Human epithelial colorectal adenocarcinoma cell lines DLD-1 and SW480 (American Type Culture Collection) were maintained in high-glucose Dulbecco's Modified Eagle's Medium (Sigma-Aldrich; Merck KGaA) supplemented with 10% fetal bovine serum (Gibco; Thermo Fisher Scientific, Inc.), 1% pyruvate and 1% penicillin/streptomycin at 37°C in a humidified atmosphere with 5% CO₂. Unless otherwise stated, cells were grown to 70–80% confluence prior to treatment. In addition, unless otherwise stated, the cells were treated with the following concentrations of drugs: 1 µM bafilomycin A1, (Cell Signaling Technology, Inc.), 5 µM 5-FU (Sigma Aldrich; Merck KGaA), 2 µM oxaliplatin (PLIVA Lachema) and 5 mM metformin (Sigma Aldrich; Merck KGaA).

AGR2-knockout (KOAGR2) DLD-1 cells were prepared using CRISPR/Cas9 technology. Briefly, the guide RNA oligonucleotide (5′-AGAGATACCACAGTCAAACC-3′) that targets exon 2 of the human AGR2 gene (ENSE00003623642) was designed using Tools for Guide Design (zlab.bio/guide-design-resources). The guide RNA specifically targeted mRNA coding 21–27 aa of the AGR2 N-terminal region, which is important for AGR2 protein-mediated cell adhesion. The GFP-scrambled sequence (5′-AACAGTCGCGTTTGCGACTGG-3′) served as a control (25). Both sequences were cloned into a LentiCRISPR-v2 vector (cat. no. 52961; Addgene) using Esp3I restriction cloning. DLD-1 cells (1×10⁶ cells) were transfected with LentiCRISPR-v2_AGR2 or LentiCRISPR-v2_scrambled (scr). After 2 days, the cells were exposed to puromycin (2 µg/ml) and the pool of resistant cells was sorted and seeded as single colonies in 96-well plates. KOAGR2 cells were tested for AGR2 expression using western blotting. Two clones with an undetectable expression of AGR2, DLD1 KOAGR2 A9 and F5, were selected for further experiments. SW480 cells (1×10⁶) were transiently transfected using an Amaxa Nucleofector II (Lonza Group Ltd.) with 2 µg pcDNA3-AGR2 or empty pcDNA3 plasmid (Invitrogen; Thermo Fisher Scientific, Inc.), which served as a control, and were subsequently selected with G-418 (400 µg/ml; Sigma-Aldrich; Merck KGaA). A pool of resistant cells was tested for positive AGR2 expression with western blot analysis.

To determine the doubling time of the transfected cancer cell lines with manipulated AGR2 gene expression compared with untransfected cells, equal numbers of cells (5×10⁵) were seeded into the complete media and maintained under the standard conditions for 48 h. The culture medium was removed, and adherent cells were detached by 0.5% trypsin (Gold Biotechnology, Inc.) and counted using CASY Model TT cell counter (Roche Diagnostics).

Cells were washed twice with cold phosphate-buffered saline (PBS). The cells were then scraped into NET lysis buffer [150 mM NaCl, 1% NP-40, 50 mM Tris (pH 8.0), 50 mM NaF, 5 mM EDTA (pH 8.0)] supplemented with Protease and Phosphatase Inhibitor Cocktail (Sigma-Aldrich; Merck KGaA) according to the manufacturer's instructions. Protein concentration was determined using the Bradford method and 15 µg proteins were loaded onto 10% acrylamide gels. Following SDS-PAGE, the samples were transferred to nitrocellulose membranes, blocked with 5% milk diluted in PBS supplemented with 1% Tween for 1 h at room temperature and incubated overnight at 4°C with primary antibodies against phosphorylated-AMPKα at Thr172 (p-AMPK; 1:1,000; cat. no. 2535; Cell Signaling Technology, Inc.), β-actin and p62 (both 1:1,000; cat. nos. sc-47778 and sc-28359; Santa Cruz Biotechnology, Inc.), microtubule-associated proteins 1A/1B light chain 3 (LC3)-II (1:1,000; cat. no. NB100-2220; Novus Biologicals, LLC) and AGR2 (in house) (26). The membranes were washed and probed with horseradish peroxidase-conjugated anti-rabbit and anti-mouse secondary antibodies (1:1,000; cat. nos. P0217 and P0161; Dako) for 1 h at room temperature. Chemiluminescent signals were developed using a mixture of 1:1 ECL-A and ECL-B solutions [ECL-A: 0.5 M EDTA (pH 8), 90 mM coumaric acid, 1 M luminol, 200 mM Tris (pH 9.4) and ECL-B: 0.5 M EDTA (pH 8), 0.123 g NaBO₃ × 4H₂O, 50 mM CH₃COONa (pH 5)] and visualized with the SYNGENE G:BOX Chem XX6 gel doc system (Syngene; Synoptics Ltd.).

Cells were plated at a density of 250 cells/well in a 6-well plate. Following 24-h incubation at 37°C with 5% CO₂, 500 µM metformin, 5 µM 5-FU and 2 µM oxaliplatin were applied and further incubation for ~10 days was performed. The medium was removed, and the colonies were stained with 1% crystal violet at room temperature for 20 min and counted. Recombinant extracellular AGR2 (eAGR2) was prepared as described previously (27) and was used in the clonogenic assay as an extracellular protein supplied in the medium (1 ng/ml) throughout the experiment.

To analyze mitochondrial depolarization, the JC-1 probe was used. At 24 h post-treatment, ~2×10⁵ cells were harvested and washed with PBS, stained with 5 µg/ml JC-1 dye (Invitrogen; Thermo Fisher Scientific, Inc.) for 30 min at 37°C and analyzed using a BD FACSAria III flow cytometer (BD Biosciences). Viable population of cells was gated according to forward scatter (FSC) vs. side scatter (SSC). Individual cells were gated by FSC-A vs. FSC-H parameters. The ratio of green (FITC) to red (phycoerythrin) fluorescence was analyzed by FCS Express version 4 software (De Novo Software) as a loss of mitochondrial potential (Δψm). Treatment with valinomycin (25 µM; Thermo Fisher Scientific, Inc.) for 2 h at 37°C was used as a positive control for mitochondrial depolarization.

CM-H2DCFDA was used as an indicator for ROS in CRC cells. Briefly, the cells were seeded in 96-well plates (4×10³ cells/well) and incubated at 37°C overnight. The next day, the cells were treated with respective drugs diluted in Hank's Balanced Salt Solution (HBSS; Sigma-Aldrich; Merck KGaA) for 6 h at 37°C. Hydrogen peroxide (50 µM; Penta s.r.o.) served as a positive control, whereas N-acetyl cysteine (10 mM; Sigma-Aldrich; Merck KGaA) was used to block ROS production. Subsequently, the drugs were removed, and 5 µM CM-H₂DCFDA (Thermo Fisher Scientific, Inc.) in HBSS was added for 30 min at 37°C. The cells were washed twice with HBSS, and the fluorescence was examined using an Infinite 100 plate reader (Tecan Group, Ltd.).

Statistical analyses were performed using the Online Web Statistical Calculators for Categorical Data Analysis (https://astatsa.com). One-way ANOVA with post-hoc Tukey HSD calculator was used to determine statistically significant differences between the groups. For western blot analyses, the protein level changes were first normalized to β-actin and were then statistically analyzed. P<0.05 was considered to indicate a statistically significant difference.

Two clones with AGR2 gene knockout, KOAGR2 A9 and KOAGR2 F5, were prepared by CRISPR/Cas9 using DLD-1 cells (Fig. 1A). SW480 cells with no detectable AGR2 expression were used as the second model; an AGR2-knock-in SW480 cell line was prepared by stable transfection (Fig. 1A). The effects of AGR2 expression on cell proliferation were tested in DLD-1 and SW480 cells. Consistently with previous studies (28,29), the results demonstrated a positive effect of AGR2 expression on cell proliferation rate in the two models (Fig. 1B). The cells were then exposed to selected drugs to determine their effects on the phosphorylation of AMPK. It has previously been reported that the actions of metformin are attributable to AMPK (30). Metformin treatment exhibited a negligible effect on AMPK phosphorylation in DLD-1 scr cells. Treatment with metformin in combination with 5-FU or oxaliplatin induced a moderate increase in AMPK phosphorylation (Fig. 1C). Notably, a significant increase in the expression of p-AMPK was observed in both KOAGR2 clones treated with 5-FU and oxaliplatin in combination with metformin (Fig. 1C). To confirm the involvement of AGR2 in the regulation of AMPK activation, the same experiment was performed using SW480 cells. A non-significant increase in p-AMPK expression was observed in SW480 cells without AGR2 (SW480-pcDNA3) treated with the combination of metformin and 5-FU (Fig. 1D). Since both A9 and F5 KOAGR2 clones showed very similar p-AMPK/AMPK expression patterns, subsequent experiments were performed with only KOAGR2 clone A9.

The effects of metformin on the ability of cells to sufficiently form colonies, based on AGR2 expression and in response to standard chemotherapy drugs, were tested using a clonogenic assay. A significant loss in the ability of cells to develop colonies was associated with the absence of AGR2 (Fig. 2). Metformin alone exhibited no effect on the number of colonies developed by DLD-1 and SW480 cells irrespective of AGR2 expression. However, metformin sensitized DLD-1 cells producing AGR2 to 5-FU and oxaliplatin, which was reflected by significantly reduced colony formation compared with untreated ARG2-negative cells (Fig. 2A).

To investigate the impact of eAGR2 on cell chemosensitivity, recombinant eAGR2 protein was added into the culture medium (Fig. 2B). Similar trends were observed as in Fig. 2A; however, with eAGR2, twice the number of colonies was formed compared with cells cultured in medium without eAGR2. In addition, the effect of eAGR2 was stronger in AGR2-negative cells treated with chemotherapy drugs, which exhibited approximately twofold increase in the number of colonies compared with cells maintained in media without eAGR2 (Fig. 2A and B).

SW480-AGR2 cells exhibited a significant decrease in the number of colonies in response to the combination of metformin and 5-FU (Fig. 2C). The presence of eAGR2 protein approximately doubled the number of colonies (Fig. 2D). Taken together, these data support the antiproliferative activity of metformin administered in combination with standard chemotherapy and suggest a significant contribution of eAGR2 to the resistance of tumor cells to chemotherapy.

Since AMPK has been demonstrated to regulate autophagy, the effects of metformin on LC3-II cleavage were determined (Fig. 3). In agreement with the elevated p-AMPK levels in AGR2-negative cells (Fig. 1), markedly higher signals of LC3-II were detected in DLD-1 KOAGR2 cells in comparison with DLD-1 scr cells (Fig. 3), which indicated that AGR2 may attenuate autophagy. No significant induction of LC3-II was observed in cells exposed to all drugs compared to untreated cells (Fig. 3A). 5-FU treatment was even associated with a sharp decrease in LC3-II, indicating that autophagy may serve an important role in survival of these tumor cells. The role of autophagy may correspond with the high sensitivity of DLD-1 cells to 5-FU, as demonstrated in Fig. 2. Although the amount of LC3-II is clearly correlated with the number of autophagosomes, LC3-II itself is also degraded by autophagy (31). Therefore, it is important to measure the amount of LC3-II delivered to lysosomes by comparing LC3-II levels in the presence and absence of lysosomal protease inhibitors e.g. bafilomycin A1. However, the addition of bafilomycin A1 was not associated with increased LC3-II cleavage in cells treated with metformin in combination with 5-FU or oxaliplatin (Fig. 3B), which indicates that autophagy was inhibited. An alternative method for detecting the autophagic flux is the determination of p62/sequestosome-1 degradation, since p62 can bind LC3, serving as a selective substrate of autophagy (32). However, no significant decrease in p62 levels was observed in response to metformin combined with 5-FU or oxaliplatin treatments; by contrast, an increase in p62 levels was observed in response to 5-FU and oxaliplatin administered alone. Western blot analysis of SW480 cells revealed a reduction in LC3-II isoform expression in response to metformin and its combinations with 5-FU and oxaliplatin (Fig. 3C). These findings support the hypothesis that autophagy-dependent clearance of misfolded proteins in non-metformin-treated patients with T2DM may be suppressed by metformin treatment (19).

Metformin inhibits mitochondrial respiratory complex I at a cellular level, possibly leading to mitochondrial membrane depolarization and the release of ROS (33,34). Therefore, alterations in mitochondrial membrane potential were investigated in cells exposed to metformin in combination with 5-FU or oxaliplatin. JC-1 staining followed by flow cytometric analysis revealed a significant mitochondrial depolarization in response to metformin alone and in combination with oxaliplatin in AGR2-negative DLD-1 cells compared with in untreated cells (Fig. 4A-C).

The impact of metformin on ROS production was monitored using an indicator for reactive oxygen species in cells, CM-H2DCFDA. All tested drugs administered alone resulted in a low or moderate induction of ROS. However, the combined administration of metformin with 5-FU or oxaliplatin significantly increased ROS production in both DLD-1 scr and DLD-1 KOAGR2 cells. The effect was significantly enhanced in AGR2-negative cells exposed to a combination of 5-FU and metformin, whereas combined treatment with oxaliplatin and metformin reached only a marginal effect (Fig. 4D).