Metformin was purchased from Sangon Biotech (Shanghai, China). Bovine serum albumin (BSA) was purchased from Sigma-Aldrich (Merck KGaA, Darmstadt, Germany). A Cell Counting Kit (CCK)-8 and trypsin were purchased from Beyotime Institute of Biotechnology (Haimen, China). Recombinant human Calml3 protein was purchased from Sino Biological (Beijing, China).

TAFs were isolated from surgically resected gastric cancer specimens from six patients treated at the First People's Hospital of Xuzhou (Suzhou, China) from January to September 2015. The six patients were male with and age of 55–60 years (median age, 57 years). Normal gastric tissues were obtained from 6 age- and gender-matched trauma patients. All patients provided written informed consent for their tissues to be used for scientific research. Ethical approval of the study was obtained from the First People's Hospital of Xuzhou (Suzhou, China). TAFs were isolated as reported previously (28) and were pooled together. Gastric cancer samples were placed in RPMI-1640 medium (HyClone; GE Healthcare, Little Chalfont, UK) containing 4% meropenem (Haibin Pharmaceutical Co., Ltd, Shenzhen, China). Tissues were cut into pieces of ~3 mm³ in volume with a scalpel, placed in 25-cm² culture flasks and covered with 1 ml fetal bovine serum (FBS; HyClone; GE Healthcare). Next, the flasks were inverted for 48 h. After 48 h, RPMI-1640 medium supplemented with 20% FBS (HyClone; GE Healthcare) was added to the flasks, and the flasks were inverted again. Culture was performed in an incubator with humidified air containing 5% CO₂ at 37°C. Following tissue attachment (2–3 days), the culture medium [Dulbecco's modified Eagle's medium (DMEM; Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) with 10% FBS] was changed twice per week for the next 2–3 weeks. Under these conditions, fibroblasts were explanted from tissue fragments, whereas other types of cell were mostly retained within the tissue. The fibroblasts formed multiple dense colonies that spread out on the culture dish. After 10–14 days, the cultured cells were briefly trypsinized in an incubator for 5 min and re-seeded into new flasks (passage 1). After reaching confluence, which occurred every 3–4 days, the cultured cells were split at a ratio of 1:2. All fibroblasts used in the experiments of the present study were between passages 4 and 9.

The SGC-7901, BGC823, GES-1 and MGT-803 (29) human gastric cancer cell lines and the human gastric TAFs were cultured in DMEM supplemented with 10% FBS, 1% penicillin-streptomycin and 1 mM sodium pyruvate. These cells were grown at 37°C in a humidified atmosphere with 5% CO₂.

After centrifugation and re-suspension, cells were seeded in 96-well plates at 4,000 cells/well. The cells were co-cultured in a Transwell chamber with TAFs pre-treated with or without different concentrations of metformin. After 48 h, CCK-8 solution (Beyotime Institute of Biotechnology) was added and the plates were cultured for an additional 2 h, according to the manufacturer's instructions. The optical density was determined at 450 nm using a microplate reader (Synergy™ Neio; Biotek, Winooski, VT, USA). The cell viability of each group was determined in triplicate.

Cells were washed with PBS, fixed with 4% formaldehyde and blocked with 1% BSA in PBS for 1 h at room temperature. Cells were then incubated overnight at 4°C with a primary antibody against α-smooth muscle actin (1:1,000 dilution; cat. no. ab5694; Abcam, Cambridge, UK) and subsequently with cyanine 3-conjugated secondary antibodies (1:2,000 dilution; cat. no. A0507; Beyotime Institute of Biotechnology, Haimen, China) for 1 h at room temperature. Nuclei were stained using DAPI and images were captured using an FV1200 confocal microscope (Olympus, Tokyo, Japan).

Gastric TAF cells were treated with 0.2 mM metformin or PBS as a control for 12 h. Equal quantities of cells from triplicate wells were mixed to generate one sample for each group. Culture medium of TAFs treated with or without metformin was collected individually. The protein solution was reduced with 10 mM DTT for 1 h at 37°C and alkylated with 20 mM iodoacetic acid (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) for 45 min at room temperature in the dark. The protein sample was diluted with 100 mM TEAB buffer to reach a final urea concentration of <2 M. Finally, trypsin was added at a 1:50 trypsin-to-protein ratio for the first digestion overnight at 37°C, and 1:100 trypsin-to-protein ratio for a second 4-h digestion at 37°C. From each sample, ~100 µg protein was digested with trypsin for subsequent analysis.

After digestion, tryptic peptides were desalted using a Strata X C18 SPE column (Phenomenex, Torrance, CA, USA) and vacuum-dried. Peptides were reconstituted in 0.5 M TEAB and processed with a 6-plex TMT kit according to the manufacturer's protocol. One unit of each TMT reagent (defined as the amount of reagent required to label 100 µg of protein) was thawed, reconstituted in 24 µl acetonitrile (ACN), and added to each sample. Next, the peptide mixtures were incubated for 2 h at room temperature. The mixtures were pooled, desalted and dried by vacuum centrifugation.

The labeled peptide sample was then fractionated by high-pH reverse-phase HPLC using an Agilent 300 Extend C18 column (5-µm particles, 4.6 mm inner diameter and 250 mm length; Agilent Technologies, Inc., Santa Clara, CA, USA). Peptides were separated first, with a gradient of 2–60% ACN in 10 mM ammonium bicarbonate in water (pH 10) over 80 min, and the eluate was collected in 80 fractions. Next, the peptides were combined into 18 fractions and dried by vacuum centrifugation.

Peptides were dissolved in 0.1% formic acid (FA) and directly loaded onto a reverse-phase pre-column (Acclaim PepMap 100; Thermo Fisher Scientific, Inc.). Next, these peptides were separated using a reverse-phase analytical column (cat. no. 164568; Thermo Fisher Scientific, Inc.). The gradient began with an increase from 8 to 22% in solvent B (0.1% FA in ACN, HPLC grade) over 26 min, followed by a further increase to 35% over 6 min, rising up to 80% over 4 min and this concentration remaining constant for the last 4 min, all at a steady flow rate of 300 nl/min on an EASY-nLC 1000 ultra (U)PLC system (Thermo Fisher Scientific, Inc.). Spectra were obtained using a Q Exactive™ plus hybrid quadrupole-Orbitrap mass spectrometer (Thermo Fisher Scientific, Inc.).

The peptides were subjected to nano spray ionization source ionization (LTQ Oribtrap; Thermo-Fisher Scientific, Inc.) followed by MS/MS in a Q Exactive™ plus that was coupled online to the UPLC. Intact peptides were detected at a resolution of 70,000. Peptides were fragmented for MS/MS using a nomalized collision energy of 33%, and the resulting ion fragments were detected at a resolution of 17,500. A top-20 data-dependent method was applied for the top 20 precursor ions above a threshold ion count of 2×10⁴ in the MS survey scan, with 30.0-sec dynamic exclusion. The electrospray voltage applied was 2.0 kV. The automatic gain control setting of 5×10⁴ ions was used to prevent overfilling of the ion trap. For MS scans, the m/z scan range was 350–1,800. A fixed first mass was set at and m/z ratio of 100.

The resulting spectra were processed using Mascot search engine (v.2.3.0; http://data-pc/mascot/home.html). Tandem mass spectra were searched against the UniProt Homo sapiens database (20,274 sequences; http://www.uniprot.org/). Trypsin/proline was specified as a cleavage enzyme allowing up to two missing cleavages. The mass error was set to 10 ppm for precursor ions and 0.02 Da for fragment ions. Carbamidomethyl on Cys, TMT-6plex (N-term) and TMT-6plex (K) were specified as fixed modifications, and oxidation on Met was specified as a variable modification. As cut-off criteria, a false discovery rate of <1% and a peptide ion score of >20 were used.

First, the mass error of all the identified peptides was checked. The distribution of the mass error was near zero, and usually <0.02 Da. Most of the identified peptides were between 8 and 16 amino acids in length, which agrees with the expected length of tryptic peptides and indicates that tryptic digestion was efficient.

GO terms may be assigned to three domains: i) Cellular component, ii) Molecular function and iii) Biological process. GO annotation of the proteome was derived from the UniProt-GOA database (www.http://www.ebi.ac.uk/GOA/). The identified protein IDs were first converted to UniProt IDs and mapped to GO IDs in the UniProt database. Proteins lacking a UniProt-GOA annotation were analyzed using InterProScan (http://www.ebi.ac.uk/interpro/interproscan.html) to annotate the protein's functional sites based on protein sequence alignment. The proteins were then classified by GO and KEGG annotation.

To perform a standard clonogenic assay, cells were seeded in 6-well plates at 1,000 cells/well. After treatment with medium containing different concentrations of recombinant Calml3 for 48 h, cells were grown for 7–10 days to allow for colony formation, subsequently fixed with pure methanol for 15 min at 37°C, and then stained with crystal violet (0.1%) for 15 min at 37°C. Clusters consisting of ≥50 cells were considered as colonies.

Values are expressed as the mean ± standard error of the mean of at least three independent experiments. The results were evaluated by one-way analysis of variance followed by the Student-Neuman-Keuls post hoc test, or Student's t-test to determine statistical significance. The statistical analyses were performed using GraphPad Prism 6 software (GraphPad, Inc., La Jolla, CA, USA). P<0.05 was considered to indicate a statistically significant difference.

TAFs were isolated from gastric cancer tissue (Fig. 1A). A TAF-specific marker, α-SMA, was detected by immunocytochemistry to confirm the identity of the TAFs (Fig. 1B). The fibroblasts from normal gastric tissues did not express α-SMA at detectable levels (data not shown). Gastric cancer cells were seeded into a Transwell chamber and co-cultured with the isolated gastric TAFs (Fig. 1C). The cell viability assay demonstrated that gastric cancer cells grew faster when co-cultured with gastric TAFs than normally cultured gastric cancer cells (Fig. 1D and E). This suggests that gastric cancer TAFs promote the proliferation of gastric cancer cells, probably through the modulation of the cancer microenvironment.

Next, it was investigated whether metformin affected the tumor-promoting role of TAFs isolated from gastric cancer. Gastric TAFs were treated with different concentrations of metformin for 24 or 48 h. At concentrations of up to 1 mM, metformin only slightly influenced the viability of the TAFs (Fig. 2A and B). Next, gastric cancer cells were co-cultured with gastric TAFs in a Transwell chamber (Fig. 2C). Co-culture with TAFs that had been pre-treated with 0.2 mM metformin for 48 h, significantly decreased the proliferation of gastric cancer SGC-7901 and MGT-803 cells, compared with gastric cancer cells co-cultured with metformin-untreated TAFs (Fig. 2D and E). Although the proliferation of GES-1 was decreased when co-cultured with TAFs pre-treated with metformin, the difference was not significant (Fig. 2F). It was observed that metformin reduced the stimulatory effect of TAFs on the proliferation of gastric cancer cells, probably by altering the extracellular proteins secreted by gastric TAFs into the culture medium.

To investigate the proteomic changes in secreted proteins from gastric TAFs following metformin treatment, the proteomic profile of TAF-secreted proteins was investigated by TMT-based protein quantification (Fig. 3A). In metformin-treated TAFs, >360 differentially expressed proteins compared with those in untreated TAFs were successfully identified, including 14 significantly upregulated and 18 significantly downregulated proteins (Fig. 3B). The upregulated proteins included calmodulin-like protein 3 (Calml3), tropomyosin α-3 chain (TPM3), tissue-type plasminogen activator (PLAT) and nucleoside diphosphate kinase A (NME1).

To further determine the function and features of the identified proteins, GO annotation was performed, including protein domain, pathway and subcellular localization enrichment analysis. The differentially expressed proteins were summarized for each GO category. The proportion of differentially expressed proteins representing each subcellular location is summarized in Fig. 4A. Among the upregulated proteins, 37% were extracellular proteins, 29% were cytosolic proteins and 14% were nuclear proteins. Among the downregulated proteins, 33% were extracellular proteins, 28% were nuclear proteins and 17% were mitochondrial proteins. The results of the protein domain enrichment analysis and KEGG pathway enrichment analysis suggested that transcriptional dysregulation was significantly enhanced in cancer, whereas the extracellular matrix-receptor interaction pathway was the most significantly downregulated pathway (Fig. 4B and C). Finally, GO enrichment analysis of upregulation and downregulated proteins was performed (Figs. 5 and 6). Proteins involved in the positive regulation of protein and actin filament depolymerization were enriched among the dysregulated proteins.

The suppressive effect of metformin on the stimulatory role of TAFs on gastric cancer cells was likely to be exerted through the upregulation and downregulation of certain proteins released into the culture medium by TAFs. The further investigations were focused on the upregulated proteins, which may have potential antitumor activity. Among the dysregulated proteins identified in TAFs subjected to metformin treatment, Calml3 was most significantly upregulated, with a 2.88-fold increase in metformin-treated TAFs compared with untreated TAFs. To explore whether Calml3 secreted from gastric TAFs mediated the tumor-suppressive role of metformin, gastric cancer cells were incubated with culture medium containing different concentrations of recombinant Calml3 protein. A clonogenic assay revealed that Calml3 not only decreased the surviving fraction of gastric cancer cells, but also decreased the size of cell clones in two gastric cancer cell lines (Fig. 7A and B). These results demonstrate that Calml3 secreted from TAFs exerts a tumor-suppressive role in gastric cancer cells. Calml3 may account for the antitumor effect of metformin on the tumor microenvironment.