EXOC4 Promotes Diffuse-Type Gastric Cancer Metastasis via Activating FAK Signal Open Access

Gastric cancer remains one of the leading causes of cancer-related death, although its incidence has recently declined (1). In recent decades, the histologic classification of gastric cancer based on Lauren's criteria, which subdivides this cancer as intestinal-type, diffuse-type, and mixed-type gastric cancer, has been most commonly used to understand clinical diagnosis, prevention, and treatment (2). Diffuse-type gastric cancer (DGC) accounts for approximately 30% of gastric cancer cases, is more likely to affect younger people, and consists of individual infiltrating neoplastic cells without glandular structures, which contributes to its greater aggressiveness with a worst prognosis and few treatment options (3, 4). The 5-year survival rate of DGC is significantly lower than that of intestinal-type gastric cancer (IGC) due to the higher likelihoods of recurrence and peritoneal metastasis (5, 6). In comparison with well-studied IGC, which has been noted for a cascade involving Helicobacter pylori infection, inflammation, oxidative damage, atrophic gastritis, intestinal metaplasia, dysplasia, and ultimately gastric adenocarcinoma (7), DGC is known as an enigmatic disorder without well-defined precursor lesions or a carcinogenesis model. Specifically, DGC is characterized by highly malignant biological behaviors with prominent infiltration and poor cellular cohesion resulting in a high frequency of metastasis (5, 6, 8, 9). Thus, more thorough investigations are required to reveal the links between signaling pathways and metastasis and relapse in DGC. Ideally, an integrated proteogenomic analysis could generate a new paradigm for detecting proteomic alterations (10) and understanding the malignant biology of this cancer. In this study, a list of proteins differentially expressed between DGCs with and without recurrence was identified on the basis of proteogenomic analyses. Among the identified proteins, exocyst complex component 4 (EXOC4) was identified as a significantly upregulated protein in DGC samples with recurrence.

EXOC4, also known as Sec8, is a component of the exocyst complex that contributes to the tethering of secretory vesicles to the plasma membrane (11–13). The exocyst is an evolutionarily conserved hetero-octameric complex, and in mammalian cells, there are two exocyst subcomplexes consisting of EXOC1–EXOC2–EXOC3–EXOC4 (subcomplex I) and EXOC5–EXOC6–EXOC7–EXOC8 (subcomplex II; ref. 14). These two subcomplexes are recruited to vesicles independently and assemble into the holo-exocyst complex upon arrival at the plasma membrane, where the vesicles are tethered and fused (13, 14). A recent report discovered that exocysts are in dynamic equilibrium with octamers and monomers (15). Every subunit may regulate cellular processes via the exocyst or independent of the exocyst complex. For example, during cell invasion, Exo70 (EXOC7), together with other exocyst components, participates in the secretion of matrix metalloproteinases. Furthermore, Exo70 interacts directly with the actin-related protein 2/3 complex independently of the holo-complex and stimulates actin branching, which in turn promotes invadopodia formation and cell invasion (16–18). Regarding Sec8 (EXOC4), in addition to its roles as a component of the exocyst and stimulating the secretion of neurotransmitters and cytokines, it has also been found to bind directly with c-JNK–interacting protein 4 to regulate MAPK signaling cascades in cervical cancer cells (19) and modulate TGFβ-induced EMT by regulating the expression of N-cadherin and Smad3/4 at the transcriptional level in lung cancer cells (20). However, the potential roles of EXOC4 in gastric malignancies remain largely unclear.

In this study, we revealed that EXOC4 is highly expressed in DGC and specifically correlated with recurrence and poor outcomes in patients with DGC. Mechanistically, EXOC4 could promote metastasis of DGC cells via activation of integrin/EGF-focal adhesion kinase (FAK)^Y397 signaling.

The study enrolled 444 patients with gastric cancer (including 127 DGC and 251 IGC) that had undergone a radical (R0) resection with a D2 lymphadenectomy from Zhongshan Hospital, Fudan University between 2004 and 2008. None of them received any neoadjuvant therapy. The demographic characteristics and the clinicopathologic parameters of each patient, including age, gender, tumor size, location, tumor differentiation, Lauren classification, vessel invasion, and TNM stage were retrospectively collected. Tumors and their adjacent normal tissues were collected within 30 minutes after operation, immediately transferred into liquid nitrogen, and then stored at –80°C until use. Tissues that were >60 mm away from the primary lesions were designated as normal mucosa tissues. Tumors and their adjacent normal tissues stained with hematoxylin and eosin (H&E) were evaluated by pathologists to confirm: (i) Lauren classification of tumors and (ii) no tumor cells in adjacent normal tissues. The study was approved by the ethics committee of Fudan University and was performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. Written informed consent was obtained from each patient.

444 pairs of normal and tumor tissues were taken from enrolled patients, and then the validation of pathologic diagnosis using H&E staining of paraffin sections based on the 2010 World Health Organization classification system was conducted. For all samples, 2-mm cores were punched from the donor blocks and transferred to the recipient paraffin block at defined array positions. Several serial sections (4 μm in thickness) were cut from all tissue microarrays, and one of these was stained with H&E as a reference (21). For IHC staining, sections were incubated with EXOC4 antibody at a 1:100 dilution (Invitrogen, PA5–31027). Both the area and the intensity of immunopositive cells were used to evaluate the IHC staining. According to a previous study (22), the score was calculated by multiplying the staining area (0, <5%; 1, 5%–25%; 2, 26%–50%; 3, 51%–75%; and 4, >75%) and the intensity of the area (0, negative; 1, weak; 2, moderate; and 3, strong). The scores equal or higher than 4 were considered as high expression based on the area under the receiver operating characteristic curves and Youden index (Supplementary Table S1). Two pathologists who were blind to patients’ clinical data in our hospital helped to assess the stained sections. The use of tissues and clinicopathologic data was approved by the ethics committee of Fudan University and was performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. Written informed consent was obtained from each patient.

Human gastric cancer cell lines HGC-27, MKN28, IM95, and AGS were from the Gastric Cancer Center of Fudan University; KATOIII and NUGC-4 were obtained from the Institute of Biochemistry and Cell Biology (Chinese Academy of Sciences); MKN45 and SNU-668 were purchased from Cobioer Biosciences Co., Ltd. All the cell lines obtained from the cell bank were tested for authentication using short tandem repeat fingerprinting and passaged for fewer than 6 months. Cells were cultured in DMEM or RPMI1640 supplemented with 10% FBS (catalog no. 16000–044; Gibco) at 37°C in a humidified atmosphere containing 5% CO₂. Cells were cultured according to the ATCC instructions and routinely tested to be negative for Mycoplasma contamination using the qPCR method (Thermo Fisher Scientific). The following antibodies were used: EXOC4 (Santa Cruz Biotechnology, catalog no. sc-136234, RRID:AB_2101573, 1:500), FAK (Cell Signaling Technology, #3283, 1:1,000), p-FAK (Tyr397;Cell Signaling Technology, #3285, 1:1,000), p-FAK (Tyr925;Cell Signaling Technology, #3284, 1:1,000), p-FAK (Tyr576/577;Cell Signaling Technology, #3281, 1:1,000), Erk (Cell Signaling Technology, #4695, 1:1,000), p-Erk (Thr202/Tyr204;Cell Signaling Technology, #4370, 1:1,000), Vimentin (Abcam, ab8978, 1:500), SNAI2 (Cell Signaling Technology, #9585, 1:1,000), SNAIL (Cell Signaling Technology, #3879, 1:1,000), GFP (Proteintech, 66002–1-Ig, 1:5,000), HA (Sigma, H6908, 1:1,000), ATP1A1 (Proteintech, 14418–1-AP, 1:1,000), Integrin α5 (Santa Cruz Biotechnology, sc-376199, 1:500), Integrin β1 (Santa Cruz Biotechnology, sc-374429, 1:500), EXOC3 (Santa Cruz Biotechnology, sc-374054, 1:200), EXOC5 (Proteintech, 17593–1-Ig, 1:500), and β-Tubulin (BPI, AbM59005–37B-PU, 1:5,000). The following chemotherapeutic reagents were used: EGF was purchased from Sigma-Aldrich; VS-4718 and gefitinib were purchased from MedChemExpress, ATN-161 (Ac-PHSCN-NH2) was manufactured by SelleckChem (S8454).

Thirteen DGC samples were processed for proteomics analysis after excluding those with low DNA quality. All patients received 3-year follow-up every 6 months from the date of surgery. Recurrence was observed in four cases. The 13 DGC samples with their matched adjacent tissues were gathered, and combined with in vitro stable isotope labeling technique (dimethyl labeling) and LC/MS-MS analysis (23, 24) to look for potential biomarkers that were relative to prognosis. Protein sample preparation, digestion, and peptide labeling were performed according to previous reports (23–25). After peptide desalting by stage-tips, labeled peptide mixtures were prefractionated by high-pH RP column (26) on Dionex U3000 system (Thermo Fisher Scientific), and five fractions were obtained for each dimethyl labeling test. Each fraction was analyzed by nanoHPLC Orbitrap Velos (Thermo Fisher Scientific). MS raw files were analyzed by MaxQuant software (RRID:SCR_014485, version 1.3.2.8) and peak lists were searched against the human Uniprot FASTA database (downloaded in October, 2015). A total of 2,417 protein groups were identified and quantified with both protein and peptide FDR < 0.01. The proteomic data have been uploaded to the National Omics Data Encyclopedia (NODE: OEP003074).

Gene-set enrichment analysis (GSEA) was used to identify enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway terms (RRID:SCR_012773, http://software.broadinstitute.org/gsea/index.jsp). The significance of fold enrichment was calculated using a P < 0.05 based on GSEA.

Study protocols involving mice were approved by the Institutional Animal Care and Use Committees of Shanghai Jiao Tong University. All sections of this report adhere to the Reporting of In Vivo Experiments (ARRIVE) guidelines. 5 × 10⁶ luciferase-expressing MKN45 cells transduced with the indicated lentiviruses or plasmids were suspended in 200 μL of sterilized PBS and tail intravenously injected into male 6-week-old Balb/c nude mice or NOD-SCID mice. Then, the mice were randomly allocated to each group (n = 5–6) and orally administered with vehicle (control) or 50 mg/kg VS-4718 (MCE, HY-13917) per day for approximately 4 weeks. Mice were housed in individually ventilated cages with free access to water and food in a specific pathogen-free room. Lung metastasis was monitored periodically by injecting 100 μL of 15 mg/kg d-luciferin (YEASEN Biotechnology). After treatment for 30 days, all the mice were sacrificed by cervical dislocation after CO₂ inhalation.

Fresh tumor samples (GAS190153, GAS190233, GAS190247, GAS190282, GAS190216, GAS190271) were collected from Chinese patients with gastric cancer undergoing gastrectomy at Zhongshan Hospital, Fudan University. All clinical sample collections were performed after obtaining informed consent from the corresponding patients and were approved by the ethics committees of Zhongshan Hospital. Tissue samples were sectioned into small pieces of approximately 2 × 2 × 2 mm³, and then were transplanted subcutaneously on the right dorsal flank of male 6-week-old NOD-SCID mice (Shanghai Lingchang Laboratory). Patients with different Lauren classification without preoperative chemoradiotherapy were selected to establish patient-derived xenograft (PDX) models, and xenografts were carried out within 3 to 5 hours following gastrectomy. A similar operation was performed to transfer the tumor piece to the next generation. After reaching generation 3 (P3), tumors were implanted subcutaneously on the right dorsal flank of 6-week-old NOD-SCID mice to do the in vivo experiment. The tumor volume was estimated as (D × d²)/2 (D, large diameter; d, small diameter). When the tumor volume reached 100 to 150 mm³, FAK inhibitor (VS-4718, 50 mg/kg) or same volume of vehicle was administrated by intragastric injection once a day. Tumor volume and body weight were measured three times weekly. After treatment for approximately 4 weeks, all the mice were sacrificed by cervical dislocation after CO₂ inhalation, and the tumors were extracted, weighed, embedded in paraffin, and sectioned.

Cell migration and invasion were measured by Transwell chambers (8-μm pores, Corning Life Sciences). Cells were transfected with indicated siRNA or plasmids before the migration and/or invasion assay. Afterwards, 1.5 × 10⁵ cells were suspended in 100-μL serum-free medium and added on the membrane while DMEM containing 10% FBS (600 μL) was added into the bottom chamber. Cells incubated for 16 or 24 hours were fixed in 0.4% paraformaldehyde and stained with 0.1% crystal violet, then photographed by an optical microscope, and were then quantified in four independent fields in each sample. Data are shown as mean ± SD of three independent experiments.

RNA interference and plasmid transfection were performed with Lipofectamine 2000 (Invitrogen) according to the manufacturer's recommendations. The siRNAs (siRNA; GenePharma) used to deplete EXOC4 [siRNA#1: 5′-AAUCGACUAGCCGAGUUGUTT-3′; siRNA#2: 5′- CGACAGCCAUUCGCACAUATT-3′. For EXOC4, EXOC3, EXOC5 stable knockdown, short hairpin RNAs (shRNA; sense: shEXOC4: 5′-CCGGAGAACCTGCTTTCATGCAACTCGAGTTGCATGAAAGCAGGTTCTTTTTTG-3′; shEXOC3: 5′-CCGGACAACGCCCTGAATGACGTCTCGAGACGTCATTCAGGGCGTTGTTTTTTG-3′; shEXOC5: 5′-CCGGGCTCAGAAATTGATGAAATACCTCGAGGTATTTCATCAATTTCTGAGCTTTTTG-3′] were cloned into pLKO.1 vector. Plasmid pEGFP-EXOC4 was purchased from Addgene (#53758). For stable EXOC4 overexpression, full-length coding sequence (CDS) of EXOC4 was inserted into pCDH-copGFP vector and lentiviruses were packaged in 293T cells. PcDNA3-FAK-HA was obtained from Dr. Zeguang Han's lab (Shanghai Jiao Tong University, China). In addition, different fragments of EXOC4 and FAK were generated by subcloning the corresponding cDNAs into the pcDNA3-Flag-vector and the pcDNA3-HA-vector, respectively. The pcDNA3-ITGA5–3×Flag, pcDNA3-ITGB1–3×Flag, pcDNA3-ITGAV-3×Flag, and pcDNA3-ITGB3–3×Flag plasmids were constructed by Youbio Company.

When cells grew to 80% to 90% density in 10-mm dishes, then the cells were washed and lysed in immunoprecipitation buffer (50-mmol/L Tris, pH 8.0, 120-mmol/L NaCl, 0.5% NP-40) mixed with phosphatase inhibitors (phosphatase inhibitor cocktail set I and II, Bimake) and protease inhibitors (Complete Mini, Roche). For immunoprecipitation assay, 1 mg cell lysates were incubated with 2 μg specific antibodies at 4°C for 2 hours, and then added with preequilibrated protein-A Sepharose beads (Sigma). After incubation overnight at 4°C with gentle rotation, the immunocomplexes were then washed four times with NETN buffer (20 mmol/L Tris, pH 8.0, 100 mmol/L NaCl, 1 mmol/L EDTA, and 0.5% NP-40). After that, the bound proteins were eluted and subjected to Western blotting (WB). For WB assay, total proteins extracted from samples were separated by SDS-PAGE and transferred onto polyvinylidene difluoride membranes. The membranes were then incubated with specific antibodies, followed by secondary antibody and detected by enhanced chemiluminescence.

Total RNA was isolated using RNA Isolater Total RNA Extraction Reagent (Vazyme, R401–01), and reverse transcribed with HiScript II Q Select RT SuperMix (Vazyme, R223–01). Real-time PCR was performed using ChamQ Universal SYBR qPCR Master Mix (Vazyme, Q711–02) with Roche LightCycler96 system.

MKN45 cells were seeded in a 10-cm dish and transfected with siNC or siEXOC4#1 for 72 hours. Cells were lysed and 1 mg total protein was used to detect phosphorylation alteration by the Human Phospho-Kinase Antibody Array (#ARY003B, R&D Systems) following the manufacturer's instructions.

Cells were treated as described previously and then were seeded on the glass coverslips, which were pretreated with Poly-l-Lysine for the attachment of cell clusters. After attaching to the coverslips, cells were fixed in 0.4% paraformaldehyde and incubated with indicated antibodies according to the standard protocols. Fluorescence images were taken using a fluorescent confocal microscope (Nikon A1Si).

Statistical analysis was performed with SPSS (RRID:SCR_002865, version 19). Evaluation of in vitro and in vivo experiments was calculated using Student t test or one-way ANOVA. Kaplan–Meier survival curve and the log-rank test was used for survival analysis. All statistical analyses were two-sided, and P < 0.05 (*), P < 0.01 (**), and P < 0.001 (***), were considered statistically significant. Benjamini–Hochberg procedure was used to test for FDR.

Tissue samples and clinicopathologic data were obtained from Zhongshan Hospital, Fudan University. Written informed consent from each patient was obtained. The research was approved by the Research Ethics Committee of Fudan University and was performed in accordance with the 1964 Declaration of Helsinki and its later amendments.

The results shown here are in whole or part based upon data generated by The Cancer Genome Atlas (TCGA) Research Network (https://www.cancer.gov/tcga), Asian Cancer Research Group (ACRG) database (https://consortiapedia.fastercures.org/consortia/acrg/), and online Kaplan–Meier Plotter (http://www.kmplot.com/analysis/index.php?p=service&cancer=gastric). The proteomic data has been uploaded to the NODE: OEP003074. All the data supporting the findings of this study are available from the corresponding authors upon reasonable request.

To identify the key effectors that distinguish recurrent and nonrecurrent DGC, we initially analyzed 100 cases of DGC deposited in the tumor tissue bank of Zhongshan Hospital. After strict quality control of the tissue samples and accurate follow-up results, 13 DGC samples, including 9 samples from nonrecurrent patients and 4 samples from patients with recurrence within 3 years after surgery, were selected for proteomic profiling (Supplementary Table S2). Among the 2,417 proteins examined, 131 differentially expressed proteins (P < 0.05 and fold change >1.2, or P < 0.05 and fold change <0.8) were identified (Fig. 1A and B). Among these proteins, 103 proteins were upregulated (P < 0.05 and fold change >1.2) and 28 were downregulated (P < 0.05 and fold change <0.8) in DGC with recurrence compared with DGC without recurrence (Fig. 1A and B). We then submitted these altered proteins for KEGG-based enrichment analysis and found that among these 131 proteins, 29 proteins were involved in the tight junction pathway (P = 0.016; FDR q-value = 0.195) and showed the most significant changes (Fig. 1C). Because the tight junction pathway is closely related to cell adhesion, the permeability of the paracellular barrier, and tumor metastasis (27), we selected these 29 proteins for further expression analysis. We found that EXOC4 was the most upregulated protein in the recurrent DGC samples (Fig. 1D). In further support of this observation, we examined the TCGA database and found that DGC recurrence was more likely to occur in patients with high EXOC4 expression (n = 19) than in those with low EXOC4 expression (n = 12, P = 0.032; Fig. 1E). However, in IGC, EXOC4 expression was not indicative of tumor recurrence (P = 0.425; Fig. 1E). Other subunits of the exocyst (EXOC1, EXOC2, EXOC3, EXOC5, EXOC6, EXOC7, and EXOC8) were also analyzed in these datasets but were not found to be correlated with tumor recurrence in either DGC or IGC (Supplementary Fig. S1SA–S1G).

Therefore, these data suggest that EXOC4 is a specific biomarker of exocysts that indicates recurrence of DGC.

To further evaluate the expression of EXOC4 in human gastric cancer, we first analyzed TCGA and ACRG databases, and found that EXOC4 expression was significantly elevated in tumor tissues (n = 233 in TCGA, and n = 300 in ACRG) compared with normal tissues (n = 21 in TCGA, and n = 100 in ACRG, P < 0.001; Supplementary Fig. S2A). In addition, we evaluated a large cohort of clinical samples from Zhongshan Hospital. The IHC results showed that EXOC4 protein expression was significantly increased in gastric tumors (n = 444, P < 0.001; Supplementary Fig. S2B). More interestingly, compared with IGC samples (n = 251 in Zhongshan cohort, and n = 69 in TCGA), EXOC4 expression was remarkably higher in DGC samples (n = 127 in Zhongshan cohort, and n = 35 in TCGA, P < 0.001; Fig. 2A and B). Meanwhile, other subunits of the exocyst did not show significant differences between DGC and IGC tissues, although most of these subunits demonstrated increased expression in whole tumor tissues compared with normal tissues (Supplementary Figs. S2C and S2D and S3A–S3G). Consistently, EXOC4 expression was much higher in DGC cell lines than in normal gastric epithelium GES-1 cells and IGC cell lines (Fig. 2C).

Moreover, we found that EXOC4 abundance was positively correlated with tumor infiltration (P = 0.046), lymph node metastasis (P = 0.016), and lymphatic vascular invasion (LVI; P = 0.018) in patients with DGC but not in patients with IGC (Fig. 2D–F; Supplementary Table S3). Specifically, the expression of EXOC4 in metastatic lymph nodes was dramatically higher than that in primary tumors (P < 0.001), indicating that tumor cells with high expression of EXOC4 might be prone to metastasis (Fig. 2G). Kaplan–Meier survival analyses demonstrated that patients with DGC with high EXOC4 expression (n = 102 in Zhongshan cohort, and n = 65 in GEO datasets) displayed lower overall survival than those with low EXOC4 expression (n = 25 in Zhongshan cohort, and n = 65 in GEO datasets, P = 0.007 and P = 0.032 respectively; Fig. 2H and I). However, survival differences associated with EXOC4 expression were not observed in patients with IGC (P = 0.2457; Fig. 2H and I). This prognostic specificity for DGC was not observed in other exocyst subunits (Supplementary Fig. S4A–S4G).

To analyze the biological function of EXOC4 in gastric cancer, we depleted EXOC4 in the DGC cell lines MKN45 and SNU668 with siRNA or short hairpin RNA, respectively. As a result, reduced cell migration and invasion were observed in EXOC4-depleted DGC cells (Fig. 3A and B; Supplementary Fig. S5A and S5B). To further assess the physiologic roles of EXOC4 in regulating DGC cell metastasis, we engineered a distant metastasis mouse model by generating stable shScramble- and shEXOC4-luciferase MKN45 cell lines. The results showed that EXOC4 depletion could significantly decrease cancer cell lung metastasis (Fig. 3C). In contrast, EXOC4 overexpression in MKN45 and SNU668 cells significantly increased cell migration and invasion in vitro, as well as tumor lung metastasis in vivo (n = 5, P < 0.05; (Fig. 3D–F; Supplementary Fig. S5C).

To identify which domain of EXOC4 might contribute to the regulation of DGC cell metastasis, we constructed a series of truncated mutants of EXOC4 (Fig. 3G). The N-terminus (aa 1–142) containing a coiled-coil domain is implicated in protein interactions with other members of the exocyst complex (28). The middle region (aa 143–402) is conserved among other members of the exocyst complex and termed the “exocyst complex component”. The C-terminal domain (aa 403–973) is an EXOC4-specific domain that was previously reported to be the PDZ-binding domain and to interact with SAP102, which is involved in the delivery of NMDARs to the cell surface in heterologous cells and neurons (29). In our study, we found that only reintroduction of full-length EXOC4, but not other truncated mutants, could fully rescue the decreased cell migration and invasion of EXOC4-depleted cells (Fig. 3H and I), indicating that full-length EXOC4 was necessary for the regulation of DGC metastasis.

To explore the potential underlying mechanism by which EXOC4 promotes DGC metastasis, we sought to determine the physiologic roles of EXOC4 in DGC. EXOC4 was previously reported to regulate cell migration by controlling ERK and p38 signaling in human oral squamous cell carcinoma (30). Therefore, we sought to determine which protein phosphorylation process or signaling pathway was affected by EXOC4 in DGC. On this premise, a human phosphokinase antibody array (catalog no. ARY003B, R&D Systems) was employed, and alterations in 43 phosphokinase sites that were likely crucial in the regulation of various cell processes, including cell migration and invasion, were evaluated. Among these phosphokinase sites, phospho–FAK-Tyr397 was the most remarkably changed in EXOC4-depleted MKN45 cells (Fig. 4A and B).

FAK is a well-established focal adhesion–associated kinase involved in cellular adhesion and spreading processes, and the phosphorylation of FAK-Tyr397 is largely involved in FAK activation, cell migration, and metastasis in many cancers (31, 32). To further verify the effect of EXOC4 expression on FAK phosphorylation in DGC cells, we measured the phosphorylation of FAK^Y397 and observed that this event but not two other phosphorylation sites, FAK^Y576/577 and FAK^Y925, was significantly reduced upon EXOC4 depletion in both MKN45 and SNU668 cells (Fig. 4C; Supplementary Fig. S6A). In contrast, p-FAK^Y397 levels notably increased under EXOC4-overexpressing conditions (Fig. 4D; Supplementary Fig. S6B). However, depletion of EXOC3 or EXOC5, two other exocytosis-related proteins frequently involved in tumor progression (13, 14, 33), did not reduce FAK^Y397, FAK^Y576/577, or FAK^Y925 phosphorylation in DGC cells (Supplementary Fig. S6C). Together, these findings strengthened the significance of EXOC4 in the regulation of the FAK signaling pathway.

Given that full-length EXOC4, but not the N-terminus of EXOC4, is involved in DGC malignancy, we hypothesize that more than the exocytosis-dependent roles of EXOC4 are involved in its pathologic functions. As such, we then conducted co-immunoprecipitation (co-IP) assays in DGC cells and 293T cells, and observed that FAK could bind to EXOC4 but not to other exocyst subunits, where EGFR and integrin were used as positive controls for binding with FAK (Fig. 4E and F; Supplementary Fig. S6D and S6E). In addition, we constructed different HA-tagged FAK protein fragments corresponding to the N-terminal domain (aa 1–415), kinase domain (aa 416–676), and C-terminal domain (aa 677–1090) of FAK. We found that FAK bound to the C-terminus of EXOC4 (aa 403–973), while EXOC4 bound to the N-terminal domain of FAK (aa 1–415), where FAK interacted with membrane receptors (Fig. 4G and H). Furthermore, we also examined FAK^Y397 phosphorylation levels in different DGC cell lines and found that p-FAK^Y397 was higher in DGC cells than in normal gastric epithelial cells and IGC cells (Fig. 4I; ref. 34). IHC staining also demonstrated that the expression of p-FAK had a strong correlation with EXOC4 in DGC (n = 127, r = 0.716, P < 0.001; Supplementary Fig. S7A and S7B).

To address whether EXOC4 promotes tumor metastasis by regulating FAK phosphorylation and activation in DGC cells, we introduced the wild-type (WT) form, or a phosphorylation-deficient mutant (Y397F) form of FAK, into FAK-depleted MKN45 cells and found that WT FAK but not Y397F FAK could promote MKN45 cell migration and invasion (Fig. 5A–C). Conversely, we introduced FAK-Y397E, a phosphomimetic mutation that retains sufficient biological activity dependent on upstream molecular activation, into EXOC4-depleted MKN45 cells and observed that FAK^Y397E but not FAK^Y397F could at least partially overcome the EXOC4 depletion–induced reductions in DGC cell migration and invasion (Fig. 5D and E).

Next, we used the FAK kinase inhibitor VS-4718, a novel selective FAK tyrosine kinase inhibitor, that has been evaluated as a cancer therapy in clinical trials (35, 36), to treat EXOC4-overexpressing DGC cells. The results showed that VS-4718 remarkably reduced the increased cell migration and invasion mediated by EXOC4 overexpression in MKN45 and SNU668 cells (Fig. 5F–H; Supplementary Fig. S8A and S8B). Furthermore, VS-4718 significantly suppressed MKN45 cell lung metastasis mediated by EXOC4 overexpression in a nude mouse model (n = 6, P < 0.001; Fig. 5I and J; Supplementary Fig. S8C). These findings indicate that FAK may be a potential therapeutic target in patients with DGC with high EXOC4 expression.

As a component of the exocyst, EXOC4 can assist in the docking of secretory vesicles to the plasma membrane and promote the release of secretory proteins by vesicles (11, 13). Therefore, we intended to determine whether EXOC4 affected the secretion or activation of FAK upstream factors, such as integrin α5/β1 and EGF, which are directly related to FAK phosphorylation (35, 37), in DGC cells. We first examined the concentrations of integrin α5 and EGF in the serum of patients with DGC by ELISA. We found that both serum integrin α5 and EGF levels were remarkably higher in patients with DGC with lymph node metastasis (N+, n = 99) than in patients without metastasis (N0, n = 28, P < 0.001 and P = 0.0094 respectively; Fig. 6A). More interestingly, we observed that the expression level of EXOC4 was positively correlated with the serum concentrations of integrin α5 and EGF in these DGC samples (P < 0.001; Fig. 6B). To investigate the secretory roles of EXOC4, we enriched vesicles derived from MKN45 cells and 293T cells by density gradient centrifugation and found that EXOC4 could promote the secretion of vesicles carrying both integrin α5/β1 and EGF (Fig. 6C–E). Consistent with these findings, FAK signaling, including the phosphorylation of FAK^Y397 and ERK^T202/Y204, was significantly activated in cells cultured with EXOC4-overexpressing cell conditioned medium; serum-free medium was used as a negative control, and 10% FBS medium was used as a positive control (Fig. 6F).

It is well established that FAK is recruited to sites of integrin clustering and interacts with integrin-associated proteins through its C-terminal domain, leading to FAK activation (38, 39). In addition, FAK can be stimulated by EGF and interacts with activated EGFR through its N-terminal domain, which has an important function in promoting EGF-stimulated cell migration (35, 38). Because EXOC4 could directly bind with FAK (Fig. 4), we sought to determine whether EXOC4 modulated the interactions of FAK and upstream signaling effectors directly. Interestingly, our co-IP results showed that the interactions of FAK with integrin/EGFR were significantly enhanced upon EXOC4 overexpression and could not be abolished by the EGFR inhibitor gefitinib or the anti-integrin peptide ATN-161 (Fig. 6G–I). Furthermore, we found that the integrin peptide antagonist ATN-161 could efficiently overcome EXOC4 overexpression–induced MKN45 cell migration and invasion (Fig. 6J and K; Supplementary Fig. S8D). On the other hand, the activated FAK signaling, increased cell migration, and invasion triggered by EGF could be markedly repressed by depleting EXOC4 in MKN45 cells (Fig. 6L and M; Supplementary Fig. S8E).

Together, these data indicate that EXOC4 can positively regulate FAK signaling by both stimulating integrin/EGF secretion and enhancing the interactions of FAK with its upstream effectors, highlighting the potential of a strategy of targeting FAK signaling to combat EXOC4 overexpression–mediated DGC metastasis.

We next tested the therapeutic potential of the FAK inhibitor VS-4718 in different PDX models derived from human gastric cancers. Six patients with different Lauren classifications without preoperative chemoradiotherapy were selected to establish PDX models, of which two samples were DGC (GAS190153 and GAS190233), two samples were mixed-type gastric cancer (GAS190247 and GAS190282), and two samples were IGC (GAS190216 and GAS190271). We verified the expression of EXOC4 and p-FAK^Y397 in the gastric cancer samples (Fig. 7A and B; Supplementary Fig. S9A) and then treated the PDX models derived from these samples with VS-4718. Approximately 3 to 4 weeks later, we found that the VS-4718 could markedly reduce tumor growth in the four EXOC4^high PDX models (P < 0.01) but had no significant effect on the EXOC4^low IGC xenograft models (Fig. 7C; Supplementary Fig. S9B–S9E). Accordingly, the expression of the proliferation marker Ki67 was significantly reduced in the PDX models with high EXOC4 expression when treated with VS-4718 (Fig. 7D); however, a similar result was not observed in IGC with low EXOC4 expression (Fig. 7E). These results suggest that FAK inhibitors may have a therapeutic effect on patients with human gastric cancer with high EXOC4 expression.

As a complex and highly heterogeneous disease, gastric cancer has been grouped by different biological and clinical features with unclear molecular mechanisms (5, 40). With the rapid development of multiomic technology, unidentified molecular factors in DGC are being revealed, including their impacts on clinical outcomes (41). Recent research results from TCGA and the ACRG have shown that DGC is predominantly found within the genomically stable group and microsatellite stable group with an overall poor prognosis, characterized by mutations in CDH1, TP53, RHOA, CTNN1A, CMTM2, and CLDN18-ARHGAP fusion (42). Conceivably, genomic alterations in adhesion and motility proteins in DGC may also contribute to its infiltrative and poorly cohesive histology. Zhang and colleagues recently reported that RHOA^Y42C with inactivation of the canonical tumor suppressor Cdh1 induced metastatic DGC in a mouse model (34). The authors revealed how the RHOA^Y42C mutation mediates FAK activation and imparts sensitivity to pharmacologic FAK inhibitors in DGC (34). In this study, by using a proteomics-based approach, we found that the expression of EXOC4, a component of the exocyst, was highly increased in DGC samples with tumor recurrence compared with nonrecurrent samples. Analysis of the TCGA database also validated the association of high EXOC4 expression with DGC recurrence, which indicated the value of EXOC4 as a potential predictive marker for DGC recurrence in patients after surgery.

The exocyst is a conserved octameric protein complex that tethers secretory vesicles during exocytosis and is implicated in multiple cellular processes, such as secretion, membrane transduction, cytokinesis, autophagy, and ciliogenesis (11, 15). Under pathologic conditions and specifically in the gastric cancer setting, the role of the exocyst has not been investigated previously. In this study, we showed that EXOC4, but not other components of the exocyst, played specific and significant roles in DGC metastasis and recurrence. More significantly, we first demonstrated that EXOC4 could stimulate the activation of FAK in DGC cells via the role of exocyst or independent of the holo–exocyst complex. On one hand, EXOC4 could directly bind with FAK, assisting FAK to move from cytoplasm to cell membrane and enhancing its interactions with the upstream integrin α5/β1 receptor and EGFR (Supplementary Fig. S10). On the other hand, EXOC4 could stimulate the secretion of FAK upstream effectors such as integrin α5/β1 and EGF in an exocyst-dependent manner and this correlated with elevated serum levels of integrin α5/β1/EGF in patients with DGC. Through these two different processes, EXOC4 markedly facilitates FAK phosphorylation at Tyr397, enhancing FAK kinase activity and contributing to DGC cell invasion, migration, and metastasis (Supplementary Fig. S10).

To validate our findings, we also employed FAK inhibitors and integrin antagonist peptides, which significantly repressed EXOC4-mediated DGC malignancy in vitro and in vivo. More significantly, we evaluated the therapeutic effect of the FAK inhibitor VS-4718 in PDX models derived from human gastric cancers. The results highlighted the potential of a strategy targeting FAK for high EXOC4 expression DGC cancer therapies. Although the result was promising, it also indicated that FAK inhibitor might need combination therapy as all of the PDXs continued to grow, albeit less than control, while on VS-4718. For one reason, EXOC4 may have other biological function to regulate tumor growth and metastasis in DGC, which need to be further explored. For another reason, gastric cancer is recognized as a complex and multifactorial disease that is associated with a wide range of molecular mechanism. As such, except for FAK, other genes or signaling may contribute to tumorigenesis of DGC.

There are certainly limitations of our study. First, the number of patients for proteomic screening is small, so there could be selection bias here, and the different recurrence pattern has not been compared. Moreover, how EXOC4 is upregulated in gastric cancer, especially in DGC, is not clarified. Then, the relationship between EXOC4 and several well-known DGC-associated genes, like CDH1, and RHOA has yet to uncover. Next, this study focuses on the function of EXOC4 itself. However, as a component of exocyst, how exocyst contributes to metastasis of gastric cancer has not been discovered. Finally, the inhibitor against EXOC4 is not developed to demonstrate the therapeutic effect targeting EXOC4 in DGC, which will be further explored in our future direction.

In conclusion, our findings not only identify a molecular mechanism for DGC recurrence via EXOC4–integrin α5/β1/EGF–FAK signaling, but also provide a promising target for the treatment of patients with DGC with high EXOC4 expression.

No disclosures were reported.

H. Li: Data curation, formal analysis, writing–original draft. X. Fu: Data curation, software, formal analysis, investigation, methodology, writing–original draft. J. Zhao: Data curation, formal analysis, methodology, writing–original draft. C. Li: Resources, software, formal analysis. L. Li: Software, validation, methodology. P. Xia: Software, formal analysis. J. Guo: Project administration, writing–review and editing. W. Wei: Visualization, methodology. R. Zeng: Resources, data curation, software, supervision, methodology. J. Wu: Conceptualization, resources, data curation, software. Y. Sun: Conceptualization, funding acquisition, writing–review and editing. L. Huang: Resources, data curation, funding acquisition, writing–review and editing. X. Wang: Conceptualization, resources, validation.

We thank Dr. Rongkui Luo (Department of Pathology, Zhongshan Hospital, Fudan University, Shanghai, China) for his kind help in evaluation, IHC staining, and Lauren classification of tissue microarrays. This work was supported in part by the National Natural Science Foundation of China (to Y. Sun; 81872425; to X. Wang 81972228; and L. Huang, 81872346), and Shanghai Pu Jiang Talents plan (to X. Wang, 2019PJD005).

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.