This retrospective study was designed to investigate the clinical significance of ATG13 in gastric cancer. A dual-strategy approach was employed: initial bioinformatics analysis of publicly available datasets, followed by validation in an independent, well-characterized clinical cohort supplemented by functional experiments.

For the bioinformatic cohort, transcriptomic and genomic data were sourced from public repositories including Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov) and Genotype-Tissue Expression (GTEx, https://www.genome.gov) projects and providing a large sample size for exploratory analysis (16, 17).

The clinical validation cohort consisted of a tissue microarray (catalog no. HStmA180Su20; Outdo Biotech., Shanghai, China) constructed from formalin-fixed, paraffin-embedded tumor specimens of 99 patients diagnosed with primary stomach adenocarcinoma (STAD). These patients underwent curative resection between January 2011 and December 2012. Additionally, 71 paired adjacent non-cancerous tissue samples were included as controls. Among 99 patients, one case lost the follow-up information and was not included for the survival analysis. The median follow-up period for the cohort of 98 patients was 40 months (range, 2–71 months) from the date of surgery. During the follow-up period, 41.8% (41/98) of patients had died because of disease recurrence and distant metastasis. For survival curve analysis, the percentage of censored cases was 58.2% (57/98). Key inclusion criteria were: (a) histologically confirmed primary STAD, (b) availability of complete clinicopathological and follow-up data, and (c) no prior radiotherapy or chemotherapy before surgery. Exclusion criteria included: (a) incomplete medical records, (b) presence of other synchronous malignancies, and (c) perioperative mortality. Tumor staging was performed according to the 7th edition UICC/AJCC TNM classification system. Tumor grade and stage were classified in accordance with the Union of International Cancer Control (UICC)/American Joint Committee on Cancer (AJCC) pathologic tumor−node−metastasis (TNM) classification, 7th edition (2010). No patients in the study had undergone preoperative radiotherapy or chemotherapy.

GEPIA (http://gepia.cancer-pku.cn) is a newly developed interactive web server for analyzing the RNA sequencing expression data of 9, 736 tumors and 8, 587 normal samples from TCGA and GTEx projects (18). GEPIA provides customizable functions such as tumor/normal differential expression analysis according to cancer types or pathological stages, patient survival analysis, similar gene detection, correlation analysis and dimensionality reduction analysis. In the present study, differential transcriptional expression of ATG13 between gastric cancer tissues and normal tissues was assessed via GEPIA database analysis.

The effect of genomic alterations of the ATG13 gene including mutations, deletion and copy-number variance on overall survival (OS), disease−free survival (DFS), disease−specific survival (DSS) and progress−free survival (PFS) of cancer patients were calculated using the cBioPortal online database (http://www.cbioportal.org) (19). The cBioPortal for Cancer Genomics provides visualization, analysis and download of large-scale cancer genomics datasets. An OncoPrint graphic from cBioPortal were utilized to show the proportion and distribution of samples with genomic alterations.

Immunohistochemical staining was performed using a standard EnVision complex method previously described (20). Briefly, 4-μm sections cut from the tissue microarray chip containing the tissue samples that had been fixed in 10% (vol/vol) neutral-buffered formalin and embedded in paraffin were deparaffinized, rehydrated and incubated with a ATG13 rabbit polyclonal antibody (catalog no. ab105392; Abcam, Cambridge, MA, USA) at a dilution of 1:200 overnight at 4°C, followed by incubation with an Envision antibody complex (anti-mouse/rabbit) using an Envision™ Detection kit (Gene Tech., Shanghai, China). Staining was visualized by incubation with 3, 3’-diaminobenzidine (DAB) as the chromogen substrate, resulting in a brown-colored precipitate at the antigen site. Nuclei were counterstained with hematoxylin. Ten random microscopic fields per slide at a magnification of 400× were evaluated. To avoid errors in the observer’s field of view, the immunohistochemical results were interpreted by two experienced pathology instructors. Global ATG13 immunostaining were graded semi-quantitatively by multiplying the percentage of positively-stained cancer cells and the score of staining intensity as the final scores. The mean percentage of positively-stained cells was scored as 0-100%. The staining intensity was scored as follows: absent (0); weak (1); moderate (2); and strong (3). The multiplication of these parameters was used as the final staining score (0-300%). Patients were separated into low and high-expression groups according to the median value (148.75%) of the staining scores. For statistical evaluation, the tumor samples with a final score <148.75% were classified as low ATG13 expression and those with a score ≥148.75 as high ATG13 expression.

Cell culture and transfection. Two gastric cancer cell lines, denoted HGC27 (catalog no. TCHu22) and MKN45 (catalog no. TCHu280) from the Cell Bank of Chinese Academy of Sciences, were routinely cultured as described previously (21). A pre-designed validated siRNA against the sequence 5’-GCCATGTTTGCTCCCAAGAAT-3’ of human ATG13 was chemically synthesized from Takara Bio Inc. (Beijing, China). Cell transfection was performed using the Lipofectamine 2000 reagent (GE Healthcare Life Sciences, Little Chalfont, UK) following the manufacturer’s instructions. Cells were harvested 72 h post-transfection and used for subsequent experiments, which was repeated at least three times.

This assay has been described previously (21). Briefly, approximately 50 µg of total proteins extracted from cell lysates were separated on SDS-PAGE, transferred to PVDF membranes (EMD Millipore, Billerica, MA, USA) and incubated a ATG13 polyclonal antibody (Abcam) at a dilution of 1:500 overnight at 4 °C. Immunoblots were then probed with the secondary antibodies conjugated to horseradish peroxidase against rabbit IgG (Abcam), and signals were visualized using an enhanced chemiluminescence system (GE Healthcare Life Sciences, Chalfont, UK) in line with the manufacturer’s protocols. The blots were re-probed with a rabbit polyclonal antibody against GAPDH (Abcam) to confirm equal loading of different samples. The intensity of ATG13 was quantified using the Image J software.

The capacities of cell proliferation and colony formation were examined using a 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyl-2H-tetrazolium bromide (MTT) colorimetric assay and a monolayer colony formation assay respectively, according to the standard methods described before (21).

Cell invasive capacity was evaluated using a Matrigel invasion chamber assay described before (21). Briefly, 1×10⁵ cells per 500 μl of Dulbecco’s modified Eagle’s medium (Hyclone, Logan UT, USA) were seeded into the upper chamber of 24-well Matrigel-coated Transwell inserts (EMD Millipore). The lower chamber of the insert was filled with 750 μl of DMEM with 10% fetal bovine serum (Hyclone). After 24 h incubation, cells on the underside surfaces were fixed with 4% paraformaldehyde, stained with 2.5% crystal violet reagent (Sigma−Aldrich; Merck KGaA, Darmstadt, Germany) and counted in five random fields under the microscope.

Statistical analyses were performed using the SPSS 17.0 statistical software package (SPSS Inc., Chicago, IL, USA). For bioinformatic processing of TCGA/GTEx/GEPIA data, the expression data are first log2 (FPKM/or TPM + 1) transformed for differential analysis and the log2FC was defined as median (Tumor)-median (Normal). The false discovery rate (FDR) using the Benjamini-Hochberg method was computed to correct for multiple hypothesis testing. Data from at least three independent cell culture experiments were presented as the mean ± standard deviation (SD). Continuous variables were compared using Student’s t-test or the Mann-Whitney U test, as appropriate based on data distribution. Categorical variables were analyzed using the Fisher’s exact test. The Kaplan-Meier method with a log-rank test was used to assess clinical outcome after surgery. Patients with missing data or loss of follow-up were not included in the above analyses. To identify independent prognostic factors, variables achieved significance in univariate analysis were included in a multivariate Cox proportional hazards regression model. The proportional hazards assumption was tested and confirmed. A two-tailed P-value <0.05 was considered statistically significant.

We set out to characterize the differences of ATG13 transcriptional levels between tumor and normal tissues across diverse cancer types by means of bioinformatics analysis. Using the date from TCGA, a graph was obtained to demonstrate the ATG13 mRNA expression profile in 33 cancer types including STAD. Among them, ATG13 mRNA expression was observed significantly higher in STAD tissues than in normal tissues (Figure 1A). Further analyses were conducted to demonstrate the differential ATG13 expression in gastric cancer tissues and their matching adjacent noncancerous tissues, using the data from GTEx. As expected, ATG13 expression was significantly increased in STAD tissues compared to the paired noncancerous tissues (Figure 1B). The above findings were supported by the observation from the GEPIA data base analysis (Figure 1C). In addition, the immunohistochemical analysis also confirmed upregulated ATG13 expression in STAD tissues compared to the matching adjacent noncancerous tissues (Figure 1D).

We next identified the prognostic association between genomic alterations of the ATG13 gene and patient survival using the cBioPortal online database. The genomic alteration rate for ATG13 was 2.6% in eight studies containing 1955 gastric cancer patients (Figure 2A). Mutation and amplification were observed to appear more frequently in both stomach adenocarcinoma and esophagogastric cancer (Figure 2B). Further correlation analyses revealed the top 10 most frequent co-altered genes whose genomic alterations were correlated with ATG13, including TTN, AMBRA1, HARBI1, CRY2, CHST1, PTPRT, ARHGAP1, MAPK8IP1, VPS13B and SYNE1 (Figure 2C). However, following analyses by Kaplan-Meier plot with a log-rank test, no significant association was found between genomic alterations of ATG13 and patient survival, as regarding OS, DFS, DSS or PFS (Figures 2C–G).

We then conducted immunohistochemistry to determine ATG13 protein expression using one tissue microarray chip containing total 99 primary STAD specimens. We found that high ATG13 immunostaining was observed in the cytoplasm of cancer cells in 39.4% (39/99) of the cancer samples tested (Figure 3). We found that high ATG13 expression was significantly associated with larger tumor size (Table 1). Kaplan-Meier survival analysis revealed that patients with high ATG13 expression had a worse OS than those with low ATG13 expression (Figure 4). Upon univariate analysis, ATG13 expression, T stage (invasion depth), N stage (lymph node metastasis), and M stage (distant metastasis) were detected to be responsible for an unfavorable OS (Table 2). Following adjusting the prognostic variables from the univariate analysis, ATG13 expression, N stage and M stage kept independent values in the multivariate analysis (Table 2).

We further investigated the biological functions of silencing ATG13 expression in gastric cancer cells. The efficiency of ATG13 silencing using a pre-design validated siRNA was examined by western blot analysis (Figure 5A). We found that HGC27 and MKN45 cells with ATG13 knockdown exhibited a significant decrease in cell proliferation and colony formation compared to their control cells transfected with non-specific control siRNAs (Figures 5B, C). Additionally, the impact of ATG13 knockdown on cell invasiveness was evaluated using a Matrigel assay. By doing so, we observed that ATG13 knockdown dramatically decrease the number of invaded cells in both cell lines (Figure 5D). These findings suggest that silencing ATG13 expression suppresses the growth and metastasis of gastric cancer cells.