The human pancreatic cancer cell lines MIA Paca-2 and PANC-1 were obtained from the American Type Culture Collection. MIA Paca-2 and PANC-1 cell lines were cultured using complete DMEM (cat. no. 11965-092; Thermo Fisher Scientific, Inc.) supplemented with 10% (v/v) FBS (cat. no. MG-FBS0820; MediRay) and 2 mmol/l L-glutamine (cat. no. 20530081; Thermo Fisher Scientific, Inc.) in a humidified atmosphere with 5% carbon dioxide at 37°C.

Cells were seeded at a density of 1.2×10⁴ cells per well in a 96-well plate and incubated for 24 h. The culture medium was then replaced with 110 µl fresh medium containing hexadimethrine bromide (final concentration, 8 µg/ml) (cat. no. H9268; Sigma-Aldrich; Merck KGaA). An aliquot of 1 µl custom-designed shRNA lentiviral particles (Sigma-Aldrich; Merck KGaA) targeting the MRP5 gene was added to MIA Paca-2 and PANC-1 cells (multiplicity of infection, 2) and incubated for 20 h. MISSION^® Non-mammalian shRNA control transduction particles (cat. no. SHC002V; Sigma-Aldrich; Merck KGaA) were used as a negative control. After a 20-h incubation, the pooled population was transferred to two 75 mm Petri Dishes and the transduced cells were selected in medium containing 0.4 µg/ml puromycin (cat. no. A1113803; Thermo Fisher Scientific, Inc.). For further investigation, two clones of each cell type were randomly picked up, which were MIA Paca-2 clones 1 (M c1) and 2 (M c2) and PANC-1 clones 1 (P c1) and 2 (P c2). The shRNA target sequence is 5′-CCACATCTTCAATAGTGCTAT-3′. The sequence is unique in the human transcriptome, which only targets MRP5 mRNA and its variants. The insert sequence of the scrambled control is 5′-CCGGCAACAAGATGAAGAGCACCAACTC-GAGTTGGTGCTCTTCATCTTGTTGTTTTT-3′.

RNA extraction from cultured cells was carried out using TRIzol^® reagent (cat. no. 15596018; Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. The total RNA was quantified using a NanoDrop^® ND-1000 UV spectrophotometer (NanoDrop; Thermo Fisher Scientific, Inc.). Primer sequences as listed in Table I (33,34) and were purchased from Integrated DNA Technologies, Inc. RT-qPCR amplification and analysis were performed with a Light Cycler 480 Instrument (Roche Diagnostics) using the LightCycler^® EvoScript RNA SYBR^® Gren I Master one-step RT-qPCR kit (cat. no. 07800134001; Roche Applied Science). For each reaction, an aliquot of 5 µl RNA sample (10 ng) was mixed with 1 µl primer mix (final concentration 0.4 µM), 4 µl master mix (5× conc.) and 10 µl PCR grade water. The reaction conditions of RT-qPCR were 60°C for 15 min (reverse transcription) and 95°C for 10 min (enzyme activation), followed by 45 cycles of 95°C for 10 sec (denaturation) and 59°C for 30 sec (annealing and extension). MRP5 mRNA expression in each sample was normalised to the reference gene GAPDH and the results were analysed using the comparative threshold cycle method (35).

MRP5 expression was confirmed by surface immunostaining of the protein. Cells were harvested and aliquoted at 1×10⁶ cells into Eppendorf tubes. After fixation in 1% paraformaldehyde (PFA) (cat. no. P6148; Sigma-Aldrich; Merck KGaA) at 4°C for 15 min, the cells underwent permeabilization in 0.2% saponin (cat. no. 47036-50G-F, Sigma-Aldrich; Merck KGaA) at 4°C for 15 min, blocking in 5% bovine serum albumin (BSA) (cat. no. NZ21-69100-038; pH Scientific) at room temperature for 15 min, and staining with anti-MRP5 rat monoclonal antibody (M5II-54) (1:20 in 2% BSA; cat. no. MA1-35684; Thermo Fisher Scientific, Inc.) or Rat IgG2α Isotype Control primary antibody (1:20 in 2% BSA; cat. no. 02-9688, Thermo Fisher Scientific, Inc.) at 4°C for 1 h. The cells were then stained with goat anti-Rat IgG H&L (Alexa Fluor^® 488) secondary antibody (1:200 in 2% BSA; cat. no. ab96887; Abcam) at 4°C for 1 h. The cells were washed three times with PBS containing 0.1% NaN₃ and 0.1% saponin between each step. After secondary antibody incubation, the cells were then washed and resuspended in 500 µl 1% PFA. Fluorescence was detected at 488 nm excitation and 525 nm emission wavelength using a flow cytometer (MoFlo XDP; Beckman Coulter, Inc.). The mean fluorescence intensity was determined using Kaluza Flow cytometry software version 2.2.1 (Beckman Coulter, Inc.).

Steady-state cellular accumulation of an MRP5 substrate, 2′,7′-bis(2-carboxyethyl)-5(6)-carboxyfluorescein acetoxymethyl ester (BCECF-AM) (cat. no. B8806; Thermo Fisher Scientific, Inc.) was determined using a flow cytometer (MoFlo XDP; Beckman Coulter, Inc.). An MRP5 inhibitor, benzbromarone (cat. no. B5774; Sigma-Aldrich; Merck KGaA), was used as a positive control to validate the function of MRP5. MIA Paca-2 and PANC-1 cells were washed with FBS- and phenol red-free DMEM (cat. no. 31053028; Thermo Fisher Scientific, Inc.) and were resuspended in the same medium to achieve a density of 0.5×10⁶ cells/ml. The accumulation of BCECF was performed by incubating 1 ml cell suspension with benzbromarone (50 µM) or the vehicle (0.1% DMSO) at 37°C for 30 min, followed by incubation with 0.25 µM BCECF-AM at 37°C for 15 min. The accumulation was terminated by adding ice-cold PBS. The cells were then washed with ice-cold PBS and resuspended in 500 µl 1% PFA. The samples were analysed with a standard laser for excitation at 488 nm and a bandpass filter at 525 nm to detect fluorescence. The mean fluorescence intensity was determined using Kaluza Flow cytometry software version 2.2.1 (Beckman Coulter, Inc.).

The gemcitabine-induced growth inhibition was first determined in 2D-cultured pancreatic cancer cells using MTT assay. Cells were seeded at a density of 5,000 cells per well in 96-well plates and incubated for 24 h to allow cells to attach to the well surface. After incubation, cells were treated with gemcitabine (cat. no. G6423; Sigma-Aldrich; Merck KGaA) at various concentrations for 72 h. Gemcitabine stock solution at a concentration of 50 mM was prepared by dissolving gemcitabine powder in Milli-Q water. After treatment, the drug solution from each well was replaced with fresh complete medium, followed by the addition of 10 µl 12 mM MTT (cat. no. M2128; Sigma-Aldrich; Merck KGaA) stock solution. The cells were then incubated for 4 h at 37°C before adding 150 µl DMSO to each well. The cell viability was quantified by measuring the absorbance at 570 nm normalised to the mean absorbance of vehicle control. The IC₅₀ values for gemcitabine-induced growth inhibition were determined using nonlinear regression in PRISM^® software (version 8.0; GraphPad; Dotmatics).

For 3D culture models, MIA Paca-2 and PANC-1 clones were allowed to grow in 96-well low-adhesion plates (cat. no. CLS7007; Sigma-Aldrich; Merck KGaA) for 10 days, followed by 72-h incubation with different concentrations of gemcitabine. After incubation, the drug-containing medium was removed and the cells were stained with 100 µl 1X PrestoBlue (cat. no. A13261; Thermo Fisher Scientific, Inc.) at 37°C for 1 h and the fluorescence was determined at an excitation of 560 nm/emission of 590 nm using a SPARK^® Microplate Spectrofluorometer (Molecular Devices, LLC).

In another experiment, the cytotoxicity of gemcitabine was assessed using a 2D colony formation assay. Cells were seeded in 96-well plates at a density of 1,000 cells per well. After 1 day, cells were treated with 2.5, 5 and 10 nM gemcitabine for 3 days, followed by incubation with drug-free medium for 7 days. The medium was changed every 3 days. After incubation, cells were fixed with 4% PFA for 15 min and stained with 0.5% crystal violet (cat. no. C6158; Sigma-Aldrich; Merck KGaA) for 15 min at room temperature. The stained cells were washed with tap water and air-dried for a few mins before being analysed under a microscope.

Cells were seeded at a density of 5,000 cells per well in 96-well plates. After overnight incubation at 37°C and 5% CO₂, MIA Paca-2 and PANC-1 cells were treated with different concentrations of gemcitabine for 24 and 48 h, respectively. To terminate the drug exposure, the drug-containing medium was replaced with 100 µl of diluted CellEvent™ Caspase-3/7 detection reagents (final concentration 2 µM) (cat. no. C10427; Thermo Fisher Scientific, Inc.) and cells were then incubated for 60 min at 37 °C. The fluorescence was measured at an excitation of 467 nm/emission of 539 nm on a SPARK^® Microplate Spectrofluorometer (Molecular Devices, LLC). Images were also acquired with a Leica DMi8 inverted fluorescence microscope (Leica Microsystems, GmbH).

Cells were seeded in 6-well plates at a density of 200 cells per well, followed by a 10-day incubation at 37°C and 5% CO₂. The medium was changed every 3 days. After incubation, cells were fixed with 4% PFA for 15 min and stained with 0.5% crystal violet for 15 min at room temperature. The stained cells were washed with tap water and air-dried for a few mins before being analysed under an inverted microscope (Zeiss AG). Colonies of >50 cells were scored for survival and counted using ImageJ 1.52g software (National Institutes of Health).

Cells in complete DMEM were seeded into 6-well plates at a density of 3×10⁵ cells per well and allowed to grow until a confluent cell layer was obtained. A linear scratch was made in the plate using a sterile 200 µl pipette tip (cat. no. 1030-260-000-9; Thermo Fisher Scientific, Inc.). Cells were washed and then incubated with DMEM supplemented with 0.5% FBS to prevent apoptosis (36). Images were acquired at 0, 24, 48 and 72 h for MIA Paca-2 cells and 0, 16, 24 and 40 h for PANC-1 cells with a Leica DMi8 inverted fluorescence microscope (Leica Microsystems GmbH). Cell migration was analysed using ImageJ 1.52g software.

Cells in serum-free DMEM were seeded in the Corning^® Costar^® Transwell^® cell culture inserts (cat. no. CLS3464; Sigma-Aldrich; Merck KGaA) at a density of 3×10⁵ cells per insert. The inserts were placed in 12-well plates and the bottom wells were filled with 1 ml complete medium. The cells were incubated at 37°C and 5% CO₂ for 48 h. After this incubation, the non-migrated cells on the upper side of the membrane were removed using a cotton swab. The migrated cells on the lower side of the membrane were fixed with 4% PFA and stained with 0.1% crystal violet in 2% ethanol. Images were taken with a Leica DMi8 inverted fluorescence microscope (Leica Microsystems GmbH) and analysed using ImageJ 1.52g software.

The main software and tools used for molecular docking studies were AlphaFold (https://alphafold.ebi.ac.uk/) (37,38), PubChem (https://pubchem.ncbi.nlm.nih.gov/), Openbabel (39), AutoDock 4.2 (40), Protein-Ligand Interaction Profiler (https://plip-tool.biotec.tu-dresden.de) (41), ProteinsPlus (https://proteins.plus) (40,42) and LigPlot (43). The ligands were downloaded from PubChem in SDF format. The protein structure of MRP5 was downloaded from AlphaFold. The ligands in SDF file format were converted to pdb format using Openbabel. All of the ligands were docked with their receptor molecule using AutoDock 4.2. Results were evaluated and visualised using Protein-Ligand Interaction Profiler, ProteinsPlus and LigPlot.

Data were presented by descriptive statistics as the mean ± standard deviation. Linear and non-linear regression were carried out using PRISM^® software (version 8.0; GraphPad; Dotmatics). The statistical analysis was performed by a one-way analysis of variance (ANOVA) with a Dunnett's post-hoc test or a two-way ANOVA with Sidak's post-hoc test. All data analysed using ANOVAs should meet the assumptions of equal variance and homogeneity. P<0.05 was considered to indicate statistical significance.

High MRP5 mRNA expression was observed in PANC-1 (ranked 7/23) and MIA PaCa-2 (ranked 11/32) cells compared to other pancreatic cancer cell lines based on the Wagner dataset stored in ONCOMINE (https://www.oncomine.org) (Fig. 1A). The surface protein expression of MRP5 on pancreatic cancer cells was assessed by staining with anti-MRP5 primary and control isotype IgG2a antibody. Fig. 1B and C show increased fluorescence staining with the MRP5 antibody (red colour) on MIA Paca-2 and PANC-1 cells (~three-fold) compared to the isotype control antibody (green colour), suggesting the expression of MRP5 protein on the surface of pancreatic cancer cells. Furthermore, the MRP5 efflux activity in MIA Paca-2 and PANC-1 cells was investigated by determining the cellular accumulation of a model MRP5 substrate, BCECF, in the presence or absence of MRP inhibitor benzbromarone at designated time-points. Benzbromarone significantly (P<0.0001) increased the steady-state accumulation of BCECF in both MIA Paca-2 and PANC-1 cells (Fig. 1D). Taken together, these results indicated that MIA Paca-2 and PANC-1 cells have high endogenous expression of MRP5.

Significant reduction in the MRP5 mRNA expression was seen for all knockdown clones transduced with shRNA lentiviral particles. In MIA PaCa-2 cells, the mRNA transcripts of the MRP5 gene were significantly decreased by 37% (P<0.001) and 43% (P<0.0001) in M c1 and M c2 compared to the scramble control, respectively (Fig. 2A). A similar trend was seen in PANC-1 cells, as P c1 and P c2 showed a significant (P<0.001) reduction in MRP5 mRNA levels by 38 and 46% compared to the scrambled control, respectively (Fig. 2B).

MRP5 gene knockdown clones were further confirmed at the protein level by cell surface staining. The mean fluorescence intensity that indicates the MRP5 surface expression was decreased by 17 and 22% for M c1 (P<0.05) and M c2 (P<0.05) and by 28 and 51% for P c1 (P<0.05) and P c2 (P<0.01) compared to the scrambled control, respectively (Fig. 2C-F). These results suggested that transduction of shRNA lentiviral particles results in a significant decrease of the MRP5 mRNA and surface protein expression of both MIA PaCa-2 and PANC-1 cells.

The integration of plasmid and the potential off-target effects of MRP5-short-hairpin RNA were examined in MRP5 knockdown cells (Fig. S1). No significant differences of ABCB1 and ABCC2 mRNA expression were found in MRP5 knockdown cells compared to scramble control. However, the accumulation of BCECF remained at the same level between MRP5 knockdown cells and scramble control for both MIA Paca-2 and PANC-1 cells (Fig. S2). This may be caused by either single-cell variability or the loss of substrate specificity in the models used in the present study.

MIA PaCa-2 and PANC-1 clones were treated with 0–150 and 0–1,500 nM gemcitabine, respectively, in the 2D culture models. The 72-h cytotoxicity of gemcitabine was determined using an MTT assay. Gemcitabine inhibited the proliferation of scrambled control and knockdown clones in a dose-dependent manner. Table II shows that the IC₅₀ values were decreased by 52% (M c1) and 42% (M c2) for MIA PaCa-2 cells (Fig. 3A) and by 46% (P c1) and 41% (P c2) for PANC-1 cells (Fig. 3B) compared to the scrambled control, respectively. Similar patterns were observed in pancreatic cancer cells by using colony formation assays (Fig. S3). The response of the two pancreatic cancer cell lines to gemcitabine was also examined in an in vitro 3D cell culture model. Gemcitabine was more potent against control pancreatic cancer cells (IC₅₀ values of 17.26 and 13.88 nM for Mia PaCa-2 and PANC-1 cells, respectively) grown in 2D cultures in comparison with those in 3D cultures (both IC₅₀ values >100 µM). Silencing MRP5 led to markedly increased sensitivity to gemcitabine in 3D culture models of both Mia PaCa-2 and PANC-1 compared to the control (Fig. 3C and D). In contrast, the knockdown clones and scramble control for both MIA Paca-2 and PANC-1 cells showed the same vulnerability to a poly (ADP-ribose) polymerase inhibitor olaparib, which is not an MRP5 substrate (Fig. S4 and Table SI).

Gemcitabine-induced apoptosis was measured by detecting caspase activity in the apoptotic population. The time course was determined based on the cell type. MIA PaCa-2 clones exposed to gemcitabine exhibited extensive apoptosis after a 24-h incubation. PANC-1 clones showed minor apoptosis after 24 h of treatment; thus, the incubation time was increased to 48 h. In MIA PaCa-2 clones M c1 and M c2 treated with 100 nM gemcitabine, the apoptotic population increased by 67.50±16.40% (P<0.01) and 44.99±5.21% (P<0.01) compared to the scrambled control, respectively (Fig. 4A and B). After treatment with 200 nM gemcitabine, the apoptotic population increased by 53.15±32.50% (P<0.05) and 34.34±5.00% (P<0.05) for M c1 and M c2, respectively (Fig. 4A and B).

A similar trend was seen in PANC-1 cells-after treatment with 100 nM gemcitabine, the apoptotic population increased by 55.21±18.25 and 65.35±37.47% for P c1 (P<0.05) and P c2 (P<0.001), respectively (Fig. 4C and D). In PANC-1 cells treated with 200 nM gemcitabine, the apoptotic population increased by 97.18±23.44 and 49.09±20.94% for P c1 (P<0.05) and P c2 (P<0.05), respectively (Fig. 4C and D). These results signified increased gemcitabine sensitivity in ABCC5 knockdown pancreatic cancer cells.

The effect of MRP5 on pancreatic cancer clonogenic capacity was examined using a colony-formation assay. Silencing of MRP5 decreased the clonogenic capacity of Mia PaCa-2 and PANC-1 cells (Fig. 5). The colonies formed within 10 days of incubation for the control Mia PaCa-2 and PANC-1 cells were significantly more numerous than those of the MRP5 knockdown cells.

Silencing MRP5 decreased wound-induced migration in Mia PaCa-2 and PANC-1 cells (Fig. 6). Complete wound healing occurred within 72 and 40 h for the control Mia PaCa-2 and PANC-1 cells, respectively; but not in the MRP5 knockdown cell lines (Fig. 6A and B).

Silencing of MRP5 decreased cell migration in the Transwell assays quantifying cell migration toward the complete medium for 48 h (Fig. 7). The migrated cells per field were reduced by >50% for both cell lines examined.

In the present study, gemcitabine and cGMP were analysed through a molecular docking study using AutoDock 4.2 software. Docking results against receptor molecules obtained from Protein-Ligand Interaction Profiler showed that both gemcitabine and cGMP formed four hydrogen bonds with ABCC5. The estimated free energy of binding of gemcitabine and cGMP were determined to be −2.77 and −6.44 kcal/mol, respectively (Fig. 8).