A549 cells were maintained in HAM-F12 medium (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% FBS (Gibco; Thermo Fisher Scientific, Inc.), penicillin (100 U/ml) and streptomycin (100 µg/ml). Cells were maintained at 37°C in a humidified atmosphere containing 5% CO₂. Cells were treated at 37°C with 10 µM cisplatin (Sigma-Aldrich; Merck KGaA) for 72 h to generate the resistant population [CIS vs. control (C)] and were exposed to 10 mM metformin (Sigma-Aldrich; Merck KGaA) (MET or CISMET) or 100 nM rapamycin at 37°C (Sigma-Aldrich; Merck KGaA) (RAPA or CISRAPA) for 24, 48 or 72 h according to different assays.

A549 cells were seeded at a density of 3×10⁴ cells/well for the C group (sensitive cells) and 9×10⁴ cells/well for the CIS population (resistant cells) in 96-well plates and treated with cisplatin for 72 h and metformin or rapamycin for another 72 h. After incubation, 12 mM MTT solution was added to each well for 2 h at 37°C. The culture medium was aspirated and the formazan crystals were solubilized with a solution of 1M HCl:isopropanol (1:25) for 15 min at 37°C, and the absorbance of each sample was measured at 570 nm.

A549 cells were seeded at low density in 6-well plates (3×10² cells/well), treated with 10 mM metformin or 100 nM rapamycin for 72 h and incubated for 7 days at 37°C. After incubation, cells were washed with PBS and stained with 3 ml methylene blue dye (0.3% in 50% ethanol) for 30 min at room temperature. The plates were washed with deionized water and staining was measured at an absorbance of 590 nm by eluting with 10% acetic acid. The size of the colonies was calculated using ImageJ software v1.53 (National Institutes of Health). Colonies >0.01 pixel² were considered as hits by the ImageJ software and were then manually divided into size ranges using Excel (Microsoft Corporation). The size ranges were hits <1 pixel², between 1 and 25 pixel² and >25 pixel². Images of the cells for morphology analysis were captured using a light microscope coupled to a camera (magnification, ×40; Leica Microsystems, Inc.), using scale bars of 100 µm.

A549 cells were seeded in 6-well plates at a density of 3×10⁵ cells/well and treated with 10 µM cisplatin for 72 h and 10 mM metformin or 100 nM rapamycin for 24 h for cell size analysis or 72 h for cell cycle analysis. After treatment, cells were washed with 1X PBS (0.137 M NaCl and 0.05 M NaH₂PO₄, pH 7.4) and resuspended in 500 µl HAM-F12 medium. Cells were centrifuged at 300 × g for 5 min at room temperature and resuspended in 100 µl propidium iodide (PI) solution (0.1% Triton X-100, 20 µg/ml PI and 10 µg/ml RNAse in PBS) for cell cycle analysis or 100 µl PBS for cell size analysis. The PI fluorescence and cell profile were determined using a BD Accuri™ C6 flow cytometer and analyzed using the BD Accuri™ software v1.0.264.21 (BD Biosciences).

Protein extracts obtained by lysing cells (50 mM Tris-Cl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, protease inhibitor cocktail and phosphatase inhibitor cocktail) were quantified using a bicinchoninic acid assay and samples containing 20-40 µg of protein were separated by SDS-PAGE (8 and 10% gels). The gels were electrotransferred to 0.45-µm nitrocellulose membranes (Bio-Rad Laboratories, Inc.) and the membranes were incubated for 2 h at room temperature with 5% non-fat powdered milk dissolved in TBS-Tween (TBST; 50 mM Tris-Cl, pH 7.5; 150 mM NaCl; 0.1% Tween-20) to saturate unspecific binding sites, followed by an overnight incubation at 4°C with primary antibodies. Membranes were washed 3 times with TBST and then incubated with HRP-conjugated goat anti-mouse IgG (Sigma-Aldrich; Merck KGaA; cat. no. AP308P; 1:2,000) and goat anti-rabbit IgG (Sigma-Aldrich; Merck KGaA; cat. no. AP307P; 1:5,000) secondary antibodies for 1 h at room temperature. Protein bands were visualized using Pierce™ ECL Western Blotting Substrate (Thermo Fisher Scientific, Inc.) and densitometry was performed using ImageJ software v1.53. Primary antibodies (all 1:2,000) against mTOR (cat. no. 2972), phospho-mTOR (Ser2448; cat. no. 2971), AMPKα (cat. no. 5831), phospho-AMPKα (Thr172; cat. no. 50081), p70-S6K1 (cat. no. 2708), phospho-p70-S6K1 (Thr389; cat. no. 9234), S6 (cat. no. 2317), phospho-S6 (Ser240/244; cat. no. 2215) and GAPDH (cat. no. 2118) were purchased from Cell Signaling Technology, Inc.

Total RNA was extracted from A549 cells using TRIzol^® (Invitrogen; Thermo Fisher Scientific, Inc.). The cDNA was synthesized using the High Capacity cDNA Reverse Transcription kit with 2,000 ng of total RNA according to the manufacturer's protocol (Thermo Fisher Scientific, Inc.). The qPCR reaction was performed with SYBR Green PCR Master Mix (Applied Biosystems; Thermo Fisher Scientific, Inc.). qPCR was performed with an initial denaturation step at 95°C for 10 min and then 40 cycles of 95°C for 15 sec and 60°C for 1 min. A melting curve was performed after the PCR from 60°C to 95°C. The 2^−ΔΔCq method was used for quantification (34). β-actin was used as the normalizing gene. Samples for this reaction were added in triplicates in a 96-well plate (MicroAmp; Applied Biosystems; Thermo Fisher Scientific, Inc.) for amplification and reading in the Step One Plus Real-Time PCR System (Applied Biosystems; Thermo Fisher Scientific, Inc.). The primers used were designed using Primer-BLAST (35) (National Institutes of Health): Superoxide dismutase 2 (SOD2) forward, 5′-AAG GAA CGG GGA CAC TTA CAA A-3′ and reverse, 5′-AGC AGT GGA ATA AGG CCT GTT G-3′; Annexin 4 (ANXA4) forward, 5′-CAG AGG AAC AAC CAG GAA CTT G-3′ and reverse, 5′-CAA GCA GAA GTT CTT CGA GGC-3′; and β-actin forward, 5′-GCC GCC AGC TCA CCA T-3′ and reverse, 5′-CCA CGA TGG AGG GGA AGA C-3′.

For heavy (H)- or light (L)-lysine labeling experiments, A549 cells were maintained in a T25 flask with SILAC™ HAMF12 medium (Thermo Fisher Scientific, Inc.) supplemented with 10% dialyzed FBS without lysine and arginine. To obtain H and L conditions, SILAC™ HAMF12 medium was supplemented with ¹³C6 L-lysine-2HCl (H) or ¹²C6 L-lysine-2HCl (L) and ¹²C6 L-arginine-2HCl (H and L). The final concentration of amino acids were 0.46 mM (lysine) and 0.47 mM (arginine). After 5 passages, C and CIS cells were treated in triplicate with 10 mM metformin for 72 h (CxMET representing the Ctrl population; CISxCISMET representing the CIS population), washed two times with PBS in a 6-well plate and collected using 100 µl lysis buffer as aforementioned.

The mixed H-/L-labeled protein sample (27.5 µg of total protein per group) was separated by electrophoresis in reducing 10% SDS-PAGE, heated for 10 sec in a microwave in Coomassie Brilliant Blue stain (Sigma-Aldrich; Merck KGaA; 2% Coomassie in 50% ethanol and 2.5% acetic acid) and incubated for 1 h at room temperature. After overnight incubation of the gel for background de-staining, each gel lane was sliced into 10 pieces. The excised sections were unstained twice with a de-staining solution (50% ethanol and 2.5% acetic acid in purified water) in 1.5-ml microtubes to start the gel bands digestion protocol using trypsin. The percentage of isotopic label incorporation was tested according to the following equation: (Ratio H/L x 100)/(Ratio H/L + 1) (36).

The protein bands were reduced, alkylated and digested with trypsin overnight at 37°C. An aliquot of the peptide mixture was separated using a 2-40% acetonitrile gradient in 0.1% formic acid using an analytical PicoFrit Column (20 cm x ID75 µm; 5-µm particle size; New Objective, Inc.), at a flow rate of 300 nl/min over 35 min. Peptides were analysed using the EASY-nLC II (Proxeon Biosystems) coupled to LTQ Orbitrap Velos (Thermo Fisher Scientific, Inc.), with electrospray ionization in positive mode and set up in the data-dependent acquisition mode. Full scan MS spectra (m/z 300-1,600) were acquired in the Orbitrap analyzer after accumulation to a target value of 1×10⁶. Resolution in the Orbitrap was set to r=60,000, and the 20 most intense peptide ions with charge states ≥2 were sequentially isolated to a target value of 5,000 and fragmented in the linear ion trap by low-energy collision-induced dissociation (normalized collision energy of 35%). Three independent experiments were performed.

For SILAC data analysis, the raw files were processed using MaxQuant 2012 version 1.3.0.5 (https://www.maxquant.org) and the MS/MS spectra were searched using the Andromeda search engine against the Uniprot Human Protein Database (37) (release 17 February 2016; 91,974 sequences; 36,693,332 amino acid residues). The initial maximal allowed mass tolerance was set to 20 ppm for precursor, 6 ppm in the main search afterward and then to 0.5 Da for fragment ions. Enzyme specificity was set to trypsin with a maximum of two missed cleavages. Carbamidomethylation of cysteine (57.021464 Da) was set as a fixed modification, and oxidation of methionine (15.994915 Da) and protein N-terminal acetylation (42.010565 Da) were selected as variable modifications. The minimum peptide length was set to 6 amino acids, including heavy label Lys6. For protein quantification, a minimum of two ratio counts was set and the 'requantify' and 'match between runs' functions were enabled. The false discovery rates (FDRs) of peptide and protein were both set to 0.01. Data processing was performed using Perseus v.1.2.7.4 available in the MaxQuant database (https://www.maxquant.org/). First, reverse and contaminant entries were excluded from further analysis. A protein ratio intensity between H and L was used to compare differential protein expression in the total extract from C and MET cells, and CIS and CISMET treated cells. The protein ratios were calculated from the median of all normalized peptide ratios using unique peptides or peptides assigned to the protein group with the highest number of peptides (razor peptides).

For statistical analysis of differentially expressed proteins, the ratios were converted into log2 and Student's unpaired t-test was applied on the C and CMET treated groups and CIS and CISMET treated groups. All MS raw files and search parameter settings associated with the present study are available for download via the PRIDE data repository at https://www.ebi.ac.uk/pride/archive/ (accession no. PXD017645). After the t-test, volcano plots were constructed to each replicate separately using the Volcano Plot Plugin of the OriginLab 2019b software (v9.65; OriginLab) considering P<0.05[Log₁₀(0.05)=1.3010] and log2 ratio >±1.0 (fold change of CxMET and CISxCISMET ratio H/L normalized) as exclusion parameters. Heatmaps were generated on Morpheus online (https://software.broadinstitute.org/morpheus) considering common proteins in both groups with P<0.05. The heatmap represents the standardized values to robust z-score clustered by Euclidian distance using the average linkage method. Proteome networks were generated using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) v11 (https://string-db.org/), using the significant proteins list after t-test analysis and setting a High Confidence value of 0.700 as a parameter. FDRs of Gene Ontology (GO) classification were calculated using STRING v11 for biological processes and performed as previously described by Szklarczyk et al (38). Biological processes were arbitrarily chosen so as not to overlap redundant classifications and obtain the largest number of classified targets. The FDR of GO and the P-value of the protein-protein interaction (PPI) enrichment were automatically calculated by STRING and the detailed description of the enrichment algorithm has been previously described (39).

Survival rates of patients with lung adenocarcinoma and lung squamous cell carcinoma were assessed using The Cancer Genome Atlas Pan-Lung Cancer database (40) (dbGaP Study Accession: phs000488.v1.p1) for ANXA4 and SOD2 in the cBioPortal software v3 5.0 (https://www.cbioportal.org/) (41). Each sample was defined as altered or unaltered for each gene based on the Onco Query Language, which considers non-synonymous mutations, fusions, amplifications and deep deletions (https://www.cbioportal.org/oql).

Statistical analyses were assessed using GraphPad Prism 8.01 software (GraphPad Software, Inc.) applying Student's unpaired t-test or one-way ANOVA followed by Tukey's post-hoc test when n sample was homogeneous between groups and Bonferroni's test when not meeting that condition. P<0.05 was considered to indicate a statistically significant difference. For proteomics analysis, Student's t-tests were performed using Perseus.

To demonstrate that 10 µM cisplatin was able to create resistance in A549 cells, an MTT assay was performed to calculate the IC₅₀. This cisplatin concentration was adopted since it had been previously reported to confer resistance to new cisplatin exposure in melanoma (42). A549 cells were subjected to 10 µM cisplatin treatment for 72 h (called CIS population or CIS pop) and subsequent doses of cisplatin (5, 10, 15, 20, 25, 35 and 50 µM) for another 72 h. The CIS pop presented a significant increase in the IC₅₀ compared with the untreated control population (Ctrl pop), suggesting the acquisition of resistance (21.24 µM for Ctrl pop vs. 32.86 µM for CIS pop; Fig. 1A and B).

To investigate the effects of MET in lung cancer cells, MTT assay was performed (Fig. 2B) in the CIS pop compared with the Ctrl pop. MET decreased cell viability more extensively in the Ctrl pop (71.2% for 24 h and 64.2% for 72 h exposure) compared with in the CIS pop (84.5% for 24 h and 83.5% for 72 h) (Fig. 2B). Additionally, the viability of cells treated with MET was significantly decreased compared with that of cells treated with rapamycin (RAPA) in the Ctrl pop after 72 h (64.2 vs. 95.5%; Fig. 2B).

Cisplatin led to increased cell size and granularity, as seen by light microscopy and forward scatter and side scatter flow cytometry parameters (Fig. 2A and C, respectively), conferring more heterogeneity to the CIS pop compared with the Ctrl pop. In the CIS pop, flow cytometry data revealed that RAPA significantly decreased cell size after 24 h, while MET significantly decreased cell size after 72 h treatment, both compared with cisplatin alone (Fig. 2D). However, only RAPA led to G₀/G₁ cell cycle arrest after 72 h treatment (Fig. 3A and B). The present data suggested that 72 h of MET treatment decreased cell viability and size in cisplatin-sensitive and -resistant cells without cell cycle impairment. In summary, MET decreased lung cancer cell viability and size without significantly changing the cell cycle.

The mTOR signaling pathway status was evaluated by western blotting after MET treatment in Ctrl and CIS pops. The analysis revealed that the mTOR signaling pathway was overactivated after cisplatin treatment, with a significant increase in S6K1 and S6 phosphorylation (Fig. 4A, D and E) compared with the control group. Subsequent MET treatment significantly decreased mTOR, S6K1 and S6 activation compared with CIS treatment, corroborating the aforementioned decreases in cell viability and size. A non-significant difference was observed in AMPK phosphorylation after MET treatment in A549 cells (Fig. 4C). The present results indicated robust activation of the mTOR signaling pathway after cisplatin treatment in A549 cells.

To determine the colony formation potential of MET- and RAPA-treated cells, A549 cells exposed to 10 mM MET or 100 nM RAPA for 24 or 72 h were evaluated in a clonogenic cell assay (Fig. 5A). Compared with the control group, MET and RAPA treatments significantly decreased the absorbance of colonies formed (MET at 72 h and RAPA at 24 and 72 h) compared with the control group (Fig. 5B), but only RAPA decreased the size of colonies compared with the control group (Fig. 5C). Both MET and RAPA significantly decreased the IC₅₀ in Ctrl pop cells (IC₅₀=15.4 and 14.5 µM, respectively, vs. 20.4 µM for Ctrl; Fig. 5D and E). Thus, the decrease of IC₅₀ indicated that previous treatment with MET or RAPA may potentiate and sensitize cells to cisplatin treatment.

In addition to the mTOR signaling pathway, MET exerts important changes in crucial pathways involved in cancer (22). To investigate other possible molecular mechanisms involved in cisplatin sensitivity induced by MET, a proteomics analysis of A549 cells after MET treatment was performed, and the proteomes in control and cisplatin-resistant cells were compared using the SILAC approach (CxMET and CISxCISMET; Fig. 6A). A total of 903 proteins were quantified in CxMET and 646 in CISxCISMET, with 511 common proteins (Fig. 6B). The Student's t-test analysis indicated 361 differentially expressed and statistically significant proteins (P<0.05) for CxMET ratios and 254 for CISxCISMET ratios (Tables SI and SII). These proteins were classified and grouped based on their involvement in biological processes (Fig. 6C and D). GO terms and FDRs generated by STRING analysis are presented in Table I. PPI enrichment P-values were <1.00⁻¹⁶ for CxMET and CISxCISMET networks (data not shown).

In the CxMET group, MET regulated 'viral process', 'mRNA metabolic process', 'IL-12 mediated signaling pathway', 'drug metabolic process', 'oxidation-reduction process', 'leukocyte degranulation', 'transport', 'regulation of cell death' and 'translation' (Fig. 6C). In the CISxCISMET group, MET regulated 'Golgi vesicle transport', 'protein folding', 'tRNA aminoacylation for protein translation', 'macromolecule catabolic process', 'oxidation-reduction process', 'leucocyte degranulation', 'mRNA metabolic process', 'viral process', 'regulation of cell death' and 'cytoskeleton organization' (Fig. 6D). The data revealed five shared pathways between groups ('mRNA metabolic process', 'oxidation-reduction process', 'leukocyte degranulation', 'viral process' and 'regulation of cell death'). The raw list of proteins identified by Perseus in both groups is presented in Tables SI and SII. Therefore, these molecular pathways may be potentially important signatures of the mechanisms of action of MET in lung cancer.

Volcano plots displayed the 1,448 proteins in CxMET and 1,158 in CISxCISMET (Fig. 7A), both separated by magnitude of evidence (P-value) and change (fold change of log2 ratio values) [cut-off values, log10(0.05)=1.3010; log2 ratio=1.00]. A total of 186 and 184 proteins were significantly downregulated and upregulated in MET compared with C, respectively, and 102 and 167 proteins were significantly downregulated and upregulated in CISMET compared with CIS, respectively. Applying the cut-off P-value, in CxMET the volcano plot expressed 3 significantly downregulated proteins, including ANXA4, and 12 significantly upregulated proteins, including SOD2. In CISxCISMET, the volcano plot expressed 7 significantly down-regulated proteins and 12 significantly upregulated proteins. Subsequently, 99 common significant proteins between CxMET and CISxCISMET analysis (ACADVL, ACTC1, ACTN1, ACTN4, AIFM1, AK3, AKR1C1, ALDH1A1, ALDH6A1, ANXA1, ANXA2, ANXA4, ARCN1, ARF3/ARF1, ARF4, C1orf57/NTPCR, CALD1, CALM2/CALM1, CANX, CCT8, CFL1, CPLX2, CRYAB, CS, CSRP1, CTSB, CTSD, DDX46, DYNC1H1, DYNC1LI2, ECH1, EIF4A1, EPRS, ETFA, ETFB, FARSB, FH, FKBP3, FLNA, G6PD, GAA, GAPDH, GCN1L1, GRPEL1, GSTP1, HADHA, HMGA1, HNRNPA2B1, HNRNPL, HSP90AA1, HSPA8, HSPD1, IARS, IGF2BP1, ILF2, ILF3, ITGB1, LARS, LDHA, LDHB, LRRFIP1, MARCKS, MDH2, MSN, MYH9, NCL, NPC2, NPM1, PDHB, PGD, PKM2, PLOD2, PPIF, PTBP1, PTGR1, PYCR1, RAB2A/RAB2B, RPL19, RPL3, RPLP0, RPS17, SERPINE1, SF3A1, SLC39A7, SOD2, SRI, STIP1, SUB1, SUCLG2, TCP1, TGM2, TKT, TPM4, TRAP1, TUFM, UBC, UGDH, VCL and ZYX), altered after MET treatment and expressed by standardized values, were grouped in clusters by Euclidian distance using the average linkage in a heatmap (Fig. 7B). These proteins were then classified into two different networks, according to their upregulation or downregulation in resistant cells compared with control cells (Fig. 7C and Table I). Cisplatin resistance decreased proteins associated with 'transport' and 'mRNA metabolic process', while it upregulated proteins involved in 'translation' (Fig. 7C), corroborating the aforementioned activation of the mTOR signaling pathway. Regulation of 'oxidation-reduction process' and apoptotic processes were common biological processes found to be upregulated or downregulated by cisplatin treatment (Fig. 7C).

Two proteins reported by the proteomics analysis were chosen for further validation: ANXA4, involved in apoptosis (43), and SOD2, involved in oxidative stress pathways (44), which presented decreased and increased expression levels in the CIS versus CISMET analysis, respectively (Fig. 7D). The mRNA expression levels of these targets were further evaluated by RT-qPCR, revealing that MET significantly decreased ANXA4 expression compared with control, CIS and CISMET groups (Fig. 7E). Additionally, MET significantly increased SOD2 expression compared with control and CIS groups (Fig. 7E). Furthermore, treatment with MET in the Cis pop (CISMET group) significantly decreased ANXA4 expression and increased SOD2 expression compared with the CIS group (Fig. 7E). RT-qPCR data corroborated the proteomics analysis presented in the heatmap. A complementary analysis of the survival rates of patients with lung cancer using the cBioPortal revealed that alterations in the ANXA4 gene decreased the median survival time after initial treatment from 43.9 months to 19.5 months (Fig. S1). The altered group for SOD2 is represented by 10 patients (plus 2 deceased), with 5 patients with deep deletions of the gene (homodeleted) and 5 patients with missense mutations (G141C, L176R, R123H, G126A and S127F) (data not shown). The altered group for ANXA4 is represented by 12 patients (plus 5 deceased), with 8 patients with gene amplification and 4 patients with missense mutations (W188L, G33C, L316V and Q52E) (data not shown). Overall, MET decreased ANXA4 expression and increased SOD2 expression in A549 cells, validating the presented proteomics data.