays a crucial role in the formation of CSCs by promoting cellular characteristics that contribute to tumor plasticity and drug resistance. [57]. Therefore, targeting the CSC population may represent a promising therapeutic strategy. Since EpCAM is known to be abundantly expressed in CSCs, it is plausible that EpCAM signaling may influence EGFR or hypoxia-induced signaling [9, 13]. In line with the idea that the Wnt/β-catenin pathway plays a critical role in CSC signaling [36], our data showed that either EpAb2-6 or dabrafenib treatment alone could significantly inhibit cancer stemness potential and Wnt activity in ATC cells, whereas the combination treatment had enhanced inhibitory effects (Fig. 7F–L).

We have previously shown that our EpAb2-6 exhibits remarkable therapeutic efficacy in orthotopic and metastatic animal models for a variety of cancer types. In addition, EpCAM promotes PD-L1 stabilization, and the combination of our antibody with Atezolizumab, a PD-L1 inhibitor, showed synergistic effects exhibiting enhanced therapeutic efficacy. Our results from the metastatic ATC animal model also indicated that EpAb2-6 might significantly inhibit metastasis and prolong animal survival. Notably, such prolongation of survival was further pronounced in the hEpAb2-6 and dabrafenib combination treatment group (Fig. 8D–I).

In this study, we demonstrate the potential therapeutic efficacy of an EpCAM-neutralizing antibody, EpAb2-6, in cell and animal models of ATC. We previously reported that in colorectal and lung cancers, EpCAM actively regulates critical cell signaling pathways, including EGFR and HGFR to promote tumor progression. These activities could be inhibited by EpAb2-6 treatment, suggesting that the antibody may have therapeutic potential for multiple cancer types [9, 11, 12, 25]. Here, we further showed that EpCAM signaling is prominently activated in tissue arrays from ATC patients as well as in ATC cell lines. The stimulation of EpCAM signaling increases TACE and γ-secretase activities, which enhance RIP of EpCAM to generate EpEX and EpICD. Excess cleavage of EpEX and nuclear translocation of EpICD support activation of EGFR, HGFR, and Wnt signaling, which promote cell proliferation, metastasis, and stemness in ATC cells. In addition, EpCAM signaling increases TNF-α expression and induces TNF-α cleavage, and clarify the crosstalk between EpCAM and TNF-α to regulate ERK-EGR1 axis signaling. EpAb2-6 can block EpCAM signaling, which comprises a positive feedback loop to promote EpCAM cleavage and subsequently increase EpEX and EpICD production (Fig. 8J). Importantly, EpAb2-6 treatment either alone or in combination with a small molecule BRAF inhibitor, dabrafenib, could successfully attenuate EpCAM-related signaling in ATC cell lines. The treatment also induced apoptosis and inhibited invasion of ATC cells, which is indicative of a promising strategy for ATC therapy. In further support of this potential therapeutic effect, EpAb2-6 exhibited efficacy in animal models of metastatic ATC, and this efficacy was amplified when the EpCAM-neutralizing antibody was used in combination with dabrafenib. In light of the fact that ATC patients currently have poor prognosis with limited therapeutic options, further studies are warranted to determine if EpAb2-6 might be a useful treatment for this disease. Additionally, the mechanistic insights gained from our study may be helpful in the development of novel anticancer therapeutics for better treatment of ATC patients.

Conclusion

This study uncovers the molecular mechanisms underlying promotion of ATC tumor progression by EpCAM signaling via the ERK-EGR1-TNF-α axis. EpEX induces TNF-α activation. The subsequent shedding of TNF-α may trigger EpCAM cleavage via the ERK-RIP axis, thereby initiating a positive feedback loop. Furthermore, EpEX regulates ERK-mediated EGR1 expression, leading to the induction of EMT. In light of the fact that ATC patients currently have poor prognosis with limited therapeutic options, further studies are warranted to determine if blockage of EpCAM signaling might be a useful treatment for this disease. Additionally, the mechanistic insights gained from our study may be helpful in the development of novel anticancer therapeutics for better treatment of ATC patients.

supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

ADAM17

A disintegrin and metalloprotease 17

ATC

Anaplastic thyroid carcinoma

Ctrl

Control

CSC

Cancer stem cell

CYT

Cytosol

EGFR

Endothelial growth factor receptor

EGR1

Early growth response 1

ELDA

Extreme limiting dilution assay

ELISA

Enzyme-linked immunosorbent assay

EMT

Epithelial-mesenchymal transition

EpCAM

Epithelial cell adhesion molecule

EpEX

Extracellular domain of EpCAM

EpICD

Intra-cellular domain of EpCAM

ERK

Extracellular signal-regulated kinase

FTC

Follicular thyroid carcinoma

HGF

Hepatocyte growth factor

KO

Knockout

MAPK

Mitogen activated protein kinase

MEM

Membrane

NUC

Nuclear

OE

Overexpression

PTC

Papillary thyroid carcinoma

qRT-PCR

Quantitative real-time polymerase chain reaction

RIP

Regulated intramembrane proteolysis

TACE

tumor necrosis factor-alpha converting enzyme

TNF-α

Tumor necrosis factor-alpha

WT

Wild type

Author contributions

C.C.L. and H.C.W. wrote original manuscript. C.C.L., Y.J.S., S.S.P., C.Y.C., C.H.W., P.C. L., C.Y., S.Y.L. H.C.W. reviewed and edited the manuscript. C.C.L., Y.J.S., S.S.P. C.Y.C., C.H.W., T.M.L., Y.L.H., R.L.H., S.H.H. performed the experiments and interpreted the data. H.C.W. performed conceptualization, investigation, funding acquisition, and supervision.

Funding

This research was supported by Academia Sinica (NSTC 112-2311-B-001-018) and the Ministry of Science and Technology (MOST) (109-2311-B-001-007-MY3) granted to Han-Chung Wu.

Data availability

All data generated or analyzed during this study are available within the article and supplementary information files.

Declarations

Ethics approval and consent to participate

The animal study was approved by the Committee on the Ethics of Animal Experiments of Academia Sinica.

Consent for publication

Not applicable.

Competing interests

Related to this work, the Institute of Cellular and Organismic Biology, Academia Sinica, has filed a patent application and transferred the technology to a biotech company; Han-Chung Wu and Chi-Chiu Lee are named as inventors. The other authors declare no competing interests.