We downloaded data we need from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov), which includes 343 GC and 30 tumor-adjacent tissues. Expression data were converted using the R software (version 4.2.1) for subsequent analysis.

After downloading clinical information, we evaluated the relationship between them and apoc1. Clinical information included age, sex, grade, and stage, T, N, M, and apoc1. After comparison, a relevant heat map was drawn based on previous results. “Limma” and “ggpubr” packages were included.

GC cell lines (AGS, MKN45) were purchased from the Type Culture Collection of the Chinese Academy of Science (Shanghai, China) and fed MKN45 into RPMI 1640 (Servicebio, Wuhan, China). AGS cells were cultured in F12K (Servicebio, Wuhan, China). All media were supplemented with fetal bovine serum (FBS) (Gibco, NY, USA) and a penicillin-streptomycin mixture. The cells were cultured in incubator with 5% CO₂ at 37°C.

apoc1 protein expression was detected in GC and adjacent tissues by IHC. Formalin-fixed paraffin-embedded tissues sections were degreased in xylene and rehydrated in different concentrations of alcohol and distilled water before antigen retrieval, then followed by antigen retrieval. The fixed sections were washed three times in phosphate buffered saline (PBS) (pH 7.3), incubated in 3% H₂O₂ for 7 min, and then washed three times with PBS again. The sections were blocked with BSA (Servicebio, Wuhan, China) for 30 min, followed by incubation with Anti-apoc1 primary antibody (Abcam, USA) at 4°C overnight, followed by incubation with anti-rabbit antibodies (Proteintech Technology, Wuhan, China) for 50 min. Subsequently, the sections were incubated with diaminobenzidine as a chromogen for 5 min. The sections were then counterstained with hematoxylin and dehydrated using ethanol and xylene. The sections were viewed under an inverted microscope (Nikon Corporation, Japan).

Total RNA was extracted from the GC cell lines using TRIzol regaent (Invitrogen, Carlsbad, USA). Hiscript III Reverse Transcriptase (Vazyme, Nanjing, China) was used for cDNA synthesis. cDNA was quantified using the ChamQ SYBR Color qPCR Master Mix (Vazyme, Nanjing, China) in StepOnePlus Real-Time PCR System (Applied Biosystems, Carlsbad, USA). Relative expression of apoc1 was calculated using the 2^−ΔΔCT method. Primers used are listed in Table 1.

Small interfering RNA targeting apoc1 and a negative control were acquired from RiboBio (Guangzhou, China). Cells were cultured overnight in serum-free medium before transfection. The cells were then transfected with si-RNAs or si-NC using Lipofectamine 3000 (Invitrogen, Carlsbad, USA).

Total cell proteins were isolated in NP-40 buffer (Beyotime, Shanghai, China), and proteins were separated via SDS-PAGE, followed by transfer onto PVDF membranes (Invitrogen, Carlsbad, USA). The membranes were then blocked with BSA solution for 2 h. Finally, the membranes were incubated with the corresponding primary antibodies at 4°C overnight. After washing with TBST, the membranes were incubated with the corresponding secondary antibodies. The antibodies used in this assay were Anti-apoc1 (Abcam, Cambridge, UK) and Anti-gapdh (Proteintech Technology, Wuhan, China).

Cells were cultured in serum-free medium and seeded into the top of Transwell chambers (Corning, Shanghai Shang, China) with or without Matrigel-coated filters (BD, NY, USA). Each chamber contained 2 × 10⁴ cells per well. A medium containing 10% FBS was added to the bottom of each chamber. 24 h after incubation, cells were stained with 1% crystal violet. Finally we counted the cell number using the ImageJ software.

First, cells were seeded into 6-well plates. When the density of cells reached at least 80% after transfection, a scratch was made in the middle of the well using a yellow pipette tip. The tip was maintained vertically at the bottom of the well. PBS was then adopted to wash the plate three times. The same scratch field was photographed at the indicated time (0, 24 and 48 h).

The animal work was approved by Nanjing Medical University. For the metastasis model, cells transfected with si-RNA or si-NC were injected into the caudal vein of anesthetized nude mice (three mice per group). 14 days later, the lungs of nude mice were removed to count metastatic foci.

We analyzed gene expression profile of GC patients using the ggpubr package, and relevant genes were defined with p < 0.05, along with corFilter ≥0.2, or ≤−0.2. After screening, scatter plots were drawn based on our results using the “ggExtra” package.

We divided the patients with GC into two groups based on the mean value of apoc1. DEGs were defined as p < 0.05, and logFC ≥1, or ≤−1. Specific genes are shown in the heat map. Limma and pheatmap packages were included.

The “clusterProfiler” package was used for functional enrichment analysis. The threshold conditions were as follows: p < 0.05, q < 0.2.

We used the “estimate” package to calculate the TME scores of the GC samples. The TME score consists of two parts: stromal and immune. We then compared these differences based on the expression of apoc1. Finally, the “reshape2” and “ggpubr” packages were adopted for the violin plot.

To determine the relationship between apoc1 and immune cell infiltration, we first used CIBERSORT (http://cibersort.stanford.edu/) to calculate the percentage of immune cells. Next, the “reshape2” and “ggpubr” packages were also used in our study for box plot and scatter diagrams. Then, we drew the correlation circle diagram according to our previous results. Finally, we acquired a list of immune checkpoints and their expression profile data, and screened out checkpoints that are associated with apoc1.

To calculate the sensitivity of the targeted drugs, we used the “pRRophetic” package. Then, apoc1 was analyzed and combined with the IC50 results to draw a boxplot. Similarly, we downloaded the immunotherapy scores from the TCIA website (https://tcia.at/) and analyzed the immunotherapy score and apoc1 expression.

Data are presented as the mean ± standard deviation, and analysis was conducted using GraphPad Prism 8.0 software (GraphPad Software Inc., La Jolla, CA, USA) and ImageJ (National Institutes of Health, USA). We also applied Student’s t-test for comparing the experimental groups and control groups. Statistical significance was set at p < 0.05.

First, we conducted a differential analysis of the data of GC patients from the TCGA database, and filtered out 1621 differential genes, with |logFC| ≥ 2 and p < 0.05. These differential genes included 872 highly expressed genes and 749 genes with low expression. Based on our results, we drew volcano and heat maps (Figs. 1A and 1B). Among these differentially expressed genes, we chose apoc1 as the target of our study. We obtained its expression profile in 33 human tumors from the TIMER database (http://TIMER2.0 (cistrome.org)) and found that apoc1 is abnormally expressed in various types of human tumors, including GC, suggesting that apoc1 is a tumor-related gene (Fig. 1C). We compared the apoc1 expression profile between GC and adjacent tumor tissues and found that apoc1 was significantly elevated in GC (Figs. 1D and 1E). Immuno-histochemical studies also demonstrated significant upregulation of apoc1 in GC tissues compared with adjacent tissues (Fig. 1F). Finally, we conducted survival analysis based on survival time and the apoc1 expression profile from the KM plot database (http://kmplot.com/analysis/), and the results indicated that apoc1 expression was inversely proportional to patient prognosis (Fig. 1G). Overall, apoc1 expression was elevated in GC, indicating a poor prognosis.

We first divided the samples into different groups based on their clinical and pathological features and compared the expression of apoc1. No statistical differences were observed in age (Fig. 2A), sex (Fig. 2B), grade (Fig. 2C), M (Fig. 2D) and N (Fig. 2G) in apoc1 expression. Compared with the T1 stage, apoc1 expression was significantly increased in the T2, T3 and T4 stages (Fig. 2F), and the same phenomenon was observed in Stages I, II, III, and IV (Fig. 2E). Finally, we created clinically relevant heat maps to confirm our conclusions (Fig. 2H).

We obtained three siRNAs targeting apoc1 and a negative control for apoc1 knockdown. To confirm this efficiency, we performed qRT-PCR and western blotting (Figs. 3A and 3B). We then conducted cell migration and cell invasion assays, and the results showed that the previously mentioned abilities of GC cells transfected with siRNAs were reduced (Figs. 3C and 3D). Furthermore, this was confirmed by the results of the wound healing assay (Figs. 3E and 3F) and the model of pulmonary metastasis (Fig. 3G). Thus, we conclude that apoc1 plays a major role in cell migration and invasion.

To explore the mechanism by which apoc1 regulates the progression of GC, we conducted analysis to investigate the correlation between apoc1 and other genes in TCGA database. With |cor| ≥ 0.2 and p < 0.05, a total of 2752 correlated genes were identified and correlation circle maps were drawn (Fig. 4A).

Next, we analyzed and adopted 816 differentially expressed genes (DEGs) based on apoc1 expression, with |logFC| ≥ 1 and p < 0.05, as the standard and drew a heat map (Fig. 4B). GO, KEGG, and GSEA analyses of the DEGs showed that apoc1 was closely related to the Wnt pathway and immune regulation (Figs. 4C and 4E). We found that apoc1 may participate in Wnt signaling via wnt2 (Fig. 4F). Overall, apoc1 may participate in GC progression via the Wnt axis and immune activities.

We further investigated its role in the TME. The results revealed that the stromal and immune scores of the apoc1 high expression group were higher (Fig. 5A).

We then studied the relationship between the fraction of immune cells and the expression profile of apoc1 and found that its expression was related to the proportion of various immune cells (Fig. 5B). Next, we analyzed the correlation between the proportion of these immune cells and apoc1. The proportion of CD8⁺T cells, memory activated CD4⁺T cells, resting dendritic cells, Macrophages M1 and Macrophages M2 were positively proportional to apoc1, while the proportion of activated dendritic cells, activated mast cells, naïve B cells, and resting memory CD4⁺T cells were negatively proportional to their expression (Fig. 5C). Subsequent analysis of the immune cell correlation confirmed our conclusion (Fig. 5D). Finally, we obtained immune checkpoints related to apoc1 (Fig. 5E).

Targeted immunotherapy is a widely used treatment for advanced GC. To determine the relationship between apoc1 expression and the sensitivity of targeted drugs, we analyzed the expression and IC₅₀ of various targeted drugs in TCGA GC samples, and found that the sensitivity of GC patients to Bexarotene, BEZ235, GNF-2, MG-132, OSI-930, Ruxolitinib, Sunitinib and Talazoparib was relevant to the expression of apoc1, and the IC₅₀ of these targeted drugs in the apoc1-high group was lower (Fig. 6A), suggesting that doctors can decide which one to utilize based on the level of apoc1 in GC patients.

Among the various immunotherapy regimens, PD-1 and CTLA-4 are the most commonly adopted. To study whether the expression of apoc1 affects the curative effect of PD-1 and CTLA-4, apoc1 expression and immunotherapy scores in GC patients were analyzed. The results indicated that the immune scores of the apoc1 highly group were higher than those in the low group, regardless of whether CTLA-4 was positive or not (Fig. 6B). However, whether CTLA-4 was positive or not and whether its curative effect was related to the expression of apoc1 was not significantly correlated with apoc1 expression. In summary, we found some targeted drugs that may have better efficacy in the apoc1 overexpression group, and apoc1 may play a major role in PD-1 therapy.