Functional Analysis of MET Exon 14 Skipping Alteration in Cancer Invasion and Metastatic Dissemination

Lung cancer is a highly heterogeneous disease associated with a spectrum of recurrent genetic and epigenetic aberrations (1). To date, numerous highly actionable genetic abnormalities have been identified in a group of receptor tyrosine kinases (RTK), including EGFR, anaplastic lymphoma kinase (ALK), mesenchymal–epithelial transition kinase (MET), which play driving roles in lung cancer initiation and development (2–5). MET is a well-characterized proto-oncogene that was originally identified in a chemically transformed osteosarcoma cell line (6). Following binding with its ligand, hepatocyte growth factor (HGF), MET undergoes dimerization and phosphorylation, leading to conformational changes culminating in activation of the tyrosine kinase and downstream signaling cascades, including those involved in proliferation, migration, and survival (5, 7). In addition to its role to maintain tissue homeostasis under normal physiological conditions, gene mutations, amplification or protein overexpression of MET can lead to aberrant activation of downstream signaling pathways that promote tumorigenesis (5, 7).

MET splice site alterations that result in exon 14 skipping (hereon referred to as METΔ14Ex) occur predominantly in non–small cell lung cancer (NSCLC) and represent a novel class of actionable oncogenic events with significant clinical impact and therapeutic perspectives (8–13). Exon 14 of MET encodes a juxtamembrane domain encompassing the Y1003 residue that serves as the key binding site for the E3 ubiquitin ligase, c-Cbl (8, 11, 12). The phosphorylation of Y1003 is a prerequisite for the recruitment of c-Cbl that subsequently facilitates MET receptor ubiquitination and degradation (8, 11, 12, 14, 15). Unlike the majority of splice site mutations that lead to loss of protein function, METΔ14Ex is postulated to increase MET protein stability and thereby ligand-dependent oncogenic activity by impairing the initiation of the ubiquitination process of MET. METΔ14Ex alterations occur in about 2%–4% of cases of NSCLC, with higher frequency in a unique, highly aggressive subtype called pulmonary sarcomatoid carcinoma (PSC; refs. 10, 13). In most cases, METΔ14Ex is mutually exclusive with other recognized oncogenic drivers. In NSCLC patients with METΔ14Ex alterations, frequent and durable responses to a variety of MET inhibitors have now been reported through initial case/cohort reports and more recently through biomarker-focused clinical studies (8, 10, 12, 13).

In the current study, we sought to dissect the intrinsic tumorigenic role of METΔ14Ex in NSCLC. We found that METΔ14Ex increased receptor stability and expression on cell surfaces, and prolonged activation of downstream signaling through impairment of receptor endocytic degradation, resulting in significantly enhanced cell scattering, migration and invasion capacity in vitro as well as metastasis in vivo. Global transcriptomic (RNA sequencing) analyses revealed that a large cluster of cell movement-related biological processes were preferentially activated in HGF-treated METΔ14Ex models (vs. MET) in a time-dependent manner. Activation of the PI3K/Akt/Rac1 pathway–mediated cytoskeleton dynamics and induction of cell-matrix disassembly were involved in the regulation of METΔ14Ex-stimulated cell movement and metastasis. Finally, we demonstrated that blockade of HGF/METΔ14Ex signaling using MET inhibitors dramatically repressed cell invasion and metastasis, highlighting potential effective therapeutic options for NSCLC patients with METΔ14Ex alterations.

Antibodies and reagents used in this study are listed in Supplementary Table S1.

H292, H125, A549, H596, and HEK293T cell lines were purchased from the ATCC. H292, H125, A549, and H596 cells were maintained in RPMI-1640 medium supplemented with 10% FBS and 1× antibiotic/antimycotic (Thermo Fisher Scientific). HEK293T was maintained in DMEM medium supplemented with 10% FBS and 1× antibiotic/antimycotic. All experiments were performed within 3–8 passages after thawing the cells. Cells were authenticated by STR profiling and routinely tested for Mycoplasma contamination.

Two gRNAs separately targeting MET exon 13 and 15 were designed and cotransfected into cells to generate METΔ14Ex cells. gRNAs were designed using the CRISPR Design Tool (http://crispr.mit.edu/) to minimize off-target effects. The following gRNAs were used in this study: gRNA1: 5′-CTTGTTAAAGACGGCTATCA-3′ and gRNA2: 5′-ACCCACTGAGGTATATGTAT-3′. Oligos were annealed and cloned into the BbsI site of plasmids pSpCas9(BB)-2A-GFP (PX458; Addgene: #48138). The plasmids were transfected into NSCLC cells by Lipofectamine 3000, and single-cell clones expressing GFP were selected using Flow cytometry. RT-PCR and Western blot analyses were further pursued to isolate the METΔ14Ex cell clone. The individual single-cell clones in which MET exon 14 was deleted in both alleles were selected as the METΔ14Ex model, whereas the alternate clones with intact exon 14 were isolated as the wild-type MET model.

Falcon cell culture inserts (Corning) with 8-μm pore size were coated with 10% growth factor reduced Matrigel (BD Biosciences). Cells were seeded onto the top chamber of an insert in serum-free medium and serum-free media supplemented with 100 ng/mL HGF were added into the lower chamber in a 24-well plate. The plate was incubated at 37°C for 24 hours, and the cells were fixed and stained with 0.2% crystal violet. Cells that invaded to the bottom of the filter were counted by light microscope (Olympus CKX-41).

All the animal experiments were performed according to general animal care and experimentation guidelines following a protocol approved by the Institutional Animal Care and Use Committee of Albert Einstein College of Medicine. Four to 6-week-old athymic nu/nu mice were purchased from The Jackson laboratory, and 4 to 6-week-old ICR-SCID mice were purchased from Envigo and housed at the Animal Housing and Studies Facility. (i) In vivo metastasis assay. A bioluminescence imaging (BLI) system was applied to monitor and quantify the formation of lung metastases. Luciferase-positive cancer cells were generated using a lentivirus expression system, which encodes the luciferase reporter used to produce a lentivirus carrying the luciferase gene. Lentivirus particles were generated in 293T cells by the cotransfection of three plasmids: pHIV-Luc-ZsGreen, psPAX2 and pMD2.G with the use of Lipofectamine 3000. After 48 hours of incubation, the medium containing the lentivirus was harvested and added to the cells along with 8 μg/mL polybrene. Single-cell expressing ZsGreen was isolated by Flow cytometry. A total of 2 × 10⁶ Luciferase-labeled tumor cells were injected into each mouse via tail vein injection. The formation of metastases was monitored every week by BLI using an In Vivo Imaging System (PerkinElmer). Mice were anesthetized with 3% isoflurane and injected intraperitoneally with 150 mg/kg body weight of d-luciferin (PerkinElmer) in PBS. At 10 minutes after the injection, bioluminescence was detected and photon flux (photons/seconds) was measured. (ii) Xenograft studies and in vivo drug administration. A suspension of 1× 10⁶ tumor cells in 100 μL of PBS was implanted subcutaneously into the left flank of nude mice. Tumor volume was determined from caliper measurements of tumor length (L) and width (W), according to the formula of V = (W² × L)/2. Both tumor size and body weight were measured three times per week. For drug administration, once tumors reached a size of 100 mm³ mice were either treated with MGCD516 (100 mg/kg) or capmatinib (25 mg/kg).

A tissue microarray (TMA) constructed using tumor tissues from a cohort of 41 PSC patients with 8 METΔ14Ex positive cases were used for IHC analysis. This TMA cohort has been previously described by Liu and colleagues (10).

No statistical method was used to predetermine the sample size. For in vitro assays, at least two independent experiments were performed, and the exact n numbers used in each experiment are indicated in the respective figure legends. The two-sided t test was used to estimate the statistical significance of differences between two groups. Two-way ANOVA with Bonferroni correction was used to determine statistical significance for real-time quantitative PCR analysis. Data are presented as the mean ± SD, unless otherwise noted. Statistical analysis was performed using GraphPad Prism (version 8.0). Statistical significance is indicated as *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

The sequencing data (RNA-seq) have been deposited in the National Center for Biotechnology Information's Gene Expression Omnibus (GEO; accession number: GSE182684).

To elucidate the role and underlying mechanisms by which METΔ14Ex contributes to tumorigenesis in NSCLC, we established several METΔ14Ex-altered lung cancer cell line models (H292, H125, and A549) using CRISPR genome editing technology (Fig. 1A). The expression of the aberrant METΔ14Ex variant was verified at both the RNA level (Supplementary Fig. S1A) and protein level (Fig. 1B, top). Using MET and METΔ14Ex cells, we first assessed the effect of METΔ14Ex on receptor kinase activity in response to short-term HGF exposure of 5 minutes. HGF treatment led to robust phosphorylation at Y1234/1235, found in the activation loop of the tyrosine kinase domain positively modulates enzyme activity (7, 16), and Y1349 tyrosine residues in both MET and METΔ14Ex cells, whereas the phosphorylation at Y1003 was only detected in MET but not in METΔ14Ex cells due to the absence of exon 14 as anticipated (Fig. 1B). Because no significant differences in phosphorylation were detected at Y1234/1235 or Y1349, this indicates that short-term HGF exposure does not significantly affect the kinase activity between MET and METΔ14Ex.

As ligand-dependent receptor dimerization is crucial for receptor kinase activity (17), we investigated whether METΔ14Ex affects HGF-dependent receptor dimerization. HGF stimulation led to strong dimerization of the MET receptor in both MET and METΔ14Ex cells without significant differences (Supplementary Fig. S1B). We next investigated how METΔ14Ex regulates receptor fate and the corresponding downstream signaling. Remarkably, METΔ14Ex delayed HGF-dependent receptor degradation and significantly increased HGF-triggered activity of the PI3K/Akt and MAPK signaling cascades in multiple cell models compared with MET (Fig. 1C; Supplementary Fig. S1C and S1D). Taken together, these results indicate that METΔ14Ex does not affect receptor dimerization and kinase activity in response to short-term HGF treatment, but prolonged HGF exposure increases and, even more importantly, protracts receptor kinase activity and downstream signaling by impairing receptor degradation.

HGF binding to MET triggers rapid internalization of MET, leading to its intracellular trafficking and ultimately degradation (18, 19). The E3 ligase c-Cbl acts as a critical regulator of MET degradation by facilitating the ubiquitination of MET kinase and its trafficking to the lysosome (14, 15, 19). We next clarified the role of c-Cbl in HGF-induced METΔ14Ex degradation using siRNA knockdown of c-Cbl in our cell model (Supplementary Fig. S1E). In comparison with control siRNA-transfected cells in which HGF treatment promoted a higher degree of receptor degradation in MET cells than in METΔ14Ex cells, c-Cbl silencing significantly impeded the receptor degradation in MET cells (Supplementary Fig. S1F). These results indicated that METΔ14Ex impeded HGF-induced receptor degradation (vs. MET) via impairing c-Cbl-dependent regulation. Interestingly, c-Cbl knockdown also impaired more receptor degradation than siRNA control in METΔ14Ex cells. This result raised a possibility that c-Cbl might also be involved in an exon 14 coding domain-independent MET degradation process.

Ligand binding triggers rapid internalization of the MET receptor, which is the first step toward initiation of receptor trafficking (19). Using a surface biotinylation internalization and recycling assay (Supplementary Fig. S1G), we found that 55% of MET became internalized following exposure to HGF treatment for 20 minutes in MET cells; however, notably less METΔ14Ex was internalized into the cytoplasm (Fig. 1D). c-Cbl is critical for clathrin-dependent MET internalization that occurs following the indirect binding of c-Cbl to MET via Grb2 (20). We found that c-Cbl silencing attenuated HGF-triggered receptor internalization to the same level in both MET and METΔ14Ex cells (Fig. 1D), which suggests that c-Cbl is tightly involved in the regulation of ligand-dependent internalization. Following internalization, the activated MET receptor is then subjected to endosomal recycling pathways that return endocytosed proteins and lipids to the plasma membrane (21). We found that MET returned to the cell membrane at comparable rates with recycled METΔ14Ex (Fig. 1E). These results indicated that METΔ14Ex leads to impaired HGF-dependent receptor internalization but not recycling, principally owing to its interaction with c-Cbl. Following HGF-induced internalization, the MET receptor is trafficked to the endosome and then the lysosome for degradation. In this study, we found that HGF treatment indeed directed internalized MET to the early endosome (Fig. 1F and G) and lysosome (Fig. 1H and I). However, compared with MET, METΔ14Ex significantly attenuated HGF-dependent trafficking to the early endosome (Fig. 1F and G) and late lysosome (Fig. 1H and I). These results indicate that METΔ14Ex impairs HGF-dependent receptor internalization and subsequent trafficking to the early endosome and lysosome.

The HGF/MET axis plays a critical role in the promotion of cell scattering and spreading, key characteristics of metastasis (22, 23). We determined the role of METΔ14Ex in the regulation of these HGF-mediated cellular events. In serum-free culture conditions, both MET and METΔ14Ex cells showed an aggregate growth character; HGF exposure stimulated prominent cell scattering in both MET and METΔ14Ex cells (Fig. 2A). HGF-promoted cell scattering was observed to be significantly increased in METΔ14Ex cells as compared with MET cells (Fig. 2A). This result clearly indicated METΔ14Ex as a positive promoter of HGF-dependent cell scattering. The same result was also observed in high cell density culture conditions (Supplementary Fig. S2A).

MET abnormalities have long been recognized to be associated with malignant transformation and with progression of invasion and metastasis (24). The driving role of METΔ14Ex in promoting cell scattering also implicates its involvement in the regulation of cell–cell dissociation and subsequent cell motility. Given these, we performed wound-healing assays and Transwell-invasion assays to assess the role of METΔ14Ex in cell migration and invasion. As expected in the wound-healing assay, METΔ14Ex resulted in approximately 95% gap closure [in comparison with HGF (−) control], significantly higher than the wound closure observed in MET cells (Fig. 2B). Similarly, METΔ14Ex enhanced HGF-dependent cell invasion relative to MET (Fig. 2C). The pronounced enhancement effect of METΔ14Ex on HGF-dependent cell migration and invasion was also similarly verified in H125 (Supplementary Fig. S2B and S2C) and A549 cells (Supplementary Fig. S2D and S2E). In addition, we showed METΔ14Ex cells in response to HGF treatment traveled a markedly longer trajectory than MET cells using live cell time-lapse imaging (Fig. 2D). Quantitative analysis of the cellular tracks revealed that the average speed was significantly elevated in METΔ14Ex cells as compared with MET cells (Fig. 2E). Taken together, these results indicate that HGF-stimulated METΔ14Ex plays a critical role in increasing tumor cell scattering, migration, and invasion.

Next, BLI was used to investigate the effect of METΔ14Ex on metastasis in vivo by injecting luciferase-expressing cells into mice via tail vein. We found that METΔ14Ex significantly increased the luciferase signal, indicating enhanced metastasis formation as compared with MET in both H292 (Supplementary Fig. S3A and S3B) and A549 cells (Fig. 2F and G). Histological analysis of lung tissues confirmed that the number of metastatic lesions produced by METΔ14Ex cells was significantly increased compared with MET cells (Fig. 2H and I). Intriguingly, METΔ14Ex-driven brain metastasis was found in 3 of total 6 (50%) mice whereas no brain metastasis was observed in wild-type MET cells injected mice (Supplementary Fig. S3C). Histological analysis of brain tissue confirmed the presence of the metastatic lesions in METΔ14Ex xenograft mice (Supplementary Fig. S3D). These findings clearly demonstrate the key role of METΔ14Ex in the promotion of in vivo tumor cell metastasis. To more comprehensively characterize the role of METΔ14Ex in tumor progression, we examined the effect of METΔ14Ex on in vivo tumor growth. We found that METΔ14Ex significantly increased tumor growth than MET in a xenograft model (Supplementary Fig. S3E and S3F). These results provide strong evidence that METΔ14Ex alteration plays a positive role in promoting tumor growth and metastasis in vivo.

To determine the involvement of key molecular pathways in the biological effects of METΔ14Ex, RNA-seq was performed on MET and METΔ14Ex cells treated with HGF for 0, 6, and 24 hours (Fig. 3A). The differentially expressed genes (DEG) between MET and METΔ14Ex samples were identified at different time points, with the criteria of adjusted P value (<0.05) and fold change (logFC > 1 or < −1). The results showed 782 DEGs at 0 hours HGF treatment, 895 DEGs following 6 hours HGF treatment, and 1,949 DEGs after 24 hours (Fig. 3B). The change in the DEG numbers in relation to HGF treatment indicates the potential dynamic effects of HGF stimulation between MET and METΔ14Ex samples.

Next, we used the DEGs as the inputs for Ingenuity Pathway Analysis (IPA) and evaluated the dominant functional pathways in which the aforementioned DEGs were principally enriched. IPA revealed distinct predicted patterns of decreasing or increasing of biological functions when cells were treated with HGF over time (at 0, 6 or 24 hours). In the absence of HGF (0 hours), the functions associated with cell movement were suppressed in METΔ14Ex sample, whereas the pathways involved in development of vasculature and angiogenesis increased after 6 hours HGF treatment. A large number of cell movement–related functions were significantly enhanced at 24 hours HGF treatment (Fig. 3C; Supplementary Table S2A–S2C). We further extracted the cell movement–related pathways from the overlapping functional categories and focused on 6 top-ranked functions (details in Materials and Methods): (i) migration of cells; (ii) cell movement; (iii) invasion of cells; (iv) chemotaxis; (v) homing of cells; and (vi) adhesion of tumor cell lines (Fig. 3D). Comparing the Z-scores of these pathways along with HGF treatment time-axis, we found that METΔ14Ex hyperactivated and led to protracted cell movement–related signaling with HGF exposure. More specifically, METΔ14Ex promotes HGF-triggered cell movement in a time-dependent manner; different from wild-type MET with a short burst of induction, METΔ14Ex leads to a protracted but pronounced induction of migratory pathways (Fig. 3C and D; Supplementary Fig. S4A).

The IPA “Regulator Effects” algorithm connects upstream regulators, dataset molecules, and downstream functions or diseases. In the top ranked Regulator Effects (Supplementary Table S3) based on a consistency score with 24 hours HGF treatment, most identified downstream biological functions are associated with cell movement. The downstream target genes were extracted from top ranked (ID #1) Regulator Effects and the expression of these genes is shown as a heatmap (Supplementary Fig. S4B). Within this list, a large group of genes upregulated in METΔ14Ex models has been well characterized as playing pivotal roles in invasion and metastasis.

A gene set enrichment analysis (GSEA) confirmed this finding and demonstrated that HGF stimulation triggered a dynamic activation of cell movement–associated biological processes (Gene Ontology) and biological pathways (Kyoto Encyclopedia of Genes and Genomes,KEGG) in METΔ14Ex cells compared with wild-type MET in a time-dependent manner (Supplementary Table S4A–S4F). Relative to the 0 or 6 hours treatment group, the transcriptomic signatures upregulated upon 24 hours HGF exposure in METΔ14Ex cells (vs. MET) were positively enriched in cell movement–associated biological processes, in which multiple gene sets, including regulation of cytoskeleton dynamics, wound healing, and cell junction organization were top ranked (Fig. 3E; Supplementary Table S4A–S4C). KEGG pathway analysis also indicated that cell invasion and metastasis-associated gene sets such as actin cytoskeleton regulation and focal adhesion (FA) were top ranked (Fig. 3F; Supplementary Table S4D–S4F). Together, these results strongly support that hyperactivation of HGF/METΔ4Ex-mediated signaling elicits a strong and concerted activation of the cell movement machinery.

We next went on to elucidate the role of METΔ14Ex in the regulation of cytoskeleton reorganization. During cell migration the cytoskeleton is dynamically remodeled (Fig. 4A) producing the force necessary for cell movement (25–27). GSEA plots revealed that the upregulated DEG signature (METΔ14Ex vs. MET; HGF treatment for 24 hours) was enriched in cytoskeleton reorganization (Fig. 4B; Supplementary Fig. S5A) and Rho GTPase-mediated signaling (Fig. 4B). A group of DEGs extracted from multiple cytoskeleton reorganization gene sets were upregulated in METΔ14Ex cells when compared with MET following 6 or 24 hours HGF treatment (Fig. 4C). These genes regulate cytoskeletal reorganization by interacting with actin and microtubule cytoskeletons or alternatively mediating signaling. We then validated the expression changes of several key genes, including MAPT, TWF2, TUBAL3, MAP1B, VAV1, and ABLIM2 by quantitative RT-PCR (Supplementary Table S5). Our results demonstrated that METΔ14Ex significantly enhanced expression of these genes as compared with MET in a time-dependent manner (Fig. 4D; Supplementary Fig. S5B).

Rac1, a Rho family GTPase, is a crucial regulator in actin and microtubule cytoskeletal dynamics (26, 27). In our study, GSEA plot showed that the METΔ14Ex-upregulated gene signature was enriched in Rho GTPase signaling (Fig. 4B) and lamellipodium assembly genes (Supplementary Fig. S5C), suggesting the involvement of Rac1 activity in METΔ14Ex-mediated cell movement. We tested the effect of METΔ14Ex on Rac1 activity and found that METΔ14Ex maintained a higher level of Rac1 activity than MET upon prolonged HGF exposure (Fig. 4E; Supplementary Fig. S5D). Membrane ruffling is the hallmark phenotype of activated Rac1. METΔ14Ex elicited pronounced membrane ruffling compared with MET; Rac1 silencing (Fig. 4F) decreased ruffling in METΔ14Ex to comparative levels observed in MET control cells (Fig. 4G). As expected, Rac1 knockdown also significantly repressed HGF/METΔ14Ex-enhanced cell invasion when compared with MET (Fig. 4H; Supplementary Fig. S5E and S5F). We further tested the therapeutic opportunity of blockade of METΔ14Ex signaling-elicited cytoskeleton remodeling using 2 Rac1 inhibitors, MBQ-167 and EHop-016. Similarly to Rac1 knockdown, MBQ-167 and EHop-016 dramatically inhibited HGF-dependent Rac1 activity and invasion in both MET and METΔ14Ex cells (Fig. 4I and J; Supplementary Fig. S5G and S5H). Therefore, these results suggested a therapeutic potency of Rac1 inhibitors on those lung cancer patients with METΔ14Ex alteration. PI3K/Akt signaling has been well characterized in the regulation of Rac1 activity (28, 29). In our study, a widely used PI3K inhibitor, LY294002, remarkably suppressed HGF/METΔ14Ex-driven Rac1 activity and subsequent cellular invasion (Fig. 4K and L). Taken together, these results clearly demonstrate that METΔ14Ex robustly activates cytoskeleton dynamics and accelerates tumor cell movement at least partially through upregulation of the PI3K/Akt-Rac1 pathway.

Epithelial-to-mesenchymal transition (EMT) programs contribute to the acquisition of invasive properties that are essential for metastasis (30, 31). TGF-β signaling has been well characterized as a potent driver of cancer progression through EMT (30, 31). In our study, GSEA plot demonstrated that METΔ14Ex-upregulated genes were significantly enriched in both hallmarks of EMT signatures and TGF-β signaling pathways (Supplementary Fig. S6A). Subsequent expression analysis of EMT markers indicated that METΔ14Ex only slightly repressed expression of E-cadherin and Snail 1 but had no effect on other markers (Supplementary Fig. S6B), suggesting that an alternative mechanism may be involved in EMT regulation. EMT in tumor populations promotes invasion not only by triggering a loss of cellular cohesion, but also by conferring migratory properties and the capacity to reorganize the extracellular matrix (ECM; refs. 30, 31). Degradation of proximal ECM by metalloproteinases at the leading edge of migrating cells is crucial in this process (Fig. 5A). The GSEA plot indicated that a group of upregulated genes in METΔ14Ex (vs. MET) was associated with ECM disassembly (Supplementary Fig. S7A). Among these METΔ14Ex-upregulated genes related to ECM degradation (Fig. 5B), the functions of matrix metallopeptidases (MMP)1, MMP2, MMP9, or ADAM8 have been well characterized for involvement in matrix degradation and metastasis (32). Using qRT-PCR, we confirmed that the expressions of several proto-typical genes, including MMP1, MMP10, MMP2, MMP9, MMP3, and ADAM8, were upregulated in HGF-treated METΔ14Ex cells relative to MET cells (Fig. 5C; Supplementary Fig. S7B). METΔ14Ex dramatically increased MMP1 expression with HGF treatment (Fig. 5D; Supplementary Fig. S7C). Subsequent MMP1 knockdown largely repressed HGF-driven cell invasion in both METΔ14Ex and MET cells (Fig. 5E and F; Supplementary Fig. S7D and S7E). These results indicate that METΔ14Ex promotes HGF-dependent cell migration and invasion, partially through upregulation of certain ECM degradation–related genes resulting in ECM degradation.

Adhesion of cells to the ECM through FA structures is a key step in migration (Fig. 5G), that is regulated by a large group of FA-related proteins (33). Our GSEA plot data revealed that a cluster of DEGs (METΔ14Ex vs. MET; HGF treatment for 24 hours) were positively enriched in cell-matrix adhesion and FA gene sets (Fig. 5H), indicating a potential role of METΔ14Ex in regulation of FAs. We extracted these DEGs from the aforementioned gene sets and the individual gene expression pattern is exhibited in Fig. 5I. Several expected DEGs, including ITGB3, ITGB4, ITGB5, SPP1, LAMB3, LAMA3, LAMC2, and PLAU, were verified to be largely enhanced in METΔ14Ex cells following 24 hours HGF treatment as compared with MET cells (Fig. 5J; Supplementary Fig. S7F).

Interaction of integrins with adaptor proteins in cell-matrix adhesion complexes (CMAC) provides both physical and regulatory links between the ECM and actin cytoskeleton during cell movement (34). Given the critical role of integrin-FAK signaling in the regulation of CMACs, we elucidated FAK regulation in this process. GSEA plot showed that the METΔ14Ex-upregulated DEGs (vs. MET) are enriched in the integrin-FAK signaling pathway (Supplementary Fig. S8A). We found that the activity of FAK kinase was significantly elevated in METΔ14Ex cells relative to MET cells (Supplementary Fig. S8B). Because FAK can be activated by multiple growth factors as well as via its interaction with integrins, we used the FAK inhibitor Y15 to test the effects of FAK on METΔ14Ex downstream signaling. Y15 treatment markedly inhibited AKT and partially repressed MAPK activity (Supplementary Fig. S8C). In contrast, the PI3K inhibitor LY294002 only slightly decreased p-FAK, whereas U0126 (MAPK inhibitor) did not affect HGF-dependent FAK activity (Supplementary Fig. S8D). These results indicate that FAK was directly activated by HGF, and activated FAK may facilitate AKT phosphorylation. We next assessed whether the inhibition of FAK could affect cell scattering and invasion. HGF significantly increased tumor cell scattering and invasion in METΔ14Ex cells in comparison with MET cells; Y15 treatment dramatically repressed cell scattering and invasion (Supplementary Fig. S8E and S8F). Altogether, these results indicate that the upregulation of FA-related genes and activation of integrin-FAK signaling play important roles in METΔ14Ex-mediated cell scattering and invasion.

Our studies demonstrated that HGF/METΔ14Ex signaling upregulated a large group of genes associated with cell movement. Next, we attempted to correlate the expression of these DEGs with METΔ14Ex alteration on a group of specimens from NSCLC patients. A total of 41 PSC patients with 8 METΔ14Ex-positive cases were subjected to IHC analysis. The target DEGs selected for IHC were set to meet the following criteria: (i) the DEGs were localized in cytoskeleton remodeling, disassembly of cell-matrix and FA-associating gene sets, and have been validated in RNA expression; (ii) the DEGs are also localized in top ranked (ID #1) Regulator Effects (Supplementary Table S3); and (iii) the expression of DEGs were more than two-fold in METΔ14Ex tumor versus normal tissue by analyzing the transcript expression profile (tumor vs. normal) of lung carcinoma patients with METΔ14Ex alteration in The Cancer Genome Atlas RNA-seq database. Following this principle, 4 genes, including ITGB3, SPP1 PLAU, and MMP9, encoding integrin β3, osteopontin (ΟPΝ), uPA, and MMP9, respectively, were selected. IHC analysis revealed that the expression of integrin β3 and OPN were significantly higher in METΔ14Ex than wild-type MET patients, whereas overexpression of uPA and MMP9 trended toward higher levels in METΔ14Ex as compared with MET patients (Fig. 6A and B). These results suggest that besides potential functional roles in PSC progression, integrin β3, and OPN may serve as candidate biomarkers for METΔ14Ex NSCLC patients.

Thus far, our studies revealed that METΔ14Ex significantly increased tumor cell scattering, invasion, and metastasis primarily via upregulation of numerous genes and signaling pathways involved in activation of cytoskeleton reorganization, promotion of cell-ECM disassembly, formation of FA, as well as enhancement of tumor-associated angiogenesis (Fig. 6C). The pivotal role of METΔ14Ex in tumor progression identified in our study led us to test the effect of two selected potent MET inhibitors, MGCD516 and capmatinib, on METΔ14Ex-mediated tumor growth and metastasis. We found that both MGCD516 and capmatinib could completely eliminate HGF-induced MET and METΔ14Ex receptor degradation in both MET and METΔ14Ex cells (Fig. 7A). Furthermore, observed METΔ14Ex-enhanced signaling output, including phospho-AKT and phospho-MAPK, were rapidly repressed to relatively low levels analogous to wild-type MET in these cells (Fig. 7A). Subsequently, in an in vivo METΔ14Ex xenograft tumor model, MGCD516 and capmatinib showed potent inhibitory effects on tumor growth (Fig. 7B and C). Our results indicated that MGCD516 and capmatinib significantly repressed cell scattering in both H292 MET and METΔ14Ex cells (Fig. 7D). In addition, MGCD516 and capmatinib also dramatically inhibited METΔ14Ex-mediated Rac1 activity and FAK phosphorylation (Fig. 7E), and consequently, migration (Fig. 7F) and invasion (Fig. 7G). We further investigated the effect of MET inhibitors on H596 cells, a lung cancer cell line harboring METΔ14Ex alteration (10). HGF treatment induced phosphorylation of MET and activated downstream PI3K/Akt and MAPK signaling cascades (Supplementary Fig. S9A). Importantly, MGCD516 and capmatinib also significantly attenuated HGF/METΔ14Ex signaling-driven PI3K/Akt and MAPK signaling (Supplementary Fig. S9B), and thereby inhibited HGF-dependent cell scattering, invasion, and wound healing (Supplementary Fig. S9C–S9E) in H596 cells. On the basis of these results, we tested the role of MET inhibitors on in vivo tumor metastasis. We found that MGCD516 or capmatinib administration significantly repressed tumor cell metastasis (Fig. 7H). These results further indicate that MGCD516 and capmatinib can dramatically inhibit METΔ14Ex cell growth as well as metastasis.

METΔ14Ex has recently been identified as a recurrent actionable driver mutation with a frequency of 2%–4% in NSCLC (8–13). Although a large body of evidence has defined the role of METΔ14Ex in tumorigenesis, the underlying mechanisms by which METΔ14Ex promotes tumor progression remain poorly understood. Although MET inhibitors such as capmatinib now have a proven role in patient management, acquired resistance is inevitable and novel approaches offering more durable outcomes will be needed. Therefore, further research dissecting the molecular signaling steps involved in METΔ14Ex-driven tumorigenesis as well as unique mesenchymal differentiation will offer opportunities to better understand the mechanisms of response to MET TKI therapy and highlight potential molecular targets that can be targeted for alternative and combination approaches for more effective therapies. We explored the tumorigenic role of METΔ14Ex through a variety of biochemical and molecular strategies in this study. Our results have led to novel insights into the mechanisms by which METΔ14Ex increases the activity of the receptor kinase, and therefore promotes cell migration, invasion, and metastasis.

Using several isogeneic cell models generated with CRISPR that yielded normal copy numbers mimicking the natural non-overexpressing scenario, we first demonstrated that ΜΕΤΔ14Εx did not affect baseline HGF-dependent receptor dimerization and auto-phosphorylation, which is logical as the extracellular Sega domain, but not the juxtamembrane region is necessary for ligand binding and dimerization (35). The absence of the portion of the juxtamembrane domain encoded by exon 14 did not interfere with dimerization and subsequent auto-phosphorylation. This also provides a simple explanation as to why METΔ14Ex does not affect MET receptor kinase activation in response to short-term HGF treatment. Consistent with previous reports (8, 11), our studies also demonstrate that the METΔ14Ex variant leads to delayed HGF-triggered receptor degradation upon prolonged exposure to HGF. As a result, impairment of METΔ14Ex degradation leads to an increase in kinase activity as well as protracted and enhanced downstream signaling activity. We subsequently demonstrated that METΔ14Ex impairs receptor internalization, diminishing the trafficking of the receptor to the early endosome and late lysosome, which clearly indicates that METΔ14Ex is retained on the membrane and protected from internalization, eventually resulting in increased MET activity. In this process, the E3 ligase c-Cbl emerges as a dominant regulator for MET endocytosis via its interaction with the MET juxtamembrane domain (14, 15, 19). c-Cbl is critical for the ligand-dependent ubiquitination of many RTKs, including MET (14, 15, 19). Once ubiquitinated, MET is recruited to the endosomal sorting complex required for transport (ESCRT) complexes where receptors are retained in specialized microdomains of endosomes coated with bi-layered clathrin and then internalized into the endosomal lumen (19). Consequently, the absence of the juxtamembrane domain in METΔ14Ex impairs its interaction with c-Cbl, leading to the escape of METΔ14Ex from HGF-dependent endocytosis.

In our study, HGF binding endows enhanced and protracted activity to METΔ14Ex due to impaired degradation (vs. MET), resulting in enhanced cell migration and invasion in vitro as well as metastasis in vivo. Overexpression of MET has previously been reported to be associated with metastasis and poor prognosis (36–38). In addition, MET mutations, including M1268T and D1246N in the kinase domain or R988C and T1010I in the juxtamembrane domain, exhibit a driving role in tumorigenicity by modulating cytoskeletal functions, enhancing cell proliferation, and motility (39, 40). All these findings imply that the hyperactivity of HGF/MET signaling is the major source leading to cell movement and metastasis. Tumor cells gain invasive capacity during the conversion from a benign to a malignant state, underscoring invasive capacity as a key hallmark of malignant tumors (41). Our results highlight the importance of METΔ14Ex as a key regulator in promoting the malignant process by conferring metastatic potential to tumor cells.

If we regard the phenotype of METΔ14Ex-triggered invasion and metastasis as “extrinsic proof,” further “intrinsic proof” comes from a molecular dissection of the underlying mechanisms. Analysis of the HGF-dependent global transcriptional signature changes under METΔ14Ex rigorously supports its role as a crucial promoter of cell invasion and metastasis. The top enriched functional and signaling pathways in both IPA and GSEA very closely associate with cell movement, in which activated HGF/METΔ14Ex signaling upregulates a large cluster of genes associating with cell cytoskeleton reorganization and cell-matrix adhesion reorganization, both well-known biological processes driving cell movement. Therefore, these findings imply that METΔ14Ex might preferentially regulate cell migration and invasion relative to cell growth after driving tumorigenesis.

The movement of cancer cells is a complex process requiring dramatic remodeling of the cell cytoskeleton, including actin microfilaments, microtubules, and intermediate filaments (25). During migration, the cells first become polarized and elongated via a protrusion that is composed of a broad lamillopodia and spike-like filopodia structure created at the leading edge (25, 42, 43). The elongated protrusions contact with adjacent ECM via a structure called FA, which transmits the traction forces required for movement. Tumor cells secrete multiple MMPs near substrate attachment sites leading to the cleavage of ECM components that clear the path for cell movement (42, 43). These steps occur sequentially or concurrently to yield a positive force that allows the cell to move in a specific direction (42, 43).

In our study, METΔ14Ex-upregulated gene signatures correlate with regulation of cytoskeleton reorganization—ranked top in both GO enrichment and KEGG pathway analysis. A large number of genes associated with cytoskeleton reorganization were upregulated in METΔ14Ex cells as compared with MET cells. Aberrant overexpression of these genes facilitates cytoskeleton remodeling through direct or indirect binding to actin or microtubule. Increased activation of Rho family GTPases-mediated signaling in METΔ14Ex cells (vs. MET, Fig. 4B) clearly promotes these processes as the crucial role of these critical GTPases in cytoskeleton remodeling has been previously well characterized (25–27). We tested an archetypal GTPase, Rac1, and found that METΔ14Ex indeed promoted HGF-dependent Rac1 activity. Activation of Rac1 initiates actin cytoskeleton reorganization that then promotes the formation of large membrane protrusions, which drive motility in many cell types (25, 26). This regulation was, at least partially, mediated via the PI3K/Akt pathway as inhibition of PI3K suppressed Rac1 activity and cell invasion. This result is consistent with previous studies where activation of Rac1 via the PI3K/Akt pathway was well characterized (28, 29, 44). Vav1, which was also found to be upregulated in our data (Fig. 4C and D), may be a key mediator. Vav1 is a guanine nucleotide exchange factor (GEF) for Rac1 that can be phosphorylated by PI3K. PtdIns(3,4,5)P3 binding to Vav1 facilitates phosphorylation of tyrosine and the opening of the PH domains, which then mediates the interaction with Rac-GDP and catalysis (28, 29, 44). Inhibition of PI3K repressed generation of membrane PtdIns(3,4,5)P3, which then blocked the activation of Vav1 and Rac1.

In addition, METΔ14Ex triggered the upregulation of a large group of genes associated with cell-matrix adhesion, including the expression of MMPs and integrins, and activity of FAK signaling. In a mesenchymal type of tumor cell migration, matrix-degrading proteases such as MMPs accumulate in an integrin‐dependent manner at the leading edge of migrating cells, leading to localized ECM proteolysis (45). Highly expressed MMPs facilitate tumor cell invasion through degradation of ECM and overcoming tissue barriers. In addition, the formation of integrin-dependent adhesions at the cell front that links the actin cytoskeleton to the substratum helps generate traction forces that move the cell forward (45). In our study, upregulation of integrin as well as ECM components, such as laminin and collagen, may contribute to form strong cell-ECM adhesion and thereby promote cell migration. Simultaneously, HGF/METΔ14Ex-activated FAK signaling boosts tumor cell motility through induction of cytoskeleton remodeling and promotion of cell-ECM adhesion dynamics (46). The activation of FAK creates a high-affinity binding site for Src, PI3K, and growth factor receptor-bound protein 7 (Grb7; ref. 46). Interaction of PI3K with phospho-FAK leads to the activation of PI3K and its downstream effectors. Our results show that inhibition of FAK dramatically represses the activation of PI3K/Akt signaling and inhibits cell invasion, which indicates that FAK-boosted tumor cell motility may be partially mediated via the PI3K/Akt pathway. Intriguingly, our data indicated that expression of integrin β3 and OPN significantly correlate with METΔ14Ex expression in PSC patients with METΔ14Ex mutation. OPN binding to integrin αvβ3 induces in vitro FA assembly and cell migration through activation of a broad range of intracellular pathways, including FAK and PI3K/Akt pathways (47, 48). The current data also suggest that integrin β3 and OPN could be potential biomarkers for NSCLC patients with METΔ14Ex alteration. Expression of integrin β3 has been linked to poor prognosis for breast cancer and melanoma patients, but the relation of integrin β3 to NSCLC is poorly understood (49). The mechanisms underlying the aggressive phenotype associated with OPN expression have been extensively investigated in NSCLC, and upregulation of OPN is proposed to be associated with advanced stages, recurrence risk, and lymph node metastases (48, 50). Moreover, the emerging evidence indicates that patients with OPN-expressing tumors have worse relapse and overall survival than OPN-negative patients, suggesting that OPN might be used as a prognostic biomarker in NSCLC (51, 52). More work is required to clarify the role of the OPN–integrin αvβ3 axis in METΔ14Ex-mediated invasion and metastasis, and clinical correlation with NSCLC patient with METΔ14Ex alteration.

Taken together, our studies highlight the oncogenic role of METΔ14Ex alteration in NSCLC and reveal that METΔ14Ex may drive the highly aggressive, metastatic nature of NSCLC through upregulation of global gene transcripts and crucial signaling pathways that promote cytoskeleton reorganization, cell-ECM disassembly, and formation of FA as well as enhancement of tumor-associated angiogenesis. RNA-seq analysis has indicated the involvement of METΔ14Ex-mediated signaling in regulating tumor-associated angiogenesis, and future work will be designed to further explore the details underlying this regulation.

METΔ14Ex is generally mutually exclusive with other recognized oncogenic drivers, suggestive of a driver oncogene role. Potent MET TKIs yield high-response rates and durable benefits, suggesting that MET inhibitors are excellent and possibly first-line therapeutic options for patients with lung adenocarcinoma and METΔ14Ex mutation (53). Our results now provide compelling preclinical evidence to better define the effect of MET inhibitors. Namely, MGCD516 and capmatinib remarkably inhibited HGF/METΔ14Ex signaling and repressed cell invasion in vitro and tumor growth and metastasis in vivo. Thus, the blockade of HGF/METΔ14Ex signaling using MET inhibitors is a potent therapeutic option for METΔ14Ex-mutated NSCLC patients as already seen in clinical studies. More notably, we identified that METΔ14Ex positively affected crucial factors of tumor cell invasion and metastasis, such as activation of Rac1, overexpression of MMP1, OPN, and integrin β3. On the basis of these findings and given the crucial role of these aforementioned factors on tumor progression, novel inhibitory strategies can now be explored to ensure more effective and enduring treatment strategies for such patients.

E. Shum reports personal fees from AstraZeneca, Genentech, and Janssen outside the submitted work. H. Cheng reports grants from ALA Lung Cancer Discovery Award, ACS Research Scholar Grant, and LCFA/IASLC Foundation Lori Monroe 802 Scholarship Award during the conduct of the study. B. Halmos reports grants and personal fees from Pfizer and Novartis, and grants from Mirati, grants and personal fees from AstraZeneca and Boehringer-Ingelheim, and grants from AbbVie, personal fees from Janssen, grants from Daiichi, personal fees from Takeda, grants from GSK, personal fees from Apollomics and TPT Pharma, grants and personal fees from Merck and BMS, and grants from Jazz, Crestone, Blueprint, personal fees from Genentech, grants and personal fees from Beigene, and grants from Yuhan outside the submitted work. No disclosures were reported by the other authors.

F. Wang: Conceptualization, resources, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. Y. Liu: Data curation, software, formal analysis, investigation, methodology, writing–review and editing. W. Qiu: Investigation, methodology, writing–review and editing. E. Shum: Investigation, visualization, methodology, writing–review and editing. M. Feng: Investigation, writing–review and editing. D. Zhao: Software, formal analysis, investigation, methodology, writing–review and editing. D. Zheng: Formal analysis, supervision, methodology, writing–review and editing. A. Borczuk: Resources, formal analysis, investigation, methodology, writing–review and editing. H. Cheng: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. B. Halmos: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

The authors thank the Albert Einstein College of Medicine (AECOM) Flow Cytometry Core, the AECOM Analytical Imaging Core, and AECOM Histotechnology and Comparative Pathology Core for their excellent technical support. This work was supported by NCI Community Oncology Research Program (NCORP) grant (to B. Halmos), ALA Lung Cancer Discovery Award (to H. Cheng), ACS Research Scholar Grant (to B. Halmos and H. Cheng), LCFA/IASLC Foundation Lori Monroe Scholarship Award (to H. Cheng).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.