Human PDAC tissue samples
All tumor samples used in this study were obtained from patients who underwent surgery for PDAC at Kyushu University Hospital (Fukuoka, Japan); all patients provided written informed consent. Clinical data including histopathological findings were obtained from electronic medical records. The study was approved by the Ethics Committee of Kyushu University (approval number: 29–401, 30–230, 2019–462, and 858–00) and conducted according to the Ethical Guidelines for Human Genome/Gene Research enacted by the Japanese Government and Helsinki Declaration.
Human PDAC organoids
All PDOs were established from human PDAC tissue samples as previously described [22, 24]. PDAC tissues were washed vigorously with ice-cold phosphate-buffered saline (PBS) and minced into 1 mm3 fragments with scalpels. The fragments were digested into single cells using the Tumor Dissociation Kit (130–095-929; Miltenyi Biotec, San Diego, CA, USA). The dissociated cells were cultured in a 24-well plate (353,504; Corning, NY, USA) with growth factor–reduced Matrigel (356,231; Corning) and cultured in niche medium (see below) at 37 °C in a humidified atmosphere containing 10% CO2. The culture media were exchanged every 2 or 3 days; for passaging, established PDOs were collected, washed with 0.5 mM EDTA-PBS, and dissociated by digestion with trypsin and mechanical shearing. The dissociated cells of PDO were replated with fresh Matrigel and cultured in niche medium. The following four types of media were used depending on the experiment. The basic medium for PDO culture comprised Advanced Dulbecco’s modified Eagle’s medium (DMEM)/F12 (12,634,010; ThermoFisher, Waltham, MA, USA) supplemented with 10 mM HEPES (ThermoFisher), 2 mM GlutaMax (35,050–061; Life Technologies, Carlsbad, CA, USA), penicillin/streptomycin (15,140,122; ThermoFisher), 1X B27 (17,504,044; ThermoFisher), 10 mM nicotinamide (N0636; Sigma-Aldrich Co., St. Louis, MO, USA), and 1 mM N-acetyl-L-cysteine (A9165; Sigma-Aldrich Co.). The niche medium comprised the basic medium supplemented with the following niche factors: 100 ng/ml human recombinant Wnt-3a (5036-WN-010; R&D Systems, Minneapolis, MN, USA), 1 μg/ml human recombinant R-spondin1 (RSPO1, 120–38; PeproTech, Cranbury, NJ, USA), 100 ng/ml human recombinant Noggin (120-10C; PeproTech), 50 ng/ml human recombinant EGF (AF-100–15; PeproTech), 100 ng/ml human recombinant FGF-10 (100–26; PeproTech), and A83-01 (2939/10; R&D Systems). The serum medium comprised the basic medium supplemented with only 5% fetal bovine serum (FBS). The combined medium comprised the basic medium supplemented with both niche factors and 5% FBS. All media were supplemented with Y-27263 (Sigma-Aldrich, St. Louis, MO, USA) to prevent anoikis [25, 26]. PDO images were acquired by fluorescence microscopy (BZ-X700; Keyence, Osaka, Japan).
Human CAFs and PCCs
Two human CAF lines (CAF-1 and CAF-2) were established in our laboratory from fresh pancreatic cancer surgical specimens using the outgrowth method [9] as previously described [10, 11]. The isolated cells were confirmed as CAFs by their spindle-shaped morphology and expression of α-smooth muscle actin (αSMA) and glial fibrillary acidic protein (GFAP); immortalization of CAFs was conducted as previously described [12]. Six PCCs were used in this study: MIAPaCa-2 (Japanese Cancer Resource Bank, Osaka, Japan), Panc-1 (RIKEN BRC, Tsukuba, Japan), Capan-2, CFPAC-1, BxPC-3 (American Type Culture Collection, VA, USA), and KP-2 (Japan Health Sciences Foundation, Tokyo, Japan). All PCCs were regularly authenticated by matched short tandem repeat DNA profiling. CAFs and PCCs were maintained in DMEM (Sigma Chemical Co., St. Louis, MO, USA) supplemented with 10% FBS, streptomycin (100 mg/ml), and penicillin (100 mg/ml) at 37 °C in a humidified atmosphere containing 10% CO2 [27].
Gene mutation analysis
DNA was isolated using a QIAamp DNA Mini Kit (Qiagen, Hamburg, Germany) according to the manufacturer’s instructions. The quality of DNA specimens was confirmed by gel electrophoresis. DNA samples of PDOs were amplified using the Ion AmpliSeq Comprehensive Cancer Panel (ThermoFisher). Amplified fragments were used for library preparation with Ion AmpliSeq Library Kit 2.0 (ThermoFisher) and sequence analysis by Ion Torrent Personal Genome Machine (ThermoFisher). The sequence reads were checked and mapped against the human reference sequence Hg19 by Ion Reporter Software (ThermoFisher). Torrent_variant_caller (http://158.129.170.67/ion-docs/Home.html) was used for detection for the variants. SnpEff 4.1 [28] was used for single nucleotide polymorphism annotation.
Transcriptome analysis
Total RNA was extracted from PDOs that had been passaged 4 to 6 rounds post-establishment and cultured with combined medium at day 5 post-passage using the High Pure RNA Isolation kit (11,828,665,001; Roche, Basel, Switzerland) with DNase I (Roche). RNA quality was evaluated using 2200 TapeStaton (Agilent Technology, Santa Clara, CA, USA). Gene expression levels were determined by SurePrint G3 Human GE Microarray 8 × 60 K v3.0 (Agilent Technology). Relative hybridization intensities and background hybridization values were calculated using Feature Extraction software (Agilent Technology). The raw signal intensities of all samples were log2-transformed and normalized by quantile algorithm with the ‘preprocessCore’ library package [29] on Bioconductor software [30]. Principal component analysis (PCA) for clustering of PDOs were performed on all genes dataset using R program with rgl package. The gene expression heatmap was generated with gplots and ggplot2 package on R (https://CRAN.R-project.org/package=gplots) and MultiExperiment Viewer version 4.9. For the validation study of Moffitt’s classification, the expression data of “Basal-like” and “Classical” signature genes were extracted and normalized to a Z-score. Total score was calculated by subtracting the total Z-score of “Classical” genes from the total Z-score of “Basal-like” genes; PDOs with total score ≥ 0 were classified as “Basal-like” subtype, and PDOs with total score < 0 as “Classical” subtype. Gene set enrichment analysis (GSEA) was performed for two clusters identified by PCA using GSEA software version 4.0.3 (Broad Institute, UC San Diego, CA, USA). For the analysis of genes with expression levels that were positively or negatively correlated with niche dependency scores, the Pearson correlation coefficients of the basal dataset on R software were calculated and the top 1000 and the bottom 1000 correlated genes were extracted. Functional annotation clustering analysis was performed using DAVID (National Institute of Allergy and Infectious Diseases).
Proliferation and organoid formation assays
For proliferation assay, the growth rate of PDOs and 3D-cultured cell lines was evaluated using the CellTiter-Glo Luminescent Cell Viability Assay Kit (CellTiter-Glo Kit, G7571; Promega, Madison, WI, USA). Cells (5,000) were seeded into a 24-well plate with Matrigel and cultured under indicated conditions. To measure the amount of cells in each well, the 3D-cultured cells were trypsinized into single cells again for each well. The cells for each well were then replaced into the 96-well plates (655,083; Greiner Bio-One International, Kremsünster, Austria) with CellTiter-Glo reagents and the luminescence was measured by a microplate reader (Infinite200, TECAN, Männedorf, Switzerland) according to the manufacturer’s instructions. A blank well containing only medium was used to define baseline luminescence of the medium, and the luminescent signal for each sample was calculated by subtracting the baseline luminescence from the luminescence of each well. In assays to compare the proliferation of PDOs in different medium, the luminescent signal of 5,000 cells were measured as a control on day 0, and the proliferation fold change in luminescent signal on day 10 relative to the control was calculated. For organoid formation assay, phase-contrast images of PDOs were captured using BZ-X700 with the Z-stack and image stitching function on day 10. Fully focused images were generated and the number and total area of PDOs for each well was quantified using the HybridCellCount software module of BZ-X Analyzer. An area of 2000 μm2 and more was identified as an organoid. Niche dependency scores were calculating by the ratio of the proliferation fold change in niche medium to that in serum medium. In niche dependency assays, PDOs were cultured in niche medium or in niche medium lacking indicated factors and the growth rate was evaluated on day 10. The proliferation fold change in niche medium lacking the indicated factors relative to niche medium was calculated. In C59 assays, PDOs were cultured in niche medium lacking Wnt3A (-Wnt medium) supplemented with vehicle (dimethyl sulfoxide, DMSO) or in -Wnt medium supplemented with 100 nM porcupine inhibitor (Porcn-i; C59, ab142216; abcam) and the growth rate was evaluated on day 10. In RSPO1 vs. RSPO3 assays, Grade1 PDOs were cultured in niche medium with the indicated concentration of RSPO1 or RSPO3 (recombinant human RSPO3, 120–44; PeproTech) and the growth rate was evaluated on day 8. The proliferation fold change of PDOs with the indicated concentration of RSPO1 or RSPO3 relative to 1000 ng/ml RSPO1 was calculated. Each experiment was performed in triplicate and repeated more than three times.
Three-dimensional co-culture assay
For direct 3D co-culture, GFP-labeled PDOs were trypsinized into single cells and seeded with or without CAFs at a 1:20 ratio (5 × 103 PDO cells and 1 × 105 CAFs/well) into a 24-well plate with Matrigel with serum medium. For indirect 3D co-cultures, CAFs were seeded into the trans-well membrane (3 μm pore size, 353,096; Corning) with Matrigel. Fluorescence and phase-contrast images of PDOs with or without CAFs were captured using BZ-X700 with the Z-stack and image stitching function at day 10. Fully focused and overlaid images were generated using BZ-X Analyzer. The total area of PDOs for each well was measured using the HybridCellCount software module of BZ-X Analyzer.
Drug treatment assays
PDOs were dissociated into single cells and 5,000 cells were seeded into a 24-well plate with Matrigel. For gemcitabine treatment, PDOs cultured in combined medium or serum medium for 7 days were treated with the indicated concentration of gemcitabine (Gemzar; Eli Lilly Japan K.K., Kobe, Hyougo, Japan) for 72 h. For simvastatin treatment, immediately after seeding with Matrigel, cells were treated for 10 days in combined medium or serum medium with the indicated dose of simvastatin (S6196; Sigma Aldrich). PDOs were then trypsinized into single cells and cell viability for each well was quantified using the CellTiter-Glo Kit as described above.
Immunohistochemistry and immunofluorescent staining
PDOs isolated from Matrigel using Cell Recovery Solution (354,253; Corning) were embedded in iPGell (PG20-1; GenoStaff, Tokyo, Japan) without damaging the 3D structures according to the manufacturer’s protocol. PDOs were then fixed with 4% paraformaldehyde and used to create paraffin-embedded blocks. Paraffin-embedded blocks of PDOs and PDAC tissues were sectioned (4 μm) and subjected to standard hematoxylin and eosin (H&E) staining and immunostaining. The following primary antibodies were used: E-cadherin (ab15148; abcam, Cambridge, UK, 1:30), Actin (ab130935; abcam, 1:100), CK19 (sc25724; Santa Cruz Biotechnology, Inc., Dallas, TX, USA, 1:50), αSMA (M0851; Dako, Santa Clara, CA, USA, 1:200), and RSPO3 (17,193–1-AP; proteintech, Rosemont, IL, USA, 1:100). The secondary antibodies were EnVision + System-HRP Labelled Polymer (K4003; Dako) and Alexa Fluor 488 anti-rabbit (A11034; ThermoFisher) and 546 anti-mouse (A11030; ThermoFisher). Nuclei were counterstained with hematoxylin or 4′,6-diamidino-2-phenylindole (DAPI; Dojindo, Kumamoto, Japan). For immunohistochemistry, staining was developed with 3,3′-diaminobenzidine substrate chromogen (11,209-1A; Kanto Kagaku, Tokyo, Japan) Bright field images were acquired using BZ-X700. The stain-positive area was quantified using HybridCellCount software module of BZ-X Analyzer (Keyence).
Transfection of fluorophores and small hairpin RNA
To obtain RFP-labeled CAFs and GFP-labeled PDOs, RFP and GFP lentiviral particles (RFP, LVP023-PBS; GFP, LVP001-PBS; GenTarget Inc., San Diego, CA, USA) were transfected into immortalized CAFs (CAF-1) and PDOs, respectively. RFP transfection into CAF-1 was performed according to the manufacturer’s instructions. GFP transfection into PDOs was performed as previously described [31]. Briefly, single-cell suspensions of PDOs with combined medium and viral particles were plated into Matrigel-coated 24-well plates and incubated at 37 °C. The next day, the medium with viral particles was carefully removed and fresh Matrigel and medium were overlaid on the cells attached to Matrigel. When PDOs became sub-confluent, blasticidin S hydrochloride (15,205; Sigma-Aldrich) was used to select GFP and RFP clones. RFP-positive CAFs and GFP-positive PDOs were sorted by the Cell Sorter SH800S (Sony Corporation, Tokyo, Japan). To obtain RSPO3-knockdown CAFs, two high-titer lentiviral particles packing small-hairpin RNA (shRNA) against RSPO3 (MISSION Lentiviral Transduction Particles; shRSPO3-1, TRCN0000056663; shRSPO3-2, TRCN0000373388; Sigma-Aldrich) or non-targeting shRNA (SHC016V; Sigma-Aldrich) as control were transfected into immortalized CAFs (CAF-1) according to the manufacturer’s instructions. Puromycin (631,305; Takara) was used for more than 3 weeks to select RSPO3 shRNA–expressing cells. Knockdown efficacy of RSPO3 shRNA was confirmed by real-time quantitative reverse transcription-PCR (qRT-PCR).
Collection of conditioned media from PDOs (PDO-CM)
PDOs cultured in combined medium for 7 days were washed twice with PBS and the fresh basic medium was replaced. After 48 h incubation at 37 °C, the medium was collected and filtered with a 0.22-μm syringe filter (Z359904; Merck, Darmstadt, Germany). After centrifugation at 1,500 rpm for 5 min, the supernatants were collected and FBS was added to a concentration of 5%. We prepared serum medium as a control medium.
RNA extraction and quantitative RT-PCR
RNA was extracted from CAFs cultured under different conditions (2D-culture with serum medium, 3D-culture with serum medium, and 3D-culture with PDO-CM) for 72 h and RFP-positive CAFs directly or indirectly co-cultured with PDO585 in serum medium for 10 days using a High Pure RNA Isolation kit (11,828,665,001; Roche) and DNase I (Roche) treatment according to the manufacturer’s instructions. RFP positive CAFs co-cultured with PDO585 were sorted by the Cell Sorter SH800S. Real-time qRT-PCR was performed using the iTaq Universal SYBR Green One-Step Kit (172–5150; BioRad, Hercules, CA, USA) and CFX96 Touch Real-Time PCR Detection systems (Bio-Rad). Transcript quantities were determined using the ΔΔCt method and values were normalized to GAPDH mRNA. The following primers purchased from Takara Bio (Kusatsu, Japan) were used in this study: RSPO1, 5′-TCCAGAGCTCCCAGTGGACA-3′ (forward) and 5′-CAGGTCACCAGCAGTCCTCAAG-3′ (reverse); RSPO2, 5′-AGAAGCCCAAACTGCCTTTGA-3′ (forward) and 5′-TCTGTAGCTGGCCTGTGAAACTG-3′ (reverse); RSPO3, 5′-CATGACAATGGTGGCAAATGAC-3′ (forward) and 5′-TTTAGCATCAAGGATTCAGACCT-3′ (reverse); RSPO4, 5′-AGCAAGTCTGTCCTCACTGCCTATC-3′ (forward) and 5′-CGGCAAATACAAATCCCGTTTC-3′ (reverse); GAPDH, 5′-GCACCGTCAAGGCTGAGAAC-3′ (forward) and 5′-TGGTGAAGACGCCAGTGGA-3′ (reverse).
Statistics
Statistical analyses were performed using Prism7 (GraphPad, San Diego, CA, USA). Data are represented as the mean ± standard error of the mean (SEM) unless otherwise indicated. For the comparisons of two groups, the unpaired two-tailed Student’s t-test was performed, and a P value < 0.05 was considered to be statistically significant. The Kaplan–Meier analysis was used to analyze survival, with curves compared using the Log-rank (Mantel–Cox) test. Area under curve (AUC) values were calculated from a log (drug) vs. response curve with robust fit.