NTRK fusion promotes tumor migration and invasion through epithelial–mesenchymal transition and closely interacts with ECM1 and NOVA1 Authors: Siqing Zeng, Ke Jiang, Jie Ge, Mimi Tang, Yuqi Wen, Xiaoting Ma, Heli Liu, Xingui Xiong Published: 2024 ABSTRACT BackgroundThe NTRK fusion gene is a rare cancer driver and a typical representative "diamond mutation". Its unique role in tumor progression is highly important for the clinical diagnosis and treatment of patients with tumors. We searched for NTRK fusion-positive patients in our hospital. As of August 2022, a total of 8 patients were affected. We discovered that NTRK fusion was associated with enhanced tumor invasion and migration ability. Previous reports also support this finding, but its underlying mechanism has not been elucidated.MethodsWe undertook a comprehensive exploration of the correlations between NTRK fusions and tumor invasion as well as migration by analysing clinical data, performing bioinformatics analysis via public databases, and conducting in vitro cell experiments.ResultsWe ascertained that within the thyroid cancer (THCA) dataset and the pancancer dataset, ECM1 and NOVA1 were coexpressed with NTRKs. Additionally, they demonstrated a significant association with the activity of the epithelial‒mesenchymal transition (EMT) pathway. Furthermore, these genes are overexpressed in various cancers and are associated with advanced clinical stage and increased aggressiveness. Our in vitro study revealed that larolutinib potentially inhibited the invasion and metastasis ability of NTRK-fused cells. Interestingly, contrary to previous findings, the repression of ECM1 increased the migration and invasion ability of NTRK-fused tumor cells.ConclusionsNTRK fusion tumors present heightened migratory and invasive potential in clinical settings. Further experiments confirmed the significant inhibitory effects of TRK inhibitors on the migration and invasion abilities of these cells. There is a complex relationship between ECM1, NOVA1 and NTRK fusion; however, further research is needed to determine whether NTRK fusion promotes tumor metastasis through these two genes.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12885-024-13271-w. Introduction The NTRK gene family comprises NTRK1, NTRK2, and NTRK3, which encode the proteins TRKA, TRKB, and TRKC, respectively. These TRK proteins possess a tyrosine kinase domain that becomes activated upon ligand binding and subsequent receptor homodimerization. This activation leads to the phosphorylation of receptors and the initiation of downstream signalling pathways crucial for neuronal development and differentiation [1, 2]. TRK proteins may be activated by a variety of mechanisms, such as NTRK mutations [3, 4], splicing variants [5, 6], and TRK overexpression [7–9], which have been found in various tumor types. Fusions involving NTRK1, NTRK2, or NTRK3 are the most common mechanisms of carcinogenic TRK activation [10]. NTRK gene fusion occurs via chromosomal inversion, deletion, or translocation events, resulting in the juxtaposition of the 3' sequence containing the kinase domain of NTRK1, NTRK2, or NTRK3 with the 5' sequence of a distinct gene. Consequently, this fusion product manifests as a chimeric oncoprotein characterized by ligand-independent constitutive activation of TRK kinase [11]. Moreover, NTRK gene fusion is an oncogenic driver in pansolid tumors and has been found in several adult and childhood solid tumors [12]. The fusion of NTRK genes is an independent oncogenic driver that is mutually exclusive with other driver mutations. Additionally, while the fusion of NTRK1 and NTRK3 is commonly observed, NTRK1 fusions are rare. Therefore, NTRK fusions represent potential therapeutic targets for various tumor types, and several inhibitors targeting NTRK, such as larotrectinib and entrectinib, have been approved and shown promising results in clinical trials [13, 14]. Although promising progress has been made in the diagnosis and treatment of NTRK fusion-positive tumors [2, 15, 16], the specific mechanism and role of NTRK fusion in cancer progression have not been fully elucidated. From October 2012 to August 2022, we identified eight patients with NTRK fusion-positive cancer of five different types in our hospital. Our clinical data revealed that seven patients had evident invasive and metastatic capabilities at the time of initial diagnosis. Therefore, we questioned whether this invasiveness and metastasis were due to NTRK fusion or simply coincidental NTRK fusion. A subsequent literature analysis of NTRK fusion cancers revealed an association between NTRK fusions and enhanced tumor invasion and migration [17–20]. NTRK fusion-positive tumors exhibit pronounced invasiveness in terms of clinical, clinicopathological, and molecular characteristics, as well as in cellular and animal models [21, 22]. However, the underlying mechanisms remain unclear, prompting further investigation into the regulation of cancer invasion and metastasis by NTRK fusion. Owing to limited clinical data, we accessed public databases such as TCGA and cBioPortal for additional information. We identified a total of 32 patients harboring NTRK fusions across diverse cancer types. Among them, 12 were diagnosed with thyroid cancer (THCA), 4 with glioma, 3 with colon cancer, and no more than 3 with any other individual cancer type. Considering the paucity of available data, we opted to employ THCA data as an initial framework for investigating the correlation between NTRK fusion and tumor invasiveness. Finally, we identified two genes, namely, extracellular matrix protein 1 (ECM1) and neuro-oncological ventral antigen 1 (NOVA1), which potentially contribute to the promotion of tumor invasion mediated by NTRK fusion events. ECM1 is a glycoprotein involved in many biological processes, is highly expressed in a variety of tumor tissues, and is characteristic of a series of epithelial malignancies, including THCA [23–25]. It can promote the invasion and migration of cancer cells by inducing epithelial–mesenchymal transition (EMT) and promoting angiogenesis [26–29]. NOVA1 has traditionally been studied as an important neuronal-specific splicing factor, and it has been shown to play a role in cancer progression [30, 31]. Similarly, NOVA1 has been shown to promote tumor migration and invasion through notch-, WNT-, and EMT-related pathways [30, 32]. Both ECM1 and NOVA1 are involved in the regulation of EMT to promote tumor metastasis. Although many studies have demonstrated the role of these two genes in promoting cancer progression and invasion across diverse malignancies, it remains unclear whether they contribute to the highly aggressive and metastatic capacity of NTRK fusion tumors through the induction of EMT. In this study, we observed the upregulation of ECM1 and NOVA1 in the NTRK fusion THCA and further explored the role of ECM1 and NOVA1 in pancarcinoma. Patients and methods Patients and samples From October 2012 to August 2022, we identified a total of 8 NTRK fusion-positive cases involving 5 types of cancer, including thyroid cancer (THCA), ovarian cancer, colon cancer, colon neuroendocrine tumor, and lung cancer, through the available genetic test reports. We analysed the admission records, medical examination results, diagnoses and treatments, surgeries, outpatient records and other medical records of these 8 patients. Datasets and data collection The cBioPortal [33] online program (http://www.cbioportal.org/) is an open access resource for the interactive exploration of multidimensional cancer genome datasets. Our NTRK fusion-positive sample information was obtained from the Tumors with TRK fusions (MSK, Clin Cancer Res 2020) dataset in the cBioPortal, and the relevant clinical and sequencing data were downloaded from The Cancer Genome Atlas (TCGA) Data Portal (https://tcga-data.nci.nih.gov/tcga/). Moreover, cBioPortal was used to evaluate the coexpressed genes (CEGs) of NTRK1 and NTRK3. The CEGs of NTRKs derived from thyroid carcinoma (TCGA, Firehose Legacy) were used to determine the intersection with differentially expressed genes (DEGs), and the CEGs of NTRKs obtained from the pancancer analysis of whole genomes (ICGC/TCGA, Nature 2020) were used for enrichment analyses. Differential gene expression analysis The raw data were processed by R (4.1.3) to establish the gene expression matrices. Differential gene expression analysis was performed via the Shiny web application for RStudio, limma online tool (http://127.0.0.1:7651/). We chose genes that fulfilled the conditions of having a count per million (CPM) value greater than 2 in at least half of the samples, accompanied by a false discovery rate (FDR)-adjusted p value less than 0.05 and a fold change (FC) greater than 1.5, as differential genes. Furthermore, we visualized these DEGs by applying the criteria of an adjusted p value less than 0.05 and a log2FC greater than 1. Functional enrichment analysis The “clusterProfiler” package was used for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis. The results were subsequently visualized via a bioinformatics website (http://www.bioinformatics.com.cn/). Survival analysis The prognostic values associated with the expression of ECM1 and NOVA1 in patients were retrieved from the Gene Expression Profiling Interactive Analysis 2 (GEPIA2) database (http://gepia2.cancer-pku.cn). Antibodies and chemicals TrkC rabbit polyclonal antibody (1:500), ECM1 rabbit polyclonal antibody (1:500), E-cadherin rabbit polyclonal antibody (1:1000), N-cadherin rabbit polyclonal antibody (1:1000), and vimentin rabbit polyclonal antibody (1:1000) were used. All the antibodies used were purchased from Proteintech, China. ECM1 shRNA plasmid (HonorGene, China), NOVA1 shRNA plasmid (HonorGene), Hunan, China, pcDNA3.4-ECM1-3xFlag (HonorGene), and pcDNA3.1-NOVA1-3xFlag (HonorGene) were used. NOVA1 rabbit polyclonal antibody (1:1000, Abcam, UK) and larotrectinib (MedChemExpress, China) were used. The shRNA sequences are shown in Supplemental Table 1. Table 1Clinical Information of 8 NTRK fusion-positive Patients in Our HospitalCaseCancer TypegenderagestageNTRK fusion typevital_status(As of August 22, 2022)1Ovarian CarcinomaFemale60IVBHNF4A-NTRK3 & NTRK1-DCST2Alive2Thyroid Papillary CarcinomaFemale54IETV6-NTRK3Alive3Colorectal Poorly Differentiated AdenocarcinomaFemale71IVBTPM3-NTRK1Alive4Non-small Cell Lung CarcinomaMale59IVNTRK1-FLAD1Alive5Thyroid Papillary CarcinomaFemale39IETV6-NTRK3Alive6Colorectal Neuroendocrine AdenocarcinomaFemale58IITPR-NTRK1Alive7Thyroid Papillary CarcinomaFemale27IIETV6-NTRK3Alive8Thyroid Papillary CarcinomaFemale37IITPR-NTRK1Alive Table 1Clinical Information of 8 NTRK fusion-positive Patients in Our HospitalCaseCancer TypegenderagestageNTRK fusion typevital_status(As of August 22, 2022)1Ovarian CarcinomaFemale60IVBHNF4A-NTRK3 & NTRK1-DCST2Alive2Thyroid Papillary CarcinomaFemale54IETV6-NTRK3Alive3Colorectal Poorly Differentiated AdenocarcinomaFemale71IVBTPM3-NTRK1Alive4Non-small Cell Lung CarcinomaMale59IVNTRK1-FLAD1Alive5Thyroid Papillary CarcinomaFemale39IETV6-NTRK3Alive6Colorectal Neuroendocrine AdenocarcinomaFemale58IITPR-NTRK1Alive7Thyroid Papillary CarcinomaFemale27IIETV6-NTRK3Alive8Thyroid Papillary CarcinomaFemale37IITPR-NTRK1Alive Clinical Information of 8 NTRK fusion-positive Patients in Our Hospital Cell culture The THCA cell line IHH4 (passage 12) was purchased from Abiowell (Abiowell Biotechnology Co., Ltd, China). IHH4 and IHH4 fusion cells were cultivated in 1640 medium (Gibco, China) supplemented with 10% fetal bovine serum and 1% penicillin‒streptomycin solution. The culture process took place in an incubator at 37 °C with saturated humidity and 5% CO2. Transfection The cells were added to 6-well plates (2 × 105/well) and grown to 40–60% confluence, at which time LipoFiter (HANBIO, China) was used to transfect the cells with the appropriate siRNAs (GenePharma, China), followed by incubation for 6 h in serum-free media, after which the media was exchanged for media containing 10% FBS. After an additional incubation period of 24 h, quantitative polymerase chain reaction (qPCR) was employed to determine the efficiency of ECM1 knockdown. At 48 h posttransfection, Western blotting and cell functional assays were performed. shRNA NC sense: 5’-TTCTCCGAACGTGTCACGT-3’, antisense: 5’-ACGTGACACGTTCGGAGAA-3’; ECM1-sh4 sense: 5’-GGGTGCCCACATTGGACAATA-3’, antisense: 5’-TATTGTCCAATGTGGGCACCC-3’; ECM1-sh5 sense: 5’-GGGAGGATACCCTTGACAAAT-3’, antisense: 5’-ATTTGTCAAGGGTATCCTCCC-3’; ECM1-sh6 sense: 5’-GAGTTCTCACCAAGCATAAAC-3’, antisense: 5’-GTTTATGCTTGGTGAGAACTC-3’. Construction of IHH4 cells with ETV6-NTRK3 fusion First, we constructed the carrier pPB[Exp]-EF1A > EGFP-CAG > ETV6-NTRK3-PGK > Puro and extracted a significant amount of transfected low-endotoxin nucleic acid. We subsequently transfected the constructed vector into IHH4 cells via the Neon™ transfection system (Cat# MPK5000). After 48 h of culture, the transfected target cells were screened with the corresponding drug according to the protocol. Upon the completion of the drug treatment cycle, the medium containing the drug was replaced with fresh medium to enable cellular recovery after a 24–48 h culture period. Finally, stable transmissible cell lines were expanded in media supplemented with drugs and verified via q‒PCR before being stored in liquid nitrogen for subsequent experiments. qRT‒PCR The Tsingzol RNA Extraction Reagent (TIANGEN Biotech, China) was utilized to extract RNA from cells and tissue samples, and cDNA was generated with Oligo (dT) Primer (Tsingke Biotechnology Co., Ltd., China) and RevertAid Reverse Transcriptase (Thermo Fisher Scientific, China). For all qRT‒PCR analyses, PerfectStart Green qPCR SuperMix (TransGen Biotech, China) was utilized on the ABI PRISM 7700 sequence detection system (Applied Biosystems, China). The 2-ΔΔCt approach was employed for the assessment of relative gene expression, and the analyses were conducted in triplicate. The sequences of the primers used were as follows (5’ → 3’): ECM1-forward: GGACAGAGTCAAGTGCAGCC, ECM1-reverse: ATGGCTCTGGTCCCCAAATG; NOVA1-forward: GGCCAAGACAGAACCAGTCA, NOVA1-reverse: GGACTTGGTGGTAGTTGGGG; NTRK3-forward: ATGGAGCTCTACACCGGACT, NTRK3-reverse: GGTGAGCCGGTTACTTGACA; ETV6-forward: CAGTTGCTTTCTGACAGCCG, ETV6-reverse: TTTGTAGTAGTGGCGCAGGG; GAPDH-forward: AGGTCGGAGTCAACGGATTT, GAPDH-reverse: TGACGGTGCCATGGAATTTG. Western blotting For the isolation of proteins from cell and tissue samples, RIPA buffer (Beyotime Biotechnology, China) enriched with 1% PMSF (Thermo, USA) was used. The protein concentration was subsequently determined via the BCA method, and the protein concentration was quantified via a BCA Kit (Beyotime Biotechnology). Equal amounts of protein were separated on a 10% SDS‒PAGE gel and transferred to PVDF membranes (Millipore, USA). These membranes were blocked for 2 h at room temperature by being fully immersed in a 5% skim milk solution. Thereafter, the membranes were incubated overnight at 4 °C with the primary antibody diluted in TBST containing blocking reagent from SeraCare KPL, USA. Next, the blot was incubated with HRP-conjugated anti-rabbit or anti-mouse IgG from SeraCare KPL, USA, at room temperature for 60 min. Finally, the protein bands were detected via a chemiluminescence imaging system and analysed for grayscale analysis via ImageJ software. Transwell assays Cell migration and invasion were measured via the Transwell assay. For the invasion assay, Matrigel (Corning, China) was used to cover a polycarbonate membrane within a transwell chamber (Corning) that contained 24,8.0 mm pores. Thereafter, 4 × 104 cells in serum-free medium were transferred to the upper compartment of the plate. Moreover, 500 μL of medium supplemented with 10% fetal bovine serum was added to the lower compartment of the plate. The cells were subsequently incubated at 37 °C for 48 h. Thereafter, the noninvasive cells on the upper surface of the membrane were eliminated by gentle scrubbing. The migrating or invading cells on the lower surface of the membrane were fixed with 4% paraformaldehyde (PFA) and subsequently stained with 0.1% crystal violet (Beyotime Biotechnology) for 10 min. An optical microscope (magnification × 100) was used to enumerate the cells, thereby determining the number of migrated or invaded cells in each experimental group. Statistical analyses All the data are presented as the means ± SDs. Statistical tests were performed via SPSS statistical software. Differences between groups were compared via t tests. P values ≤ 0.05 were considered to indicate statistical significance. Results The clinical presentation of patients harboring NTRK fusion suggests a potential association between NTRK fusion and enhanced invasive and migratory capacity The comprehensive clinical details of the eight patients with positive NTRK fusion at our institution are presented in Table 1. Notably, with the exception of Patient 2, who exhibited no apparent invasion of adjacent tissues or lymph node involvement, all the other patients manifested varying degrees of local invasion, lymph node involvement, or distant metastasis, thereby demonstrating a high level of invasiveness and metastatic potential. Notably, Patient 1 presented with multiple metastatic sites at the time of initial diagnosis. Consequently, our objective was to investigate the associations between NTRK fusion and these aggressive characteristics. To address this question comprehensively, we extensively reviewed the literature on NTRK fusion cancers and reported a significant correlation between NTRK fusions and increased tumor invasiveness and migratory ability [17–20]. Pathology images of seven representative patients are provided in Supplementary Fig. 1. However, since patient 3 had their pathology examination conducted at an alternative medical establishment, there was a scarcity of accessible imaging data. We subsequently acquired the clinical (Supplementary Table 1) and genomic data of NTRK fusion patients from the TCGA Pancancer Atlas for further analysis. Owing to the incomplete clinicopathological characteristics of these NTRK fusion-positive patients (n = 30), only 22 patients with available data were included. The baseline staging characteristics of the 493 patients without NTRK fusion are summarized in Table 2. The association between NTRK fusion and clinical stage was examined via cBioPortal (Fig. 1A-B), revealing a significant correlation between NTRK fusion and advanced disease stage. Table 2Staging characteristics of patients with and without NTRK fusionCharacteristicsTHCAFusionP valuen49322T.stage, n (%) < 0.001T2160 (31.1%)6 (1.2%)T1a19 (3.7%)0 (0%)T3167 (32.4%)9 (1.7%)T1b78 (15.1%)2 (0.4%)T144 (8.5%)1 (0.2%)T49 (1.7%)0 (0%)T4a14 (2.7%)0 (0%)TX2 (0.4%)1 (0.2%)T2b0 (0%)2 (0.4%)T4b0 (0%)1 (0.2%)N.stage, n (%) < 0.001N0225 (43.7%)9 (1.7%)N1b71 (13.8%)4 (0.8%)N1a92 (17.9%)0 (0%)NX50 (9.7%)0 (0%)N155 (10.7%)7 (1.4%)N2c0 (0%)1 (0.2%)N20 (0%)1 (0.2%)M.stage, n (%)0.029M0274 (56%)10 (48%)MX209 (42%)9 (43%)M19 (1.8%)1 (4.8%)M1b0 (0%)1 (4.8%)Pathologic.stage, n (%) < 0.001Stage I276 (53.9%)9 (1.8%)Stage III112 (21.9%)0 (0%)Stage II51 (10%)2 (0.4%)Stage IVA44 (8.6%)4 (0.8%)Stage IVC6 (1.2%)0 (0%)Stage IV2 (0.4%)3 (0.6%)Stage IIB0 (0%)2 (0.4%)Stage IIA0 (0%)1 (0.2%)Fig. 1Clinical relevance of NTRK fusion (A, B) Statistical plot of clinical staging in the dataset titled Tumors with TRK fusions in cBioPortal (MSK, Clin Cancer Res 2020) Table 2Staging characteristics of patients with and without NTRK fusionCharacteristicsTHCAFusionP valuen49322T.stage, n (%) < 0.001T2160 (31.1%)6 (1.2%)T1a19 (3.7%)0 (0%)T3167 (32.4%)9 (1.7%)T1b78 (15.1%)2 (0.4%)T144 (8.5%)1 (0.2%)T49 (1.7%)0 (0%)T4a14 (2.7%)0 (0%)TX2 (0.4%)1 (0.2%)T2b0 (0%)2 (0.4%)T4b0 (0%)1 (0.2%)N.stage, n (%) < 0.001N0225 (43.7%)9 (1.7%)N1b71 (13.8%)4 (0.8%)N1a92 (17.9%)0 (0%)NX50 (9.7%)0 (0%)N155 (10.7%)7 (1.4%)N2c0 (0%)1 (0.2%)N20 (0%)1 (0.2%)M.stage, n (%)0.029M0274 (56%)10 (48%)MX209 (42%)9 (43%)M19 (1.8%)1 (4.8%)M1b0 (0%)1 (4.8%)Pathologic.stage, n (%) < 0.001Stage I276 (53.9%)9 (1.8%)Stage III112 (21.9%)0 (0%)Stage II51 (10%)2 (0.4%)Stage IVA44 (8.6%)4 (0.8%)Stage IVC6 (1.2%)0 (0%)Stage IV2 (0.4%)3 (0.6%)Stage IIB0 (0%)2 (0.4%)Stage IIA0 (0%)1 (0.2%) Staging characteristics of patients with and without NTRK fusion Clinical relevance of NTRK fusion (A, B) Statistical plot of clinical staging in the dataset titled Tumors with TRK fusions in cBioPortal (MSK, Clin Cancer Res 2020) Functional annotation of NTRK and its coexpressed genes In light of the occurrence of NTRK gene fusions in various cancers, genes coexpressed with NTRK1 and NTRK3 were obtained from cBioPortal. The top 150 genes from these coexpressed gene sets, along with NTRK1 and NTRK3, were subsequently selected for Kyoto Encyclopedia of Genes and Genomes (KEGG) and gene ontology (GO) enrichment analyses. (Fig. 2). Our enrichment analysis revealed that ntrk and its associated coexpressed genes significantly contribute to invasion-related pathways, including cell‒cell adhesion, extracellular matrix structural constituents conferring tensile strength, extracellular matrix structural constituents, and fibronectin binding. These findings strongly support our hypothesis, suggesting the active engagement of NTRK and its associated coexpressed genes in the processes of tumor invasion, migration, and metastasis.Fig. 2Functional annotation of NTRK and NTRK-coexpressed genes. A Top 10 CC, BP, and MF GO enrichment terms. B The top 10 enriched KEGG pathway terms. C Emapplot showing the relationships between enriched pathways Functional annotation of NTRK and NTRK-coexpressed genes. A Top 10 CC, BP, and MF GO enrichment terms. B The top 10 enriched KEGG pathway terms. C Emapplot showing the relationships between enriched pathways Screening of DEGs related to NTRK fusions Given that the thyroid cancer (THCA) dataset is relatively extensive, a comprehensive dataset was retrieved from the TCGA database to further examine the role of NTRK fusion proteins in tumorigenesis and development. In accordance with the clinical information of the samples, the data were subsequently categorized into three distinct groups: the NTRK fusion-positive group, the THCA group, and the normal group. Comparative analyses were conducted to discern disparities between the NTRK fusion-positive group and the other two groups. Figure 3A shows 175 DEGs in the NTRK fusion-positive group vs. the THCA group, and Fig. 3B shows 1622 DEGs in the NTRK fusion-positive group vs. the normal group. Moreover, we ranked the coexpressed genes (CEGs) of NTRK1 and NTRK3 in thyroid carcinoma (TCGA, Firehose Legacy) obtained from cBioPortal according to q values. The top 30 CEGs for each gene were subsequently selected for the analysis of genes intersecting with the other two groups of DEGs (Supplementary Table 2 and Fig. 3C). The expression of 10 intersecting genes—ECM1, NOVA1, ADAMTS9, TMEM108, VSTM4, RMST, PRR29, PNMA8B, SSPP2 and CAPN3—all of which are upregulated CEGs of NTRK3 that exhibit significant positive correlations and coexpression patterns with these genes among normal controls and fusion patients (Fig. 3D-F).Fig. 3Differentially expressed genes and their correlation and coexpression with NTRKs were screened. A. DEGs in the NTRK fusion group vs. the THCA group. B DEGs in the NTRK fusion group vs the normal group. C Venn diagram of the intersection of the CEGs of NTRK1 and NTRK3 with the previous two sets of DEGs. Heatmaps of the (D) correlation and (E–F) coexpression of NTRK1 and NTRK3 with 10 intersecting genes (*P < 0.05) Differentially expressed genes and their correlation and coexpression with NTRKs were screened. A. DEGs in the NTRK fusion group vs. the THCA group. B DEGs in the NTRK fusion group vs the normal group. C Venn diagram of the intersection of the CEGs of NTRK1 and NTRK3 with the previous two sets of DEGs. Heatmaps of the (D) correlation and (E–F) coexpression of NTRK1 and NTRK3 with 10 intersecting genes (*P < 0.05) NTRK fusion may promote the invasion and migration of THCA cells through the EMT pathway induced by ECM1 and NOVA1 To further clarify the connections between these ten differentially expressed genes (DEGs) and the invasion and metastasis of thyroid carcinoma (THCA), an exhaustive literature survey of these genes was performed. ECM1 and NOVA1 are extensively involved in the process of tumor metastasis and invasion [26, 28, 30, 32, 34] and are significantly coexpressed with NTRK1 and NTRK3 (Fig. 4A). Consequently, we selected ECM1 and NOVA1 for further investigation of their roles in THCA pathogenesis. In the TCGA-THCA cohort, ECM1 was significantly correlated with NTRK1 and NTRK3. In contrast, NOVA1 was significantly correlated solely with NTRK3 (Fig. 4B). In addition, these genes were significantly associated with the EMT pathway (Fig. 4F) and key EMT-related genes (Fig. 4G), which supported our hypothesis that ECM1 and NOVA induce EMT progression in THCA cells. An analysis of the associations between these genes and clinical stage was subsequently conducted on the basis of gene expression levels and gene set variant analysis (GSVA) scores via the GSCA website. The findings revealed a significant correlation between these two genes and advanced-stage THCA (Fig. 4C-E).Fig. 4ECM1 and NOVA1 induce EMT to promote invasion and migration in NTRK fusion-positive patients. A Heatmaps showing the associations of NTRK3 coexpressed with NTRK1, ECM1, and NOVA1 in the normal and fusion groups. B Scatter plots of the correlations of ECM1 and NOVA1 with NTRK3 and NTRK1 in the TCGA-THCA dataset. C-E The relationships among ECM1, NOVA1, and thyroid cancer (THCA) disease clinical stages were analysed on the basis of gene expression levels and gene set variation analysis (GSVA) scores on the GSCA website. F The GSCA website was utilized to generate a heatmap illustrating the correlation between ECM1 and NOVA1 and tumor functional states in THCA. (G) Heatmap of the correlation between ECM1 and NOVA1 and the emt marker gene. (*P < 0.05) ECM1 and NOVA1 induce EMT to promote invasion and migration in NTRK fusion-positive patients. A Heatmaps showing the associations of NTRK3 coexpressed with NTRK1, ECM1, and NOVA1 in the normal and fusion groups. B Scatter plots of the correlations of ECM1 and NOVA1 with NTRK3 and NTRK1 in the TCGA-THCA dataset. C-E The relationships among ECM1, NOVA1, and thyroid cancer (THCA) disease clinical stages were analysed on the basis of gene expression levels and gene set variation analysis (GSVA) scores on the GSCA website. F The GSCA website was utilized to generate a heatmap illustrating the correlation between ECM1 and NOVA1 and tumor functional states in THCA. (G) Heatmap of the correlation between ECM1 and NOVA1 and the emt marker gene. (*P < 0.05) NTRK gene fusion promotes tumor invasion and migration through the EMT pathway induced by ECM1 and NOVA1, which may also occur in other solid cancers The presence of NTRK gene fusions in various solid tumors has been established. To gain insights into the potential roles of ECM1 and NOVA1 across cancers, we conducted further analysis. Our comprehensive pancancer analysis of whole genomes (ICGC/TCGA, Nature 2020) revealed a comparable tendency for coexpression among NTRK1, NTRK3, ECM1, and NOVA1 (Fig. 5A). Furthermore, a positive correlation was detected between the expression levels of ECM1, NOVA1, and several EMT marker genes, namely, CDH2 and FN1 (Fig. 5B, C). To further investigate this association, we selected 33 cancers from the GSCA database to conduct correlation analyses between the expression of ECM1 and NOVA1 and pathway activity. Our results indicated that both genes were significantly correlated with EMT in multiple cancer types (Fig. 5D). Additionally, the CancerSEA cohort data for ECM1 associations in thirteen included cancers were examined on their website. We discovered strong associations between ECM1 expression and the functional status associated with EMT, angiogenesis, invasion and metastasis (Fig. 5E). Therefore, we postulate that NTRK fusion might facilitate tumor invasion and migration by inducing the activation of the EMT pathway through ECM1 and NOVA1, irrespective of tumor histology.Fig. 5Coexpression of ECM1 with NOVA1 and NTRKs in pancarcinoma and its correlation with EMT. A Scatterplot of the coexpression of NTRK1 and NTRK3 with ECM1 and NOVA1. B The correlation between the expression of ECM1 and the expression of key genes involved in NOVA and EMT. C Scatter plots of the expression of ECM1-, NOVA- and EMT-related genes. D GSVA score and activity of cancer-related pathways associated with ECM1 and NOVA across cancers. E Relationship between ECM1 expression and tumor phenotype in pancarcinoma. (*P ≤ 0.05; #FDR ≤ 0.05) Coexpression of ECM1 with NOVA1 and NTRKs in pancarcinoma and its correlation with EMT. A Scatterplot of the coexpression of NTRK1 and NTRK3 with ECM1 and NOVA1. B The correlation between the expression of ECM1 and the expression of key genes involved in NOVA and EMT. C Scatter plots of the expression of ECM1-, NOVA- and EMT-related genes. D GSVA score and activity of cancer-related pathways associated with ECM1 and NOVA across cancers. E Relationship between ECM1 expression and tumor phenotype in pancarcinoma. (*P ≤ 0.05; #FDR ≤ 0.05) ECM1 and NOVA1 across cancers We subsequently analysed the potential roles of ECM1 and NOVA1 across various cancer types via multiple publicly available databases. First, we assessed the mRNA expression levels of ECM1 and NOVA1 in 33 different cancer types (Fig. 6A, B) and used GEPIA 2 to analyse cancers without corresponding normal tissues (Supplementary Fig. 2, Fig. 2A). Survival analysis via the GEPIA 2 database revealed that the overall survival (OS) of patients in the high-expression group was significantly greater than that of patients in the low-expression group (Fig. 6C). The combined survival analysis of all the cancer data is shown in Fig. 6D. In addition, advanced clinical stage and aggressiveness were associated with the expression of ECM1 and NOVA1 in diverse types of cancer (Fig. 6E-P, Supplementary Fig. 2, Fig. 2B-D).Fig. 6ECM1 and NOVA1 across cancers. (A, B) Pancancer analysis of ECM1 and NOVA1 expression. Survival analysis results for ECM1 and NOVA1. (C) Survival map of the relationship between high expression of two genes and survival in multiple cancers and (D) Kaplan‒Meier curve. (E-P) Relationships among ECM1, NOVA1, and cancer clinical information. (*P < 0.05; **P < 0.01; ***P < 0.001) ECM1 and NOVA1 across cancers. (A, B) Pancancer analysis of ECM1 and NOVA1 expression. Survival analysis results for ECM1 and NOVA1. (C) Survival map of the relationship between high expression of two genes and survival in multiple cancers and (D) Kaplan‒Meier curve. (E-P) Relationships among ECM1, NOVA1, and cancer clinical information. (*P < 0.05; **P < 0.01; ***P < 0.001) Expression of NTRK3, ECM1, and NOVA1 To validate these findings, we constructed a stable IHH4 cell line with the ETV6-NTRK3 fusion gene. The cells were subsequently divided into three groups according to various treatment modalities: IHH4-fusion (We constructed IHH4 cells with the ETV6-NTRK3 fusion), IHH4 (Untreated IHH4 cells), and IHH4-F-L (IHH4 cells with ETV6-NTRK3 fusion were treated with 1000 nM Larotrectinib for 24 h). The mRNA levels of NTRK3, ECM1, and NOVA1 were measured via qRT‒PCR. In comparison with that in the IHH4 group, NTRK3 expression was upregulated in the IHH4-fusion group, whereas ECM1 expression was downregulated and NOVA1 expression remained unaltered. In contrast, compared with that in the IHH4 fusion group, NTRK3 expression was downregulated in the IHH4-F-L group, whereas ECM1 and NOVA1 expression was upregulated. In contrast, treatment with larotrectinib significantly increased the mRNA expression of ECM1 and NOVA1 in ETV6-NTRK3 fusion IHH4 cells (Fig. 7A). The discrepancy between the experimental results and predictions may be attributed to the differences between the human environment and the cellular environment. However, this finding also implies a complex and intimate relationship between the NTRK fusion protein and these two genes.Fig. 7In vitro cell validation experiments. A The mRNA expression levels of ECM1, NOVA1, and TRKC were detected via qPCR. B-D ECM1, NOVA1, TRKC, N-cadherin, vimentin, and E-cadherin expression levels in cells were assessed via Western blotting. E, F Cell migratory and invasive activities were evaluated through Transwell assays. The data are presented as the means ± SDs (* P < 0.05; **P < 0.01; ***P < 0.001) In vitro cell validation experiments. A The mRNA expression levels of ECM1, NOVA1, and TRKC were detected via qPCR. B-D ECM1, NOVA1, TRKC, N-cadherin, vimentin, and E-cadherin expression levels in cells were assessed via Western blotting. E, F Cell migratory and invasive activities were evaluated through Transwell assays. The data are presented as the means ± SDs (* P < 0.05; **P < 0.01; ***P < 0.001) We successfully downregulated the expression of ECM1 in NTRK3-ETV6IHH4 cells by employing three ecm1-specific short hairpin RNA (shRNA) constructs, as validated by qPCR analysis (Supplementary Fig. 2, Fig. 2E). However, none of the nine shNOVA1 plasmids that we procured demonstrated a significant knockdown effect on the NOVA1 gene upon qPCR detection. As a result, we only carried out experiments related to ECM1. Western blot analysis revealed no significant differences in the expression levels of NTRC, ECM1, NOVA1, vimentin, TRKC, or E-cadherin. Compared with that in the IHH4-fusion group, E-cadherin expression was significantly upregulated in the IHH4-F-L group (Fig. 7B-D), suggesting that larotrectinib could modulate EMT induction through E-cadherin. Although the mRNA expression of ECM1 and NOVA1 was significantly increased in the cells of the IHH4-F-L group, their protein expression did not consistently increase. Compared with that in the negative control group, the expression of vimentin slightly decreased, whereas the expression of E-cadherin slightly increased; however, these changes were not statistically significant (adj. 0.2994 for vimentin and adj. 0.1960 for E-cadherin). The role of NTRK fusion in cell migration and invasionin vitro To further explore the potential relationship between NTRK gene fusions and the migratory capability of thyroid carcinoma (THCA) cells, we performed Transwell assays to assess their impact on the migration and invasion of IHH4 cells. Our results revealed no significant differences in migration or invasion between the IHH4-fusion group and the IHH4 group. However, treatment with larotrectinib attenuated the migration and invasion of the IHH4-fusion group in comparison to those of the untreated IHH4-fusion group. Furthermore, when the IHH4-fusion + shNC group was compared with the IHH4-fusion + sh-ECM1 group, increased cell migration ability and invasion capability were noted, although the proportion of upregulated cells was comparatively lower (Fig. 7E-F). In general, compared with IHH4 cells, in vitro constructed ETV6-NTRK3 fusion IHH4 cells did not exhibit enhanced migratory or invasive capabilities. However, treatment with larotrectinib significantly decreased the migration and invasion ability of ETV6-NTRK3 fusion IHH4 cells. Similarly, contrary to our initial hypothesis, the downregulation of ECM1 expression in ETV6-NTRK3 fusion IHH4 cells resulted in increased cell migration and invasion compared with those in si-NC cells. Discussion The NTRK gene fusion represents a potent oncogenic driver that is observed across various tumor entities, albeit rarely occurring. In light of their gene-specific properties, targeted inhibitors, such as larotrectinib and entitinib, have been extensively researched and put into practice [14, 35–38]. Moreover, TRK inhibitors display extraordinary efficacy in the treatment of patients who possess NTRK fusion proteins across diverse cancer types, offering distinct superiority over other gene fusion targets. Current data indicate that larotrectinib has the potential to evoke prompt and sustained responses in patients with solid tumors harboring NTRK gene fusions, resulting in a substantial prolongation of patient survival, a favourable safety profile, and the possibility of long-term administration. Despite extensive investigations on the diagnosis and management of NTRK fusions, the underlying mechanisms that drive these events in cancer remain poorly understood. Although NTRK fusion rarely occurs in tumors (typically less than 1%), NTRK gene fusion frequently occurs (greater than 75%) in some rare cancers, such as infantile fibrosarcoma [39, 40] and secretory breast cancer [41]. Consequently, investigating the underlying mechanism of NTRK fusion in oncogenesis is highly important to achieve potential breakthroughs in tumor diagnosis and therapeutic interventions. We conducted an analysis of the relevant cases identified in the hospital and reported that the overwhelming majority (7/8) of patients exhibited local invasion and distant metastasis upon initial diagnosis. Numerous previous studies have also shown that NTRK fusion can induce aberrant activation of TRK, thereby facilitating tumor proliferation and dissemination, which is associated with enhanced migratory and invasive properties of tumors [42, 43]. However, owing to its low incidence, NTRK fusion is currently not advocated as a dependable biomarker for early-stage cancer patients. This paradoxical circumstance may cause patients to be deprived of the most suitable treatment at the initial stage, consequently resulting in disease progression. In the future, with increased detection rates or improved accuracy of NTRK fusion detection methods, this contradiction could be resolved. To further examine the associations between NTRK fusion proteins and tumor invasion and migration, we performed a comprehensive bioinformatics analysis utilizing a publicly accessible database. Numerous studies have consistently demonstrated that ECM1 and NOVA1 are strongly correlated with tumor invasion and migration, thereby serving as independent prognostic risk factors for certain types of tumors [44–47]. These two genes may play a role in promoting tumor invasion and migration through NTRK fusion. Moreover, our bioinformatic analysis demonstrated that the expression of ECM1 and NOVA1 was correlated with increased clinical classification and enhanced aggressiveness across diverse cancers. Furthermore, their expression significantly activated processes related to the EMT pathway. To validate our findings, we conducted in vitro cell experiments. Interestingly, although IHH4 cells with the ntrk fusion did not exhibit significantly greater migration or invasion than did normal IHH4 cells with the NTRK fusion, treatment with the TRK inhibitor larotrectinib notably decreased the migration and invasion abilities of IHH4 cells harboring the ETV6–NTRK3 fusion protein. This finding suggested that TRK is indeed involved in THCA cell migration and invasion [44, 48, 49]. Regarding ECM1, the migration and invasion capabilities of the IHH4 cells with ETV6-NTRK3 fusion increased after inhibiting the expression of ECM1. This might be attributed to the fact that the process of constructing the fused IHH4 cells had an influence on the cells and failed to simulate the in vivo situation more precisely. Or, there might be certain interactions between ECM1 and NOVA1, which could not be explored for the intrinsic connection in this experiment. In conclusion, we discovered that ECM1 is indeed associated with the migration and invasion of NTRK fusion cells. However, the underlying mechanism is far more complex than our speculation, and more efforts are required for further exploration. Moreover, there are several limitations to this study. Considering the efficiency and success rate of in vitro NTRK fusion cell models and the associated financial implications, compared with existing methods of promoting the integration of donor templates by inducing double-strand breaks (DSBs) at participating sites [50] or by cloning appropriate gRNA pairs into plasmids expressing CRISPR-associated nuclease 9, chromosomal rearrangement can be induced in the relevant cellular environment [22] to obtain a cellular model of NTRK fusion. We constructed an overexpression vector of the fusion gene and introduced it into cells by means of electroporation for resistance screening, thereby yielding a stable cell line. This approach may have given precedence to results while neglecting potential alterations that take place during the fusion production process. Such potential alterations could contribute to disparities between in vitro experimental findings and bioinformatic analysis results. Moreover, establishing an appropriate cellular context to model cancer-relevant aberrations is crucial. Owing to the constraints of experimental resources and the time limitations associated with the construction of gene-fused cells, we decided to employ a single cell line for our research. However, this approach has the drawback of preventing comparisons between different cell lines, which could provide more comprehensive data and draw more conclusive conclusions. Furthermore, due to the scarcity of patients with identified NTRK fusions, the medical records we collected pertained to patients who had been hospitalized previously and discharged. The acquisition of clinical tissue samples for them was challenging. Thus, we were unable to validate the expression patterns of the relevant genes in the tissue samples of clinical patients and compare them with those of non-NTRK fusion patients. With the increasing popularity of genetic examinations, we will consciously collect the biological tissues of some special cases in advance, hoping to remedy this deficiency in the future. An important aspect of this study is that we performed bioinformatic analyses on both individual cases of THCA and multiple cancer types, resulting in consistent findings. Therefore, similar to how NTRK gene fusion acts as a strong oncogenic driver in different tumor types, it is reasonable to speculate that there could be a common underlying mechanism among all solid tumors with NTRK fusions. However, confirming this hypothesis may present difficulties, and new conclusions may arise from future research on the mechanisms underlying NTRK fusion. In fact, the effect of NTRK fusion on tumors is complex and multifaceted, and additional mechanisms need to be discovered and verified. For example, our enrichment analysis revealed that the CEG of ntrk is also closely related to immune-related biological processes, so the fusion of NTRK may also change the immune state of tumors in some way, thus affecting the occurrence and development of tumors. In addition, NOVA1 is a member of the NOVA family of alternative splicing factors [51, 52]. NOVA activity leads to up to 700 alternative splicing events per cell [52], and NOVA1 may be involved in the fusion of NTRKs with gene partners as an alternative splicing factor. In summary, numerous aspects of NTRK fusion require further exploration and investigation. Supplementary Information Supplementary Material 1: Supplementary Figure 1. Pathology images of 7 patients in our hospital (Patient 3 received a pathological diagnosis at another hospital, so no images of her were available).Supplementary Material 2: Supplementary Figure 2. ECM1 and NOVA1 expression and its relationship with clinical stage. (A) ECM1 and NOVA 1 expression levels in pancancer samples from the TCGA database were analysed via GEPIA 2. The relationships among ECM1, NOVA1 and clinical stage were analysed via GSCA. (B) Expression tendencies according to pathologic stage. (C) Trends in GSVA scores among stages. (D) GSVA scores according to cancer stage. (D) qPCR technology was used to verify the results of plasmid knockdown. (*P<0.01).Supplementary Material 3.Supplementary Material 4.Supplementary Material 5.Supplementary Material 6. Supplementary Material 7.Supplementary Material 8.Supplementary Material 9. Supplementary Material 1: Supplementary Figure 1. Pathology images of 7 patients in our hospital (Patient 3 received a pathological diagnosis at another hospital, so no images of her were available). Supplementary Material 2: Supplementary Figure 2. ECM1 and NOVA1 expression and its relationship with clinical stage. (A) ECM1 and NOVA 1 expression levels in pancancer samples from the TCGA database were analysed via GEPIA 2. The relationships among ECM1, NOVA1 and clinical stage were analysed via GSCA. (B) Expression tendencies according to pathologic stage. (C) Trends in GSVA scores among stages. (D) GSVA scores according to cancer stage. (D) qPCR technology was used to verify the results of plasmid knockdown. (*P<0.01). Supplementary Material 3. Supplementary Material 4. Supplementary Material 5. Supplementary Material 6. Supplementary Material 7. Supplementary Material 8. Supplementary Material 9.