The NSCLC cell line, RAL, is derived from a metastatic lesion of lung adenocarcinoma of a 52-year-old female previously treated with cisplatin (14). The identity of the patient was irreversibly anonymized prior to specimen processing. The cell line is characterized by the following: EGFR-wt, KRAS mutation at exon 1 (p.G12C, missense, not functional, deleterious), TP53 mutation at exon 7 (p.G244C, missense, not functional, deleterious) and no ALK rearrangement. The cells were grown in Dulbecco's modified Eagle's medium/HAM F12 (1:1) supplemented with 10% fetal bovine serum, 2 mM of L-glutamine (EuroClone) and 10 µg/ml of insulin (Sigma-Aldrich; Merck KGaA) in a humidified atmosphere with 5% CO₂ at 37°C. The population doubling time was 65 h, as previously described (14).

Cisplatin (Ebewe Pharma-Sandoz) and pemetrexed (Eli Lilly Italia Spa) were tested at plasma peak concentrations (10 and 320.5 µM, respectively). Cell viability was assessed by MTS assay according to the manufacturer's protocol (CellTiter 96^® AQueous One Solution Cell Proliferation assay, Promega Corp.). Experiments were run in octuplicate and each experiment was repeated 3 times. To ensure that the exposure time was compatible with the half-life of the drugs administered as in a clinical setting, the cells were treated with pemetrexed or cisplatin alone and washed out with drug-free medium after 3 and 6 h of exposure, respectively, followed by up to 96 h of recovery, as previously described (15). For the combination treatment, the cells were exposed to cisplatin and pemetrexed for 6 h and washed out with drug-free medium, followed by 96 h of recovery (96 h-post wo).

Untreated cells, and those at 96-h post wo and 21 days post-treatment washout (21 days-post wo) were collected and fixed with ice-cold 70% ethanol. The cells at 21 days-post wo were allowed to recover by replacement of the fresh medium 3 times/week. No split was carried out, in order to permit the RAL cells to form resistant clones and repopulate the flask. Flow cytometric analysis was performed using a FACSCanto flow cytometer (BD Biosciences). Data acquisition and analysis were performed using FACSDiva software (Version 6.1.3, BD Biosciences). Samples were run in triplicate (10,000 events for each replica). Data report the mean value of 3 experiments with standard deviations <5%.

Cells were fixed in 1% paraformaldehyde in phosphate-buffered saline (PBS) on ice for 15 min, suspended in cold ethanol (70%) and stored overnight at −20°C. The cells were then washed twice in PBS, resuspended in PBS containing 0.1% Triton X-100 for 5 min at 4°C, incubated in 50 µl of solution containing TdT and FITC-conjugated dUTP deoxynucleotides 1:1 (Roche Diagnostics GmbH) in a humidified atmosphere for 90 min at 37°C in the dark, washed in PBS, counterstained with propidium iodide (PI, 200 µg/ml, MP Biomedicals) and RNase (10,000 U/ml, Sigma-Aldrich; Merck KGaA) for 30 min at 4°C in the dark, and analyzed by flow cytometry.

The cells were fixed in ethanol (70%) and stained in a solution containing 100 µg/ml of PI (Sigma-Aldrich; Merck KGaA), 10,000 U/ml of RNase (Sigma-Aldrich; Merck KGaA) and 0.01% NP40 (Sigma-Aldrich; Merck KGaA). After 24 h, the samples were analyzed by flow cytometry. Data were elaborated using ModFit (DNA Modelling System) software (Version 4.1.7, Verity Software House Inc.), and expressed as fractions of cells in the different cell cycle phases.

The cells were washed once in PBS, incubated with 250 µl of Annexin V-FITC in binding buffer (eBioscience) for 15 min at 37°C in a humidified atmosphere in the dark, and then washed in PBS and suspended in binding buffer. Immediately before flow cytometric analysis, PI was added to a final concentration of 5 µg/ml to distinguish between total apoptotic cells (Annexin V⁺ and PI⁻ or PI⁺) and necrotic cells (Annexin V⁻ and PI⁺).

CDKN1A Silencer Select Validated siRNA (Thermo Fisher Scientific, Inc.) and a validated Negative Universal Control™ (Invitrogen; Thermo Fisher Scientific, Inc.) were used. The cells were seeded in 25 cm²-flasks at a density of 2.5×10⁵ cells. The transfections were carried out using the TransIT-X2 Dynamic Delivery System (Mirus Bio LCC) and Opti-MEM GlutaMax medium (Invitrogen; Thermo Fisher Scientific, Inc.) without antibiotics. siRNA/TransIT-X2 was incubated with the cells for 25 min. The total incubation time before the combinational treatment was 24 h.

Cells (7×10⁶) that had either been treated with cisplatin and pemetrexed or under the control condition were washed in hypotonic buffer (10 mM Tris, 5 mM MgCl₂). Triton X-100 (10% stock) was added to a 0.3% final concentration. The suspension was passed twice through a 22-gauge needle with a syringe and centrifuged at 800 × g for 10 min at 4°C to remove debris. The supernatant was collected. Protease/phosphatase inhibitor cocktail (Sigma-Aldrich; Merck KGaA) was added and protein lysate was saved as the cytosolic fraction. Nuclei were isolated following 2 subsequent centrifugations at 800 × g for 10 min at 4°C and lysed in RIPA buffer containing 1 µg/ml of DNAse (Voden Medical Instruments). Western blot analysis was performed as previously described (16). The list of the specific antibodies and conditions used is available upon request.

Total RNA was extracted using TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.) and purified with the RNeasy MinElute Cleanup kit (Qiagen GmbH) according to the manufacturer's instructions. DNase enzyme digestion (Qiagen GmbH), was performed to exclude genomic DNA contamination. Purified RNA was eluted with 14 µl of RNase-free water (Qiagen GmbH). A total of 500 ng of purified RNA were reverse transcribed into cDNA using the RT² First Strand kit (SABiosciences Corp.) in a volume of 20 µl reaction and diluted with 91 µl of nuclease-free water, following the manufacturer's instructions.

The Human Oncogenes and Tumor Suppressor genes RT² Profiler™ PCR Array (complete list of genes is presented in Table SI) was used to profile the expression of 84 different key genes that promote oncogenesis. A mix containing 102 µl of cDNA, 1,248 µl of water and 1,350 µl of 2X RT² SYBR Green Master Mix (SABiosciences Corp.) was prepared and 25 µl were added to each well. qPCR was performed using the Real-Time Cycler 7500 (Applied Biosystems) with the following thermal profile: 95°C for 10 min; 40 cycles: 95°C for 15 sec and 60°C for 1 min; dissociation curve at 95°C for 1 min, 55°C 30 sec and 95°C for 30 sec. Raw data from qPCR were exported and analyzed using online software at the GeneGlobe Data Analysis Center website (Qiagen GmbH). Fold changes in gene expression were calculated using the 2^−ΔΔCq method (17).

Differentially expressed genes (with >2-fold changes in expression) were validated with 3 biological independent repli-cates by RT-qPCR (Table SIIA), using 1 µl of cDNA, 0.4 µM of specific primers, 12.5 µl of RT² SYBR-Green Master Mix and water up to a final volume of 25 µl. The relative gene expression was calculated as previously mentioned.

A protein-protein interaction (PPI) network and functional module were established using the Search Tool for the Retrieval of Interacting Genes (STRING) (18). A confidence score of ≥0.4 and a maximum number of interactors of (=) 0 were set as the cut-off criteria.

CDKN1A and RASSF1 gene promoter methylation was analyzed by bisulfite sequencing (BS). A total of 1 µg of genomic DNA was bisulfite-treated using the EZ DNA Methylation Kit (Zymo Research), which induces the chemical conversion of unmethylated cytosine to uracil, and eluted in 20 µl of elution buffer. A total of 3 µl of bisulfite-treated DNA were amplified by PCR in a total volume of 20 µl, with 4 mM of MgCl₂, 0.5 mM of each dNTP, 0.2 µM of each primer, and 0.5 unit of Taq Hot Start Thermostable DNA polymerase (EuroClone). The primers were designed using the Methyl Primer Express^® Software v1.0 (Applied Biosystems) and their sequences are presented in Table SIIB. Amplifications were performed using the following thermal profile: 95°C for 2 min followed by 38 cycles at 95°C for 30 sec, 59°C for 30 sec and 72°C for 30 sec. The PCR products were purified from agarose gel using the EuroGold Gel Extraction kit (EuroClone) and cloned using the pGEM-T Easy Vector System (Promega Corp.) following the manufacturer's instructions. In total, 10 white colonies for each time point were resuspended in 50 µl of water and tested by colony PCR using the same PCR conditions as before, adding an extra step at 95°C for 10 min to induce cell lysis. PCR products were then purified by QIAquick PCR Purification kit (Qiagen GmbH). Sequencing of PCR products was performed using the Big Dye Terminator Cycle Sequencing kit (Applied Biosystems) on the Applied Biosystems 3130 Avant Genetic Sequencer.

ChIP assay was performed as previously described (19). Washes of immuno-precipitated (IP) samples were carried out for 1 min using the HulaMixer^® Sample Mixer (Thermo Fisher Scientific, Inc.), setting the rotation as follows: Orbital (rpm) 58-1, Reciprocal (deg) 13°-01, Vibro (pause) 5° off. IP purified DNA and inputs were amplified using genomic primer sets specific for RASSF1 and CDKN1A gene promoters designed to overlap the regions investigated by BS (Table SIIB). RT-qPCR was performed in a total volume of 20 µl including 5 µl of IP-DNA, 0.1 µM of each primer and 14 µl of SsoAdvanced™ Universal SYBR^®-Green Supermix (Bio-Rad Laboratories, Inc.). Amplification was carried out in the CFX Connect™ Real-Time PCR Detection System (Bio-Rad Laboratories, Inc.) using a two-step protocol consisting of 95°C for 3 min, 95°C for 10 sec and 64°C for 30 sec, 40 cycles, followed by a melting curve step to confirm the specificity of the PCR products. ChIP data were analyzed using the Percent Input Method: input Ct values corresponding to 1% of initial chromatin were adjusted to 100% and used to normalized Ct values of IP-DNA samples.

Cell blocks were prepared by fixing untreated cells, and cells at 96-h post wo and 21 days- post wo in 10% neutral-buffered formalin for 24 h and then embedding them in paraffin using Bio Agar gel (Bio Optica). Sections 4-µm-thick were cut and placed on positive-charged slides (Bio Optica). Immunostaining was performed with the Ventana Benchmark XT system (Ventana Medical Systems, Tucson, AZ) and the Optiview DAB Detection kit (Ventana Medical Systems), an indirect, biotin-free system for mouse IgG, mouse IgM and rabbit primary antibodies detection. The kit aims to identify targets by immunohistochemistry (IHC) and ICC in formalin-fixed, paraffin-embedded sections that are stained on the Ventana automated slide stainers and visualized by light microscopy. Incubation time and temperature are set by the stainer automatically. Antibodies against CDKN1A (DCS-60.2, Cell Marque, Ventana Medical Systems), thymidylate synthase (TS; clone 4H4B1, Life Technologies; Thermo Fisher Scientific, Inc.) and the DNA excision repair cross-complementation group 1 (ERCC1; clone 8F1, Neomarkers) were used. The CDKN1A antibody was pre-diluted by the supplier and incubated at room temperature for 16 min. For TS and ERCC1 detection, antibodies diluted at 1:300 and 1:100 were used, respectively, and incubated at room temperature for 32 min. Sections were automatically counterstained at room temperature for 16 min with hematoxylin II (Ventana Medical Systems). Hematoxylin and eosin staining, at room temperature for 16 and 1 min respectively, was also performed for each sample. Biomarker expression was semi-quantitatively analyzed as the percentage of immunopositive tumor cells out of the total number of tumor cells. All samples were evaluated by a trained pathologist and analyzed using an upright light microscope in bright field (Axioscope, Zeiss).

Three biological replicates for each time-point (untreated cells, and cells at 96 h-post wo and 21 days-post wo) were analyzed for each experiment and the results were reported as the mean values and standard deviation (SD) or standard error (SEM). Time points were compared using the non-parametric Friedman's test followed by the Bonferroni post-hoc test. Reported P-values were two-sided and P<0.05 was used as the threshold for significance. Correction for multiple testing was not performed due to the explorative aim of this analysis. Statistical analyses were carried out using STATA/MP 15.0 for Windows (Stata Corp. LP).

The anti-proliferative effects of cisplatin and pemetrexed used as single agents or in combination were evaluated using the RAL cells by MTS assay at 24, 48, 72 and 96 h after drug washout. The RAL cells exhibited no or only slight sensitivity to pemetrexed and cisplatin, with 100 and 71% of cell survival at 96 h, respectively. Although a higher cytotoxic effect was observed at the end of the exposure time to both drugs, the half-maximal inhibitory concentration (IC₅₀) value was never reached (64% of cell survival at 96 h-post wo) (Fig. 1A).

TUNEL assay (Fig. 1B) revealed a percentage of apoptotic cells consistent with that of non-surviving cells detected by MTS assay when the two drugs were used in combination. In particular, a significant increase in the number of apoptotic cells was observed at 96 h-post wo (44.6%, P=0.002). Notably, the cells had fully recovered 21 days after the end of treatment and began to grow and divide normally. Cell cycle distribution analysis (Fig. 1C) revealed a significant decrease in the number of cells in the G₀G₁ phase (1.7%) and a marked increase in the number of cells in the G₂M phase at 96 h-post wo (76%), whereas no differences with respect to the untreated samples were detected in the cells in the S phase. These cell cycle perturbations had fully recovered at 21 days-post wo when the cell percentages in the G₀G₁ or G₂M phase were similar to those of the untreated cells (56.7% vs. 56.5%, and 18.2% vs. 17.1%, respectively).

The expression of 84 oncogenes and tumor suppressor genes (TSGs) was analyzed in the untreated cells, and in the cells at 96 h-post wo and 21 days-post wo. In total, 11/84 genes were not expressed (ESR1, FOXD3, HGF, KIT, MOS, MYCN, ROS1, RUNX3, TNF, TP73 and WT1) and 2/84 were unevaluable (BCL2 and WWOX). Among the genes included in the panel, 8 (CDKN1A, CDKN3, EGF, FHIT, JAK2, MYC, RASSF1 and RB1) and 3 (CDKN1A, CDKN3 and RASSF1) were differentially expressed by >2-fold in the cells at 96 h-post wo and 21 days-post wo, respectively (Fig. 2A and B). In particular, RT-qPCR analysis of the cells at 96 h-post wo confirmed that CDKN1A (P= 0.002), EGF (P= 0.001) and JAK2 (P=0.009) were significantly upregulated with a >2 fold change, whereas FHIT was downregulated (P=0.008). A significant increase in CDKN1A mRNA expression was also maintained and confirmed in the cells at 21 days-post wo (P=0.011) (Fig. 2C). The STRING database used to visualize protein-protein interaction (PPI) revealed a network with high degree of connectivity between the differentially expressed genes, KRAS and AKT, with a PPI enrichment P-value equal to 0.00106 (Fig. S1).

Subsequently, it was investigated whether epigenetic modifications (DNA methylation and histone marks) are associated with gene expression, by performing BS and ChIP on the RAL cells at 96 h and 21 days following the combined treatment. The results for the CDKN1A gene were compared with those obtained for RASSF1, a tumor suppressor gene often silenced and hypermethylated in RAL cells (15). The promoter regions of CDKN1A and RASSF1 are shown in Fig. 3A and B, respectively.

DNA methylation analysis was performed by BS in 10 clones corresponding to the untreated cells, and cells at 96 h-post wo and 21 days-post wo. The methylation percentage of each cytosine was calculated by the average of the methylation status of the 10 clones. The promoter region of the CDKN1A gene was completely unmethylated (data not shown), whereas the RASSF1 gene promoter exhibited a hypermethylated (>40%) CpG island (Fig. 3C). No significant differences were detected among the treated and untreated cells.

Three post-transcriptional histone modifications were investigated by ChIP assay: Two of these were associated with transcriptional active chromatin, i.e., the acetylated form of the H3 histone tail (acH3) and trimethyl-Lysine 4 of H3 histone tail (H3K4me3), and one enriched in transcriptional silenced chromatin domains, trimethyl-Lysine 27 of H3 histone tail (H3K27me3). The chromatin histone marks (acH3 and H3K4me3) corresponding to a transcriptionally open chromatin region were ≥7-fold enriched in the promoter region of the CDKN1A gene compared to RASSF1 (Fig. 3D and E). Conversely, the repressive histone mark H3K27me3 showed comparable results in the two genes (Fig. 3F). Isotype rabbit IgG was used as the background control (Fig. 3G). No substantial differences between the untreated cells and cells surviving the combination treatment were detected in histone modifications associated with the promoters of both genes, with the exception of a significant 4-fold increase in acH3 enrichment in the CDKN1A promoter detected in the cells at 96 h-post wo with respect to the untreated cells, and cells at 21 days-post wo (Fig. 3D).

An upregulation of CDKN1A protein expression was found in the cells at 96 h-post wo with respect to the controls (85 and 10% immunopositive cells, respectively), whereas CDKN1A expression in the cells at 21 days-post wo was equal to baseline (5% immunopositive cells) (Fig. 4A). In particular, it was observed that 10% of cells at 96 h-post wo exhibited a strong cytoplasmic expression (Fig. 5A) that was virtually absent in the untreated cells, and in cells at 21 days-post wo.

The results of western blot analysis of whole lysates (Fig. S2) were in line with the ICC results, confirming that the cells at 96 h-post wo exhibited a marked upregulation of CDKN1A protein expression, while in the cells at 21 d-post wo, the protein level decreased with respect to the cells at 96 h-post wo.

Moreover, western blot analysis of protein lysates from subcellular compartments revealed that CDKN1A overexpression observed following treatment was mainly related to a protein increment in the cytoplasmic fraction. Again, the increased CDKN1A expression was more evident in the cytoplasm of the cells at 96 h-post wo (Fig. 5B).

We also characterized the RAL cells for the expression of TS and ERCC1, known to be potential markers of response to pemetrexed and cisplatin, respectively. The RAL cells exhibited a high basal nuclear and cytoplasmic expression of TS, 80% of which were strongly immunopositive. Strong TS stain intensity was also observed in the cytoplasm (Fig. 4B). An additional upregulation of cytoplasmic TS was observed in the cells at 96 h-post wo, whereas only 40% of the cells were immunopositive at 21d-post wo. As regards ERCC1 protein expression, almost 100% of the untreated cells and cells at 96 h-post wo were strongly positive, whereas the cells at 21 days-post wo exhibited a somewhat lower positivity (75%) (Fig. 4C). The TS and ERCC1 protein expression levels were also confirmed by western blot analysis (Fig. S2). In all the immunocyto-chemical evaluations, the cells at 96 h-post wo appeared larger than their untreated counterparts, according to the well-known acute effects of cisplatin treatment (20,21).

The role of CDKN1A gene in the cellular response to cisplatin and pemetrexed was investigated by transient silencing. The CDKN1A mRNA levels decreased significantly with respect to those observed in cells transfected with an appropriate negative control (scrambled oligonucleotide), confirming CDKN1A knockdown by siRNA. Moreover, its protein level was undetectable in cells transfected with CDKN1A siRNA (Fig. S3).

The RAL cells in which CDKN1A was silenced and which were exposed to the cisplatin-pemetrexed combination and evaluated 96 h after washout exhibited a consistently significant increase in cell death compared to the scramble control-transfected cells at 96 h-post wo. TUNEL assay revealed that there were 73.6% of apoptotic cells when CDKN1A was knocked down compared with 44% of apoptotic cells in the untreated cells transfected with CDKN1A siRNA (Figs. 6A and S4). As was expected, when scramble RNA was used instead of the CDKN1A siRNA probes, the cisplatin-pemetrexed combination induced a considerable increase in cell death compared to the untreated control conditions (40.1% vs. 22.1%, respectively) (Fig. 6A). The increase in cell death detected by TUNEL assay in the cells at 96 h-post wo with CDKN1A silencing was also confirmed by Annexin V assay, which identified both early and late apoptosis. In the 96 h-post wo condition, CDKN1A silencing was associated with 51.9% of cells in early apoptosis compared with 23.2% cells in the untreated siRNA counter-parts and 9.4% in the cells at 96 h-post wo transfected with the scramble control (Figs. 6B and S4).

Finally, a cell cycle distribution analysis revealed that the cisplatin-pemetrexed combination led to a significant G2 phase block (Figs. 6C and S4) that was more evident in the cells at 96 h-post wo than in the untreated cells transfected with either CDKN1A siRNA (58.6% in 96 h-post wo vs. 10.9% in untreated cells) or scramble RNA (56% in 96 h-post wo vs. 18.2% in untreated cells). Furthermore, as CDKN1A inhibits the activity of cyclin-CDK2, -CDK1 and -CDK4/6 complexes, functioning as a regulator of cell cycle progression during G1 and S phases, a significant increase in the cells in the G1-phase was also detectable in the cells transfected with siRNA at 96 h-post wo with respect to their scramble control counter-parts (11.7 and 2.4%, respectively). No significant differences in percentages of S phase cells were detected in the untreated cells. Treatment with the cisplatin-pemetrexed combination caused an increase in the number of cells in the S phase (41.6% vs. 28.3% in the cells at 96 h-post wo and untreated cells, respectively). When CDKN1A was knocked down, the percentage of cells in the S phase decreased to the level of the untreated cells (29.7%).