The Roles of Beclin 1 Expression in Gastric Cancer: A Marker for Carcinogenesis, Aggressive Behaviors and Favorable Prognosis, and a Target of Gene Therapy

Beclin 1 is encoded by Becn1, and plays a role in tumorigenesis, neurodegeneration, apoptosis and autophagy. Here, the aggressive phenotypes and relevant proteins were examined after Beclin 1 expression was altered in gastric cancer cells. We also observed the effects of Beclin 1 on gastric carcinogenesis using Becn1 knockout mice. Finally, clinicopathological significances of Beclin 1 expression were analyzed using meta- and bioinformatics analyses. Becn1 overexpression was found to inhibit proliferation, glucose metabolism, migration and invasion of gastric cancer cells, whereas its knockdown caused the opposite effects. Beclin 1 suppressed the tumor growth by decreasing proliferation and increasing apoptosis. The heterozygous abrogation of Becn1 in gastric pit, parietal and chief cells could not cause any epithelial lesion. Beclin 1-mediated chemoresistance was closely linked to the autophagy, Bax underexpression, and the overexpression of Bcl-2, LRP1, MDR1, and ING5. Bioinformatics analysis showed higher Becn1 mRNA expression in intestinal- than diffuse-type carcinomas (P<0.05), and in male than female gastric cancer patients (P<0.05). Becn1 hyperexpression was positively associated with both overall and progression-free survival rates of the cancer patients (P<0.05). Meta-analysis showed that down-regulated Beclin 1 expression in gastric cancer was positively with lymph node metastasis, TNM staging, dedifferentiation and poor prognosis (P<0.05). Becn1-related signal pathways in gastric cancer included prostate, lung, renal, colorectal, endometrial and thyroid cancers, glioma, and leukemia, the metabolism of amino acid, lipid and sugar, and some signal pathways of insulin, MAPK, TRL, VEGF, JAK-STAT, chemokine, p53, lysosome, peroxidome and ubiquitin-mediated protein degradation (P<0.05). These suggested that Beclin 1 might be considered as a potential marker of gastric carcinogenesis, aggressiveness and prognostic prediction, and as a target of gene therapy in gastric cancer.


INTRODUCTION
There are a worldwide decrease in morbidity and mortality and rapid development in diagnostic and operative techniques, but gastric cancer still challenges the human health. The clinical outcome and prognosis of gastric cancer are generally worse because it has often metastasized by the time of discovery and most cancer patients are elderly at presentation (1,2). Therefore, deep clarification of molecular mechanisms about gastric carcinogenesis and subsequent progression may remarkably improve early finding, diagnosis and treatment.
Previously, Beclin 1 overexpression in ovarian cancer was reported to negatively correlate with the differentiation and higher cumulative and relapse-free survival rates (16). Either Beclin 1 mRNA or protein hyperexpression was seen in colorectal and gastric cancers in comparison to precancerous lesion or mucosa. Beclin 1 expression was inversely associated with liver metastasis and distant metastasis of colorectal cancer, or venous invasion, lymph node metastasis, TNM staging, dedifferentiation and favorable prognosis of gastric cancer (17,18). In colorectal cancer cells, Beclin 1 inhibited cell viability, migration and invasion, lamellipodia formation, and tumor growth, induced autophagy and apoptosis (19). Reportedly, Beclin 1 overexpression augmented the apoptotic induction of cis-diamminedichloroplatinum via enhancing Caspase-9 activity in gastric cancer cells (18). Gao et al. (20) reported that radiotherapy also induced autophagy and increased Beclin-1 expression via p53 pathway. Biallelic loss of Becn1 is embryonically lethal for knockout mice, and promotes spontaneous tumorigenesis of lymphomas, liver and lung cancers (21)(22)(23)(24). Here, we analyzed the effects of Beclin 1 expression on the aggressive behaviors and phenotypes of gastric cancer cells, and clarified relevant mechanisms. To explore the role of Becn1 knockout in gastric carcinogenesis, we established conditional Becn1 knockout mice in gastric pit, parietal and chief cells using the Capn8, Atp4b and PGC promoter to initiate Cre recombination respectively. Finally, clinicopathological significances were analyzed using meta-and bioinformatics analyses.

Cell Line and Culture
Gastric cancer cell lines (BGC-823 and MKN28) were kindly presented by Prof. Su of Jinzhou Medical University. These cells were grown in RPMI 1640 medium containing FBS, penicillin, and streptomycin in 5% CO 2 at 37°C. BGC-823 and MKN28 cells were subjected to pcDNA3.1-Becn1 or pcDNA3.1 vector transfection, and selected by G418 with monoclone collection. The target sequences of sh-Becn1 were 5'-CACCGGAATGGA ATGAGATTAATGCTTCAAGAGAGCATTAATCTCATT

CCATTCCTTTTTTG-3' and 3' -CCTTACTTACTCT A A T T A C G A A G T T C T C T C G T A A T T A G A G T A A
GGTAAGGAAAAAACCTAG-5'. Additionally, we treated cells with MG132 (proteasome inhibitor), paclitaxel (mitotic inhibitor), or SAHA (histone deacetylase inhibitor).

Proliferation Assay
We used cell counting kit-8 (CCK-8) to determine cell proliferation. In brief, 2.0 × 10 3 cells/well were cultured on 96well plate. After adhering to plate.10 ml of CCK-8 solution was added into each well of the plate at different time points, and the absorbance was measured at 450 nm after 3 h incubation.

Cell Cycle Analysis
We trypsinized, collected, and fixed the cell using ethanol for 2 h. After RNase treatment for 1 h, cells were pelleted and stained by propidium iodide (PI) for 30 min. At final, flow cytometry was used to examine PI signal.

Apoptosis Assay
FITC-labeled annexin V staining (Kagen, China) was employed to indicate phosphatidylserine externalization of early apoptosis in terms of manufacturer's instructions.

Measurement of Extracellular Acidification Rate and Oxygen Consumption Rate
We measured the metabolic parameters using wing discs of abxUbxFLPase of Seahorse metabolic flux analyzer. Discs were measured in bicarbonate-free Schneider medium with 12 mM glutamine and 11 mM glucose. The glycolysis can be indicated by extracellular acidification rate (ECAR), and the mitochrondrial respiration by oxygen consumption rate (OCR). We also normalized ECAR and OCR values using protein amount according to BCA assay.

Transwell Chamber Assay
To assess invasion, we cultured 2.0 × 10 5 cells in FBS-free medium in the matrigel-coated insert on the chamber top, and added 10%-FBS-containing medium to the chamber bottom as a chemoattractant. After 24-h incubation, the upper member of insert was scrubbed, and the lower portion fixed in methanol and followed by Giemsa dye. To measure migration, we repeated the above-mentioned protocol excluding membrane-control insert.

Animals
We housed three mice per plastic cage with paper chips, standard rodent food and water in pathogen-free and temperaturecontrolled condition with a 12-h light/dark illumination cycle. All animal experiments were conducted using protocols approved by the Committee on Animal Experimentation of The Affiliated Hospital of Chengde Medical University. Female 8-week Balb/c nude mice were used for subcutaneous implantation by injection of 2.0× 10 6 cells per mouse to axilla (n=10 mice/group). Finally, the mice were anesthetized, photographed, and killed for tissue sampling. The part of tumors was subjected to routine pathological block preparation, and the remaining was stored in liquid nitrogen until protein, RNA or protein extract. The tumor volume was calculated as width 2 × length ×0.52.
Additionally, we performed cre-mediated deletion of floxed alleles by mating Becn1 conditional mutant mice (Mouse Biology Program, University of California, Davis USA) with Capn8-cre, Atp4b-cre (kindly presented by Prof. Yang) or PGC-cre (prepared in our Lab., unpublished) transgenic mice. At least 5 mice were killed at 9 month, and their stomachs were analyzed.

Real-Time RT-PCR
We extracted total RNA from cells or tissues using Trizol (Takara). Reverse transcription of 1 µg RNA was performed using random primers and AMV reverse transcriptase. PCR primers were designed according to the sequences in GenBank. Oligonucleotide primers for PCR were 5'-TTACCACAGC CCAGGCGA-3' and 5'-GCCACCATCAGATGCCTC-3' for mouse Becn1, and 5'-ACATACTCAG CACCGGCCTC-3' and 5'-TATGACTCCACTCACGGCAAA-3' for mouse GAPDH. SYBR Premix Ex Taq II kit (Takara) was employed to amplify target cDNAs using GAPDH as an internal control.

Western Blot
Protein was extracted in RIPA lysis buffer and determined by BCA assay. We separated denatured protein in SDSpolyacrylamide gel and transferred to Hybond membrane. After blocked in 5% skim milk, the membranes were incubated with primary antibodies ( Table 1) and then with anti-rabbit, or anti-mouse IgG conjugated to horseradish peroxidase (Dako). We visualized bands using ECL-Plus detection reagents.

Histological Analysis
Tissues block were sectioned and consecutive 5mm-thick sections were stained with HE, alcian blue, PAS and HID. Sections were deparaffinized, dehydrated, and subjected to immunostaining using primary antibodies (Table 1) as previously described (17), or TUNEL using ApopTag Plus Peroxidase In Situ Apoptosis Detection Kit (Millipore) as reported (19).

Identification of Eligible Studies
We searched the articles using BIOSIS, PubMed, SciFinder, and Web of Science until July 10, 2020. The following strategy was (Becn1 OR Beclin 1) AND (stomach OR gastric) AND (adenocarcinoma OR carcinoma OR cancer). The articles were included to observe Beclin 1 immunostaining in gastric cancers and compare Beclin 1 expression with their pathobiological features and prognosis. We excluded the abstract, comment, review and meeting, and papers about Western blot, RT-PCR, cDNA chip, or transcriptomic data for Beclin 1 expression.

Data Extraction and Quality Score Assessment
Both reviewers independently collected useful information from all eligible articles, including first author, publication year,

Name
Source Company Rabbit Abcam

Bioinformatics Analysis
The transcriptomic and clinicopathological data of gastric cancer patients were downloaded from TCGA database by TCGA-assembler of R software. We analyzed Becn1 mRNA level between gastric normal and cancer tissues using Oncomine and TACGA data. Becn1 expression was compared with clinicopathological and prognostic data of gastric cancer patients. GSEA was performed with GSEA-3.0. We used Becn1 level as a phenotype label and analyzed pathway enrichment. Kaplan-Meier plotter was employed to analyze the prognostic significance of Becn1 mRNA.

Statistics Analysis
We used Revman software 5.3 to carry out Meta-analysis, and evaluated HWE using Chi-square test. If there was no significant heterogeneity, the fixed effect model would be employed. Experimental and TCGA data was handled with SPSS 10.0 software using t test or survival analysis. We regarded P<0.05 as statistically significant.

RESULTS
The Effects of Beclin 1 Overexpression on the Aggressiveness of Gastric Cancer Cells No alteration was observed in the expression of GST-p, NF-kB and CD147 ( Figure 2H).
Moreover, Beclin 1 overexpression suppressed tumor growth of gastric cancer cells in nude mice ( Figure 3A, P<0.05). There was a higher Beclin 1 expression, a stronger signal of TUNEL and a weaker ki-67 expression in tumors of Becn1-overexpressing BCG-823 ( Figures 3B, C). Beclin 1 down-regulated the expression of VEGF, E-cadherin, Bcl-2, but up-regulated the expression of LC-3B, NF-kB, PI3K, Akt1/2/3 and MDR in xenograft tumor of gastric cancer cells ( Figure 3C). Moreover, we matched the conditional Becn1-knockout (KO) mice with Atp4b-cre, Capn8-cre or PGC-cre mice, and designed the primers to confirm the monoallelic deletion of Becn1 using DNA from tail and gastric mucosa ( Figure 4A). According to PCR results, we obtained the monoallelic deletion of Becn1 in gastric mucosa in Becn1-PGC-cre, Becn1-Capn8-cre and Becn1-Atp4b-cre mice ( Figure 4B). There was a weaker expression of Becn1 in gastric epithelium of Becn1-PGC-cre, Becn1-Capn8-cre and Becn1-Atp4bcre than those of wild-type (WT) mice at both mRNA and protein levels ( Figure 4C, D). However, no remarkable lesions were observed in the gastric mucosa of these three kinds of mice although weaker Beclin 1 was immunohistochemically observed in gastric chief, pit and parietal cells of Becn1-PGC-cre, Becn1-Capn8-cre, and Becn1-Atp4b-cre than those of WT mice ( Figures 4E-G). Additionally, there were no differences in sialic acid, neutral and sulfuric acid mucus between conditional KO and WT mice according to alcian blue, PAS and HID staining ( Figure 4H).

Association Between Beclin 1 Expression and Clinicopathological Parameters of Gastric Cancer
We collected 986 cancers and 786 controls and found downregulated Beclin 1 expression in gastric cancer ( Figure 5A, P=0.02). As shown in Figure 5B, there was no difference in Beclin 1 expression between age >60 year and <60 year cancer patients (P>0.05), between male and female ones ( Figure 5C, P>0.05) or between T 1-2 than T 3-4 gastric cancer ( Figure 5D, P=0.19). Beclin 1 expression was negatively associated with lymph node involvement ( Figure 5E, P=0.03), TNM staging ( Figure 5F, P=0.03), and dedifferentiation ( Figure 5G, P=0.01). Wellmoderately adenocarcinoma had a higher Beclin 1 expression than the poorly-differentiated subtype ( Figure 5H, P=0.0004). As indicated in Figure 5I, there was significant association between Beclin 1 expression and favorable overall survival in patients with gastric cancer (OR=1.34, 95% CI: 1.15-1.56, P=0.0002). Sensitivity analysis was performed to evaluate individual study's influence on the pooled results by deleting one single study (data not shown).

The Clinicopathological and Prognostic Significances of Becn1 Expression in Gastric Cancer
According to TCGA databases, Becn1 mRNA expression was found to be lower in gastric cancer than normal mucosa ( Figure  6A, P<0.05). There was a higher Becn1 mRNA expression in intestinal-than diffuse-type adenocarcinoma according to DErricodata ( Figure 6B, P<0.05). In TCGA data, Becn1 expression was higher in male than female cancer patients ( Figure 6C, P<0.05). According to Kaplan-Meier plotter, Becn1 expression was found to positively correlate with overall and progression-free survival rates of all cancer patients ( Figure  6D, P<0.05), even stratified by Lauren's classification, TNM staging and treatment (

Becn1-Related Signal Pathways in Gastric Cancer
As summarized in Table 4, we conducted a GSEA to analyze Becn1-related signal pathways in gastric cancer. The significant The transfectants showed a decrease in growth (B) and chemosensitivity to MG132, Paclitaxel and SAHA (C) in comparison with the control or mock. After Becn1 was overexpressed, the morphological appearance was observed under microscope (D). Flow cytometry showed that Beclin 1 expression induced the G 1 arrest of BGC-823 cells, but S arrest of MKN28 cells (E). Apoptosis was increased by Beclin 1 overexpression, as evidenced by Annexin V assay (F). Oxygen consumption and transwell assays were used to monitor the glucose metabolism, migration and invasion of gastric cancer cells respectively (G, H). The phenotype-related proteins were screened by Western blot (I). *P < 0.05, compared with the transfectants.
Here, Beclin 1 was demonstrated to suppress proliferation, migration, invasion and tumor growth, and induce cell cycle arrest, apoptosis, autophagy and chemoresistance in gastric cancer cells, whereas versa for Becn1-silencing cells, suggesting that Beclin 1 can be used as a gene therapy target of gastric cancer if its chemoresistant induction could be avoided or alleviated. Additionally, gastric lesions were not observed from monoallelic conditional KO mice of Becn1 in pit, parietal and chief cells    respectively although spontaneous lymphomas, liver and lung cancers were detected in Becn1-/+ mice (23). Reportedly, microadenoma, macroadenoma to invasive well-differentiated adenocarcinoma were rapidly observed in the gastric Lgr5+ stem cells with the double deletion of Smad4 and PTEN using Lgr5-Cre mice (46). Our findings support the opinion that Becn-1mediated gastric cancer might originate from local stem cells with genetic alteration, but not the differentiated cells.
Additionally, Becn1-induced gastric carcinogenesis possibly needs the exposure to chemical carcinogen. Bax can open the mitochondrial voltage-dependent anion channel during apoptosis, which is inhibited by interaction with Bcl-2 on the mitochondrial membrane (47). Consequently, Bax overexpression and Bcl-2 underexpression in gastric cancer cells may account for the inductive effect of Beclin 1 on apoptosis via mitochondrial pathway. Reportedly, activated p38 MAPK phosphorylates MAPKAP kinase 2 to phosphorylate the transcription factors (e.g. ATF2, Mac, and MEF2) and subsequently to mediate cell survival. PI3K/Akt activation results in drug resistance and cell proliferation. We observed p-p38 overexpression in Becn1-ovexpressing or silencing gastric cancer cells, but both p-PI3K and p-Akt hyperexpression was seen in Becn1-ovexpressing, but not Becn1-knockdown cells, suggesting that PI3K/Akt signal pathway might be involved in the effects of Beclin 1, but not p38. According to the literature, MDR1 and LRP1 are mainly responsible for drug resistance due to their ATPdependent efflux pumping (47). ING5 overexpression also caused chemoresistance in neuroblastoma, glioma, gastric cancer, lung cancer, ovarian cancer and breast cancer cells (48)(49)(50)(51)(52). Cellular stress or increased metabolic demand activates autophagy, which can cause therapeutic resistance (47). Our results hinted that the Beclin 1-mediated chemoresistance might be due to autophagy and the hyperexpression of MDR1, LRP1, and ING5. Becn1 overexpression was found to markedly attenuate the ability of gastric cancer cells to migrate and invade possibly due to VEGF hypoexpression because VEGF may promote mobility and proliferation of gastric cancer cells. In the study, E-cadherin hyperexpression, and Twist1 and Zeb2 hypoexpression in gastric Becn1 transfectants were closely linked to MET because Zeb2 and twist promote EMT process with E-cadherin underexpression. Therefore, the inhibitory of Beclin 1 on migration and invasion was dependent on MET in gastric cancer cells.
TGF-b1-induced autophagy linked b-catenin and Smad signaling to promote EMT of mouse kidney proximal tubular epithelial C1.1 (SV40 transformed) cells by disrupting Ecadherin/b-catenin-mediated cell-cell contact via ILK overexpression (53). SIRT1, SIRT6 and SPHK1 induced EMT  by up-regulating autophagy-linked lysosomal degradation of Ecadherin in melanoma and hepatocellular cells via Beclin 1-Ecadherin cascade respectively (54)(55)(56). FSTL1 was demonstrated to induce EMT and airway remodeling by activating autophagy in asthma (57). HMGB1 also induced apoptosis and EMT in association with autophagy through the upregulated expression of DDR1 and the downregulation of the phosphorylation of mTOR following H/R injury in H9c2 cells (58). Persistent hypoxia induced autophage disorders, which could cause down-regulated E-cadherin and down-regulated MMP-9, thus promoting invasiveness of placenta trophoblasts (59). GRIM-19 suppressed hypoxia-triggered invasion and EMT by inhibiting hypoxia-induced autophagy through inactivation HIF-1a/ STAT3 signaling axis (60). HIF-1a-mediated autophagy promoted EMT and metastatic ability of CD133+ pancreatic cancer stem-like cells during intermittent hypoxia (61). Endoplasmic reticulum stress induced EMT through autophagy via c-src kinase activation (62). In contrast, Catalano et al. (63) reported that autophagy induction impaired migration and invasion by reversing EMT in glioblastoma cells, in line with our inductive effects of Beclin 1 overexpression on E-cadherin expression in gastric cancer. In combination of these findings, we speculated that the regulatory correlation between Beclin 1 and E-cadherin depended on the cell types.
Reportedly, loss and down-regulation of Beclin 1 expression was immunohistochemically observed in esophageal adenocarcinoma (64), colorectal cancer (65), breast cancer (66), but versa in pancreatic cancer (67). Here, we performed meta-analysis and found that down-regulated Beclin 1 expression in gastric cancer was positively linked to lymph node metastasis, TNM staging, dedifferentiation and poor prognosis, in agreement with our bioinformatics findings. Reportedly, loss of Beclin-1 in cancer cells and Beclin 1 overexpression in stromal mesenchymal cells were closely linked to local recurrence and lymph node metastasis in breast cancer (68). Beclin-1 expression was related to HBV infection status and the grade of hepatocellular carcinoma (HCC) (69). In the hypoxic group, it was negatively correlated with high tumor grade, advanced stage, large size, and multifocal tumors of HCC, while positively with TNM stage and liver metastasis of gallbladder carcinoma (70), and with worse recurrence in intrahepatic cholangiocarcinoma (71). Beclin 1 expression was reported negatively correlate with tumor grade, lymph node involvement, TNM stage, tumor size, dedifferentiation, and recurrence of lung cancer (72). These findings indicated that its down-regulation contributed to gastric carcinogenesis and progression as a molecular marker.
Becn1 expression loss was demonstrated to act as a negative prognosticator in ovarian cancer patients receiving platinum-based chemotherapy (73). Tang et al. (74) found that low Becn1 expression was associated with poor prognosis in breast cancer as an independent predictor, but the converse was true for breast cancer patients receiving tamoxifen treatment (75). Beclin 1 was considered as a dependent factor for favorable prognosis of the colorectal cancer patients (65). It was the same for ovarian clear cell carcinomas (76) and multiple myeloma (77). In contrast, Beclin 1 was significantly correlated with short disease-free survival and overall survival of pancreatic ductal adenocarcinoma (67). Here, meta-or bioinformatics analysis showed that either Beclin 1 protein or mRNA expression was positively linked to the favorable prognosis of the patients with gastric cancer. Taken together, Beclin 1 might be considered as a potential marker for the prognosis of the gastric cancer patients at either mRNA or protein level.
Down-regulated Becn 1 expression in gastric cancer was positively correlated with aggressiveness and worse prognosis at either mRNA or protein level as a molecular marker. Becn1 KO in pit, parietal or chief cells couldn't induce gastric carcinogenesis. It might be employed as a potential target for gene therapy of gastric cancer patients if its chemoresistant induction would be avoided and alleviated.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study was reviewed and approved by The Affiliated Hospital of Chengde Medical University.

AUTHOR CONTRIBUTIONS
H-cZ, SZ, HX, E-hZ, H-mJ, and C-lH designed and carried out the experiments, and HZ wrote the draft. All authors read and approved the final manuscript. All authors contributed to the article and approved the submitted version.