The human gastric cancer cell line MKN-45 was purchased from the Cell Resource Center, Institute of Basic Medical Sciences, Chinese Academy of Sciences, Peking Union Medical College. Cell culture was performed at the Clinical Medical Research Center of the Inner Mongolia Medical University. MKN-45 cells were cultured in RPMI-1640 medium (Invitrogen; Thermo Fisher Scientific, Inc.) with 10% fetal bovine serum (FBS; HyClone; GE Healthcare Life Sciences) and 1% penicillin-streptomycin (Invitrogen; Thermo Fisher Scientific, Inc.) and maintained in a humidified CO₂ incubator at 37°C. MKN-45 is a poorly differentiated human gastric adenocarcinoma cell line, and 90% of MKN-45 cells exhibit stem cell characteristics (19).

Extraction and purification of bioactive peptides were performed as previously reported (20). Additionally, the optimal concentration of 20 µg/ml bioactive peptides was determined and selected for the treatment of MKN-45 cells (10,11).

OXA was purchased from Jiangsu Aosaikang Pharmaceutical Co., Ltd. and dissolved in DMSO as a stock solution. The yield of cultured MKN-45 cells in the laboratory was 1×10⁶ cells/ml. After being cultured for 24 h, 20 µg/ml of induced ACBP, 15 µg/ml OXA, and a combination of 10 µg/ml induced ACBP and 7.5 µg/ml OXA were added to the cell culture medium. The negative control group was treated with phosphate-buffered saline (PBS). PBS is a phosphate buffer, which acts as a dissolving protective agent, does not affect cell growth and causes no damage to cells (11,21,22). After 36 h of incubation with all three treatments, the cells were collected for further analysis (13,20,23). All experiments were performed in triplicate.

Following addition of an appropriate amount of SDT-lysis buffer, the cells were ultrasonicated at 80 W for 10 repeated cycles, which included sonication for 10 sec, pausing for 15 sec, and boiling for 15 min. The supernatants of the cell lysates were collected after centrifugation at 14,000 × g for 40 min. Proteins were quantified by the bicinchoninic acid assay. The samples were transferred to a dispensing pack and stored at −80°C. Three biological replicates were performed for each group (24).

Protein samples (20 µg) were mixed with 5X loading buffer and boiled for 5 min. Subsequently, SDS-PAGE was conducted on a 12.5% (v/w) polyacrylamide gel. Three biological replicates were performed for each group.

Protein sample solution (30 µl) was mixed with DTT to a final concentration of 100 mM and then boiled for 5 min. After cooling to room temperature, 200 µl of UA buffer was added and the mixture was transferred to a 10-kDa ultrafiltration centrifuge tube. Following centrifugation at 14,000 × g for 15 min at 37°C, the filtrate was discarded. This step was repeated once. The tube was supplemented with 100 µl IAA buffer (100 mM IAA in UA), followed by shaking at 4,000 × g for 1 min at 37°C. Subsequently, the mixture was incubated in the dark for 30 min at room temperature, and then centrifuged at 14,000 × g for 15 min. Following addition of 100 µl UA buffer, centrifugation was performed at 14,000 × g for 15 min. This step was repeated twice. Next, the tube was loaded with 100 µl of 10-fold diluted dissolution buffer and centrifuged at 14,000 × g for 15 min. This step was repeated twice. After loading with 40 µl trypsin buffer (4 µg trypsin in 40 µl dissolution buffer), the tube was shaken at 4,000 × g for 1 min and incubated for 16–18 h at 37°C. Then, the tube was substituted with a new collecting tube, which was centrifuged at 14,000 × g for 15 min. Following addition of 40 µl of 10-fold diluted dissolution buffer, the filtrate was collected after centrifugation at 14,000 × g for 15 min. The peptides were desalted with a C18 cartridge and redissolved with 40 µl dissolution buffer after lyophilization. Finally, the peptide samples were quantified by measuring absorbance at 280 nm (OD₂₈₀) (24).

Peptides (~100 µg) in each group were labeled with the iTRAQ Labeling Kit (AB SCIEX Co.) according to the manufacturer's instructions. Three biological replicates were performed for each group.

The labeled peptides of each group were mixed and fractionated using an AKTA Purifier 100 (GE Healthcare). Buffer A (pH 3.0) containing 10 mM KH₂PO₄ and 25% ACN was used as the mobile phase. Buffer B (pH 3.0) containing 10 mM KH₂PO₄, 500 mM KCl and 25% CAN was used as the eluent. Subsequently, the column was equilibrated with buffer A. The peptide samples were then separated by the column at a flow rate of 1 ml/min. The linear gradient of buffer B was from 0 to 8% in 22 min, 8 to 52% in 25 min and 52 to 100% in 3 min, maintained at 100% for 8 min, and then reset to 0%. The elution profile was monitored by UV absorbance at 214 nm. Finally, the fractions were collected every 1 min and desalted using a C18 cartridge.

All samples were separated using an HPLC system Easy nLC at a nanoliter flow rate. Buffer A was a 0.1% formic acid-water solution, whereas buffer B was a 0.1% formic acid-84% acetonitrile-water solution. The column was equilibrated with 95% buffer A. Samples loaded in the autosampler were transferred onto the loading column (Acclaim PepMap100, 100 µm ×2 cm, nanoViper C18; Thermo Fisher Scientific, Inc.) and then separated by an analytical column (EASY-Column, 10 cm, ID 75 µm, 3 µm, C18-A2; Thermo Fisher Scientific, Inc.) at a flow rate of 300 nl/min. The linear gradient of buffer B was from 0 to 35% in 50 min, 35 to 100% in 5 min, and maintained at 100% for 5 min.

HPLC-fractionated samples were subjected to MS using a Q-Exactive mass spectrometer (Thermo Fisher Scientific, Inc.). The parameters used in MS were as follows: Detection mode, positive ion; analysis time, 60 min; scanning range of parent ion, 300–1,800 m/z; MS1 resolution, 70,000 at 200 m/z; AGC target value, 3.0×10⁻⁶; first-order maximum IT, 10 msec; dynamic exclusion, 10 msec; The mass-to-charge ratios of peptides and their fragments were recorded. Ten fragmentographies were acquired from MS2 scans. MS2 activation type, higher energy collisional dissociation (HCD); isolation window, 2 m/z; resolution, 17,500 at 200 m/z; microscans, 1; second-order maximum IT, 60 msec; normalized collision energy, 30 eV; and underfill ratio, 0.1%.

The raw MS data were extracted from RAW files. Mascot version 2.2 (Matrix Science) and Proteome Discoverer version 1.4 (Thermo Electron) were used for molecular identification and quantitative analysis. The MS data were analyzed with the UniProt protein database. Carbamidomethylation of cysteines, iTRAQ labeling at the N-term and lysine side-chain amino groups were set as the fixed modifications, while the oxidation of methionine and iTRAQ4plex (Y) was set as a variable modification. The false discovery rate for each peptide was adjusted to 1%, and the minimum peptide length was specified to 6. In addition, the enzyme specificity was set to trypsin, and up to two missed cleavages were allowed. Mass tolerance was set as 20 ppm for precursor ions and 0.1 Da for fragment ions.

GO annotation of the target proteins was conducted by Blast2GO (25), which consisted of four steps: i) Sequence alignment (BLAST), ii) GO entry extraction mapping, iii) GO annotation, and iv) data augmentation. First, the target proteins were aligned against the specific protein sequence database using the localized sequence alignment tool NCBI BLAST+ (ncbi-blast-2.2.28+-win32.exe), and the first 10 aligned sequences with an E-value ≤1e-3 were retained for subsequent analysis. Next, Blast2GO Command Line was used to extract the GO entries related to the target protein sets and the aligned proteins or homologous proteins with high sequence identity (database version: go_201608.obo, www.geneontology.org). During the annotation process, Blast2GO Command Line annotated the target proteins with the GO entries extracted from the mapping process based on the sequence similarity between the targeted and aligned proteins, the reliability of GO entry sources, and the structure of the GO-directed acyclic graph. Following annotation, the conserved motifs that matched with target proteins were searched against the EBI database using InterPro Scan to improve the annotation efficiency. Functional information related to the motifs was obtained and annotated to the target protein sequences. ANNEX was used to enhance the annotation information, and a link between different GO categories was established to improve the accuracy of GO annotations.

KO (KEGG Orthology) in the KEGG database is a classification system for genes and their products. Orthologous genes with similar functions are grouped with their products on the same pathway, and a KO (or K) tag was assigned for their interaction. KO classification was performed on the target protein sequences by BLAST against the KEGG GENES database using KEGG Automatic Annotation Server software. Additionally, the details on the pathways associated with target protein sequences were obtained according to the KO classification.

For clustering analysis, the quantitative information of the target protein set was first normalized to the [-1, 1] interval. Next, Cluster 3.0 software (http://bonsai.hgc.jp/~mdehoon/software/cluster/software.htm) was used to classify the two dimensions of the sample and protein expression (distance algorithm, Euclidean; connection, Average linkage) simultaneously. Finally, a hierarchical clustering heat map was constructed using Java TreeView software, version 3.0 (http://jtreeview.sourceforge.net).

Gene symbols obtained from the database of target protein sequences were used to investigate the direct and indirect interactions between target proteins and experimental evidence via the IntAct database (http://www.ebi.ac.uk/intact/main.xhtml). CytoScape version 3.2.1 (http://www.cytoscape.org/) and String database (https://string-db.org/) were used to generate the PPI network and analyze the network topologies.

To verify the protein expression levels obtained by iTRAQ analysis, the expression levels of selected proteins were quantified by LC-PRM-MS analysis (26). Briefly, peptides were prepared according to the iTRAQ reagents protocol. An AQUA stable isotope peptide was spiked in each sample as an internal standard reference. For desalting purposes, tryptic peptides were loaded into C18 stage tips on an Easy nLC-1200 system prior to reversed-phase chromatography. A LC gradient of acetonitrile ranging from 5 to 35% in 45 min was used. PRM analysis was performed using a Q-Exactive Plus mass spectrometer. The optimization of collision energy, charge state and retention times for the most significantly regulated peptides was conducted by unique peptides with the highest intensity and confidence. A full MS scan was carried out in a positive ion mode mass spectrometer with 70,000 resolution (at 200 m/z), an AGC target value of 3.0×10⁻⁶, and a maximum ion injection time of 250 msec. Then, 20 PRM scans were performed at 35,000 resolution (at 200 m/z), an AGC target value of 3.0×10⁻⁶ and a maximum injection time of 200 msec. The targeted peptides were isolated using a 2 Thomson (Th) window. Peptide fragmentation was induced by higher-energy collisional dissociation (HCD) at a normalized collision energy of 27. The raw proteomics data were analyzed with Skyline software, version 19.1 (MacCoss Lab, University of Washington) (27), where the signal intensities for the identified peptide sequences were relatively quantified and normalized with a reference standard.

To validate the results of MS, 6 differentially expressed proteins (TPX2, NUSAP1, TOP2A, YAP, MKi-67 and GPC4) were selected for PRM analysis. The criteria for the validation of proteomics data were as follows: i) Potential biological functions and significant differential expression; ii) the number of peptide fragments detected by LC-MS/MS was >1; and iii) novel oncoproteins that were decreasingly expressed in MKN-45 cells after treatment with ACBP-OXA compared with ACBP or OXA treatment alone.

Student's t-test was used to analyze the statistical software SPSS (version 22, IBM Corp.). Data are expressed as means ± standard deviation of three independent biological replicates. A P-value of <0.05 was considered to indicate statistically significant differences (Tables II, III, SII and SIII).

In the present study, the iTRAQ technique was applied to analyze the proteomics profile of MKN-45 cells treated with a combination of ACBP and OXA. The entire experimental procedure is illustrated in Fig. 1. A total of 6,210 proteins were detected (Table SI). Quality control of protein data revealed that the molecular mass of proteins fell in the range of 5–100 kDa (Fig. 2A), and the majority of the peptides were ~7–15 amino acids in length (Fig. 2B), which appeared to be similar to the known properties of tryptic peptides.

Compared with the control group, MKN-45 cells treated with ACBP, OXA and ACBP-OXA exhibited 17 (10 up- and 7 downregulated), 111 (27 up- and 84 downregulated) and 128 (53 up- and 75 downregulated) differentially expressed proteins, respectively (Tables I and SII). Proteins with a >1.2-fold change in expression (either up- or downregulation) and P<0.05 were considered to be differentially expressed. The protein expression remained unchanged in the ACBP- or OXA-treated cells by the best screening criteria for the differentially expressed proteins (>1.2-fold change in expression and P<0.05). However, in ACBP-OXA-treated cells, the protein expression changed or changed in the opposite direction compared with the ACBP- or OXA-treated groups. According to the criteria mentioned above, a total of 77 differentially expressed proteins were subjected to further screening (Table II). Of these 77 proteins, 40 were significantly upregulated and 37 were significantly downregulated. Specifically, the top 5 differentially upregulated proteins were hemopexin, vitamin D-binding protein, fatty acid-binding protein, α-fetoprotein and α-2-macroglobulin, whereas the top 5 differentially downregulated proteins were ribosome biogenesis protein BRX1, dedicator of cytokinesis protein 6, targeting protein for Xklp2, DNA topoisomerase 2-α and proliferation marker protein Ki-67.

To improve the accuracy of protein identification, the PRM mode was first used to specifically monitor the peptide sequences of 6 target proteins in mixed samples. The mixed samples were repeatedly tested 3 times using LC-PRM/MS, and the PRM data were analyzed by Skyline software, version 19.1 (MacCoss Lab, University of Washington). The results demonstrated that the 6 target proteins had credible peptides. The corresponding peptides were selected for PRM quantitative analysis. The 3-time repeated tests of LC-PRM/MS stably detected 10 candidate peptides of the 6 target proteins, with a relative standard deviation of ~12%. These results support that PRM is an accurate method for the quantification analysis of peptides (Table SIII). Three daughter ions of the most abundant and consecutive peptides were selected for different analyses, such as quantitative analysis at the protein and peptide levels, data calibration and biostatistical analysis (Table SIV). First, the peak area of the daughter ion was integrated to obtain the original peak area of peptides. Second, the peak area of the internal standard peptide was labeled with heavy isotopes for correction purposes. The relative expression levels of each peptide in all samples were then measured. Finally, the mean relative expression level of the target peptide was calculated and statistically analyzed (Table SV).

Based on the corresponding peptide fragments, the differences in the relative expression level of target proteins were further calculated in different samples. In other words, the mean ratios of multiple peptides were calculated (Table III). The LC-PRM/MS results revealed that quantitative information of target peptide fragments was obtained for all samples. Subsequently, the target proteins and peptide fragments were subjected to relative quantitative analysis by adding heavy isotope-labeled peptide fragments. The results demonstrated that differential protein expression was observed among the 6 target proteins. Notably, the protein expression levels were significantly decreased in MKN-45 cells treated with ACBP-OXA, but not cells treated with ACBP or OXA alone.

All differentially expressed proteins detected by MS were further analyzed using bioinformatics methods.

Hierarchical clustering results are presented as heat maps, where the red color indicates upregulation and the green color downregulation. All samples were classified into four categories: C1-C3, ACBP1-ACBP3, OXA1-OXA3 and ACBP-OXA1-ACBP-OXA3. It was observed that the proteins may be divided into two categories via vertical comparison. The differential changes in protein expression between ACBP-OXA-treated MKN-45 cells and the control group are shown in Fig. 3. In addition, the differentially expressed proteins were significantly altered in ACBP-treated and OXA-treated MKN-45 cells (Figs. S1 and S2, respectively) compared with the control group, suggesting the rationality of differential expression patterns of selected target proteins. Therefore, clustering analysis may be used to assess the reasonability of screening differentially expressed proteins, e.g., whether changes in the expression levels of these target proteins can indicate the therapeutic effect of biological agents on cancer cells.

GO is a functional classification system that provides a set of dynamically updated standardized vocabulary to describe the properties of genes and gene products based on three different perspectives: i) The involved biological process, ii) molecular function and iii) cellular component. Differentially expressed proteins in ACBP- (n=17), OXA- (n=111) and ACBP-OXA-treated cells (n=128) exhibited a total of 734, 2,295 and 2,402 functional annotations, respectively (Tables SVI–SVIII). Similarly, the protein functions predicted from secondary GO enrichment analysis were divided into three categories: i) Molecular function, ii) cellular component and iii) the involved biological process. Each protein contained at least one functional GO annotation. Furthermore, GO functional annotations of differentially expressed proteins were compared between the four groups. Compared to the control group, 17 differentially expressed proteins in the ACBP group tended to be located at the macromolecular complex and were closely associated with catalytic activity, protein-binding and nucleic acid-binding transcription factor activity. These proteins were involved in cellular processes, metabolic processes and responses to stimuli (Fig. 4). The 111 differentially expressed proteins of the OXA group were found in the nucleus and cytosolic part and were associated with structural molecule activity. These proteins participated in diverse biological processes, such as metabolic processes, cellular component organization or biogenesis and localization (Fig. 5). The 128 differentially expressed proteins in the ACBP-OXA group were located in the nucleolus, membrane-enclosed lumen and organelle part and were involved in catalytic activity and structural molecule activity, which can affect signaling, cellular component organization or biogenesis and negative regulation of biological processes, indicating that the combination of two drugs exerts complementary and synergistic therapeutic effects on gastric cancer cells (Fig. 6).

Candidate proteins regulate complex pathological and physiological processes through interaction and intercoordination with other proteins. KEGG pathway analysis was conducted to identify the key signaling pathways and related regulatory processes underlying the therapeutic effects of ACBP, OXA and ACBP-OXA (Tables SIX–XI). The identified signaling pathways in the ACBP group were associated with glycosylation biosynthesis, steroid hormone synthesis, mineral element absorption and cell cycle, among others (Fig. 7). KEGG signaling pathways in the OXA group were associated with cancer-related signaling pathways, adhesions, transcriptional error regulation in cancer, RNA transcription, ECM-receptor activation and PI3-AKT signaling, among others (Fig. 8). The KEGG signaling pathway in the ACBP-OXA group was associated with ribosomes, cancer-related signaling pathways, chemokines, PPAR and AMPK, among others (Fig. 9).

Direct interaction patterns between differently expressed proteins can be beneficial for extracting important information on target proteins. As shown in Fig. 10, the yellow color indicates the differentially expressed target proteins, and the blue color represents the proteins that interacted with these differentially expressed target proteins. In addition, PPI analysis revealed the connectivity degree of the protein interactions. A high connectivity degree may be more indicative of a protein complex. Through intergroup comparison between the ACBP-OXA and control groups (Fig. S3), it was demonstrated that TPX2 (Fig. 10A), TOP2A (Fig. 10B), MKi-67 (Fig. 10C) and GPC4 (Fig. 10D) exhibited a high degree of connectivity, which was located at the center of the network.