Isolation of mesothelioma-associated fibroblasts

Tissue samples were collected from surgery specimens of PM patients who underwent surgical resection at the Department of Thoracic Surgery of the Medical University of Vienna. The study was approved by the ethics committee of the Medical University of Vienna (EK Nr. 904/2009) and written informed consent to use their tissue for research purposes has been obtained from all patients. To establish primary Meso-CAF cultures, the tumor tissue was minced into small pieces and incubated in cell culture flasks (T25) in RPMI-1640 growth medium (#R6504; Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Biowest, France) and antibiotics (100 U/ml penicillin and 100 μg/ml of streptomycin (1% P/S)) (Sigma-Aldrich, St. Louis, MO, USA). The tissue pieces were incubated in a humidified atmosphere (37 °C, 5% CO2) without agitation (for the first 7 days) to allow attachment of tumor pieces to the culture device and cell outgrowth. The phenotype of out-growing cells was validated by light microscopy. Flasks with fibroblastoid cells were selected for further cultivation and expansion. The cells were regularly washed with sterile phosphate-buffered saline (PBS, 1x) to remove cell debris.

Cell culture

The fibroblastoid cells, which could be verified as Meso-CAFs (Meso109F, Meso125F, VMC59F) were all maintained in RPMI-1640 medium supplemented with 10% FBS within uncoated flasks. All primary Meso-CAFs used for experiments were at fewer than 15 passages. The human mesothelioma cell line MSTO-211H was purchased from the American Type Culture Collection (ATCC, Rockville, MD, USA), SPC212 was provided by Prof. R. Stahel (University of Zurich, Switzerland), and Meso84 as well as VMC23 were established at the Medical University of Vienna as recently described by Pirker et al. [11]. All PM cells were kept in RPMI-1640 medium supplemented with 10% FBS. Establishment of the primary cancer-associated fibroblasts CAF-3 from colorectal adenocarcinoma cultured in endothelial cell growth medium (EGM 2 MV; #CC-3202; Lonza, Basel, Switzerland) was previously described [12]. The human primary normal lung fibroblasts (NLFs) MRC-5 and Wi-38 were obtained from the ATCC, and cultivated in the respective growth medium according to the supplier’s protocol (MRC-5 in Dulbecco’s modified Eagle’s medium (DMEM) and Wi-38 in Eagle’s minimal essential medium (MEM) both supplemented with 10% FBS). All cells were incubated under normoxic cell culture conditions (37 °C, 5% CO2) in a humidified incubator, regularly passaged by trypsinization and routinely checked for Mycoplasma contamination. Human cell authentication was carried out for all cells using short tandem repeat DNA profiling analysis. A summary of all cell types used in the study is provided in Supplementary Table S1.

Determination of doubling times

To evaluate the doubling times of Meso-CAFs, 7.5 × 104 cells per well were seeded in a 12-well plate. Duplicates were trypsinized every 48 h and cell numbers were counted at different time points using a Neubauer cell counting chamber. Final doubling times of the cells were calculated from duplicated cell counts of at least four time points using the formula: doubling time [h] per time point = (time [h]) × log(2)/(log(cell count at time point)-log(seeded cell number)).

Array-based comparative genomic hybridization

Microarray-based comparative genomic hybridization (array CGH) analysis of Meso-CAFs and CAF-3 was performed as described by Pirker et al. [11] and Mathieu et al. [13]. Genomic DNA was isolated from primary fibroblast cultures of about 80% cell confluence using QIAamp DNA Blood Mini Kit (Qiagen, Valencia, CA, USA). The genome analysis was carried out on 4x44K whole genome oligonucleotide-based microarrays (Agilent, Santa Clara, CA, USA) according to the manufacturer’s protocol, scanned on a G2505B Micro Array Scanner (Agilent, Santa Clara, CA, USA) and analyzed using the software Genomic Workbench (version 7.0) (Agilent, Santa Clara, CA, USA). Array CGH profiles of fibroblasts were checked for genomic alterations and compared to genome profiles of PM cell lines from a previous publication [11]. Array CGH data from Meso-CAFs and CAF-3 are available at ArrayExpress (https://www.ebi.ac.uk/biostudies/arrayexpress) under the accession number E-MTAB-12179.

Quantitative real-time reverse transcription PCR

Commonly used CAF and PM markers were selected from the literature and their expression levels were quantified in Meso-CAFs, CAF-3 and NLFs (MRC-5, Wi-38), as well as in PM cells of different histological subtypes (VMC23, SPC212, MSTO-211H, Meso84) using quantitative real-time reverse transcription PCR (qRT-PCR). Total cellular mRNA was isolated from cell cultures of about 80% confluence with innuPREP RNA Mini Kit (Analytik Jena, Jena, Germany) and cDNA was synthesized with 2 μg of RNA using RevertAid RT Kit (Thermo Scientific, Thermo Fisher Scientific, Carlsbad, CA, USA), both according to the respective manufacturer’s instructions. The qRT-PCR was carried out with 1 μl of cDNA and appropriate primer pairs on a C1000 Touch Thermal Cycler (Bio-Rad Laboratories, Hercules, CA, USA) using SYBR Green PCR Master Mix (Applied Biosystems, Thermo Fisher Scientific, Carlsbad CA, USA). All primer sequences are listed in Supplementary Table S2. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and β-Actin were used as reference genes for normalization and gene expression levels were calculated with the ΔCt method.

Immunohistochemistry

Generation of paraffin-embedded cell blocks was done as previously described [14]. Sections were incubated with primary antibodies (BAP1, sc-28383 (Santa Cruz Biotechnology, Dallas, TX, USA), 1:200; calretinin, E7R60 (Cell Signaling Technology, Danvers, MA, USA), 1:50; CK8/18, C51 (Cell Signaling Technology, Danvers, MA, USA), 1:50; WT1, D817F (Cell Signaling Technology, Danvers, MA, USA), 1:100) at 4 °C overnight and antibody binding was detected with the UltraVison LP detection system (Lab Vison Corporation, Freemont, CA, USA).

Next generation sequencing

DNA and RNA were isolated from the Meso-CAFs as described above for array CGH and qRT-PCR, respectively. Sequencing was performed with a 523 gene panel (TruSight Oncology 500 panel, Illumina Inc., San Diego, CA, USA) on an NextSeq 550 instrument (Illumina Inc). The list of investigated genes is shown as Supplementary Table S3. The sequences were analyzed with the Clinical Genomics Workspace bioinformatics pipeline from Pierian (Creve Coeur, MO, USA).

Sanger sequencing

For Sanger sequencing, PCR amplicons spanning all loci containing variants in Meso109F were generated with Q5 proofreading polymerase (New England Biolabs, Ipswitch, MA, USA) using DNA from Meso109F cells and from whole blood of the same patient (isolated as described above for array CGH) as template. PCR products were purified and Sanger sequencing was done by Microsynth Austria GmbH (Vienna, Austria). PCR primers are listed in Supplementary Table S2 and were also used as sequencing primers.

Whole genome gene expression microarrays

Total RNA of Meso-CAFs and CAF-3 was extracted from primary cell cultures of about 80% confluence with RNeasy Mini Kit (Qiagen Sciences, Germantown, MD, USA) according to the respective protocol of the manufacturer. Whole-genome gene expression analysis was carried out as described previously by Pirker et al. [11] and Mathieu et al. [13] using 4x44K whole genome oligonucleotide-based gene expression microarrays (Agilent, Santa Clara, CA, USA) and a G2505B Micro Array Scanner (Agilent, Santa Clara, CA, USA). Gene expression data of Meso-CAFs was compared to CAF-3 and to data of 31 PM cell lines available from a previous publication [11]. Analysis and comparison of expression data was conducted in R [15], and differentially expressed genes between cell types (Meso-CAF, CAF-3, PM) were determined using “limma” package [16] as previously described by Mohr et al. [17]. Genes with multiple oligonucleotide probes on the array were summarized to the probe with maximal interquartile range using the package “genefilter” [18]. Additional plots were created using GraphPad Prism 8.0 (GraphPad Software, San Diego, CA, USA). Whole genome gene expression array data from Meso-CAFs and CAF-3 are available at ArrayExpress (https://www.ebi.ac.uk/biostudies/arrayexpress) under the accession number E-MTAB-12177.

Mass spectrometry-based proteomics

Three different protein fractions (supernatant, cytoplasmic and nuclear proteins) of Meso-CAFs (Meso109F, Meso125F, VMC59F) and NLFs (MRC-5, Wi-38) were analyzed in three biological replicates. The cells were cultivated in T25/T75 flasks until exhibiting 80–90% of confluence and the growth medium was exchanged to the respective FBS-free medium 6 hours prior to protein extraction. Protein fractioning was conducted as described by Slany et al. [19]. The supernatant containing the secreted proteins was initially collected, and the cells were lysed using isotonic lysis buffer supplemented with protease inhibitors and mechanical shear stress application with a 23 g syringe. The nuclei were separated from the cytoplasmic protein fraction by centrifugation and all protein fractions were individually precipitated in ethanol at − 20 °C overnight. After centrifugation, protein pellets were dissolved in sample buffer (7.5 M urea, 1.5 M thiourea, 4% CHAPS, 0.05% SDS, 100 mM dithiothreitol) and concentrations of distinct protein fractions were assessed using Bradford assay (Bio-Rad Laboratories, Hercules, CA, USA). Enzymatic in-solution digestion of proteins into peptides with Trypsin/Lys-C (Promega Corporation, Fitchburg, WI, USA) and sample clean-up on SDB-RPS StageTips was performed based on the protocol from Humphrey et al. [20] with minor changes. Peptide samples were subjected to LC-MS/MS analyses on a Dionex Ultimate 3000 nano LC-system (Thermo Scientific, Thermo Fisher Scientific, Carlsbad, CA, USA) coupled to a timsTOF pro mass spectrometer (Bruker Daltonics, Bruker Corporation, Billerica, MA, USA). All samples were run as technical replicates and data analysis for protein identification was performed using MaxQuant 1.6.17.0 [21] employing the Andromeda software searching against the UniProt Database for human proteins (version 12/2019 with 20,380 entries). The mass spectrometry proteomics data were submitted to the ProteomeXchange Consortium via the PRIDE partner repository and can be accessed with the dataset identifier PXD035987, PXD036017 and PXD036127 [22]. Statistical data analysis and the generation of figures was conducted in R with the package “DEP” [23] and by the use of the software Perseus (version 1.6.14.0) [24, 25]. Additional plots were created using GraphPad Prism 8.0.

The proteins of the supernatant, cytoplasmic and nuclear fraction were additionally filtered for actively secreted, membrane-associated and DNA-binding proteins, respectively, by comparing the protein data sets with published human data from protein databases (UniProt, Human Protein Atlas). Moreover, the secretome data of lung-CAFs [26], colon-CAFs [27] and breast-CAFs [28] was collected from the literature, filtered for actively secreted proteins as described above and the annotated genes of the identified proteins in the secretomes of different CAFs and Meso-CAFs were compared using Venny 2.1 [29]. To measure the similarity between the secretomes, Jaccard similarity coefficients between different gene sets were calculated. Gene ontology (GO) enrichment analysis was performed using DAVID functional annotation tool [30, 31] and biological processes associated with differentially expressed proteins of Meso-CAFs, NLFs and CAFs from other tumor entities were determined.

A more detailed methodologic description of the proteome analysis is available as online supplement (Supplementary Methods).

Transduction of tumor cells with green fluorescent protein

The cDNA of green fluorescent protein (GFP) was integrated into the genome of PM cells using retroviral transduction to introduce stable expression of the fluorescence protein. Generation of retrovirus particles in HEK293 cells and transduction of target cells was performed as previously described [32]. Briefly, HEK293 cells, which were maintained in DMEM supplemented with 10% FBS were transfected by calcium phosphate co-precipitation to take up the vector pQCXIP (Clontech, Mountain View, CA, USA) containing the GFP sequence as well as a puromycin resistance gene and the two helper plasmids pVSV-G (Clontech, Mountain View, CA, USA) and p-gag-pol-gpt [33]. Supernatants of the transfected HEK293 cells containing the retroviral particles were used to transduce SPC212 and MSTO-211H cells in the presence of polybrene (8 μg/ml), followed by a selection of GFP-positive cells with puromycin treatment (0.8 μg/ml).

Generation of 2D co-cultures

GFP-tagged tumor cells were either seeded alone or together with Meso-CAFs in a ratio of 1:100 (96-well plate: 150 tumor cells, 1.5 × 104 Meso-CAFs; 48-well plate: 300 tumor cells, 3 × 104 Meso-CAFs). Bright field and fluorescence images of the wells were taken every 24 h for 3 days, either manually using a Nikon Eclipse Ti2 microscope with a DS-Fi3 camera (Nikon, Tokyo, Japan) or automatically using the IncuCyte S3 Live-cell Analysis System (Sartorius, Göttingen, Germany). Definiens Developer XD Software (Definiens, Carlsbad, CA, USA) was used to quantify the number of tumor cells on the images via automated cell counting based on cell size and GFP signal. Tumor cell growth over time was evaluated by determining the cell number percentage at different time points compared to 0 h.

Generation of 3D co-cultures

3D cell cultures were generated as described by Dolznig et al. [12]. GFP-tagged tumor cells were either embedded alone or together with Meso-CAFs in a matrix of collagen in a ratio of 1:100 (3 × 103 tumor cells, 3 × 105 Meso-CAFs). The collagen solution was prepared on ice by mixing 2 mg/ml collagen type I (rat tail; Corning, Bedford, MA, USA), 10x PBS and 0.05% methylcellulose, as well as NaOH (1 M) to neutralize the pH (7.2–7.4). The cells were gently resuspended in the solution and the cell suspension was transferred into a silicone gel casting mold inside a 6 cm plastic dish. A nylon mesh insert was centrally placed in the middle of each culture to prevent collagen gel contraction by the fibroblasts. After incubating the cultures for 1 h at 37 °C (5% CO2), the collagen gels were polymerized and the casting molds were removed. The remaining collagen gel cylinders with the cells were maintained in RPMI-1640 medium with 10% FBS under regular culture conditions and the growth medium was exchanged every second day. Bright field and fluorescence images of the cultures were taken every 24 h for 6 days using a Nikon Eclipse Ti2 microscope with a DS-Fi3 camera. The pictures were taken at five different non-overlapping positions of the cultures and several layers per location to reflect the whole 3D culture. Quantification of tumor cells and determination of cell growth over time was performed as described above.

Analysis of cell migration

2D cultures were generated as described above and GFP+ tumor cells were monitored by videomicroscopy using the IncuCyte S3. Images of the wells were taken every 30 minutes for a total duration of 60 h. Movements of single GFP+ tumor cells were manually tracked using ImageJ to obtain coordinates for each individual cell and time point. The DiPer migration tool for Microsoft Excel was used to analyze the migratory behavior of tumor cells in presence or absence of Meso-CAFs including migrated distance, mean squared displacement (MSD), directionality ratio (DR), and plots of origin [34].

Analysis of effects mediated by conditioned medium

Conditioned medium (CM) was prepared by cultivating Meso-CAFs or NLFs under conventional culture conditions and collecting the supernatant. When the cells exhibited a confluence of about 80%, they were incubated in fresh RPMI-1640 medium supplemented with 10% FBS for 72 h. The collected supernatant was centrifuged for 10 minutes at 1500×g to remove cell debris and either immediately used for experiments or stored at − 80 °C for later use. GFP+ tumor cells were seeded in 48-well plates (300 tumor cells per well) and treated for 30 h with a mix of CM and fresh RPMI-1640 medium supplemented with 10% FBS in a ratio of 1:1 (500 μl of each per well). Fresh growth medium was added to exclude effects from depleted growth medium constituents in the CM alone and ensure full nutrition of the tumor cells over the experiment. Fresh growth medium alone was used as control and all experiments were performed in triplicates. The generation of 2D cultures as well as the evaluation of cell growth and migration was carried out as described above using IncuCyte S3 and Definiens or ImageJ. Images were taken every 5 h or every 30 minutes for the analysis of growth or migration, respectively.

Analysis of inhibitor treatment effects

2D cultures in 48-well plates were treated with various small molecule inhibitors of signaling pathways and were imaged every 24 h for 72 h using IncuCyte S3. Tumor cell growth over time was determined as described above and effects of distinct signaling inhibition in presence or absence of Meso-CAFs were evaluated. Final working concentration of each inhibitor was determined by a stepwise increase of concentrations until a reduction of tumor cell growth of at least 20% could be observed either in the cultures with or in those without Meso-CAFs (assessed by automated image analysis) without leading to a visually assessed decrease in the cell confluence of the Meso-CAFs. The following inhibitors were used at indicated final concentrations: TGFβ receptor type I inhibitor galunisertib (Selleckchem, Houston, TX, USA, 10 μM), the EGFR inhibitor erlotinib (Selleckchem, 1 μM), the PI3K inhibitor LY-294002 (Selleckchem, 5 μM), the c-Met inhibitor crizotinib (MedChem Express, Monmouth Junction, NJ, USA, 1 μM (SPC212) / 2 μM (MSTO-211H)), the MEK inhibitor U0126 (Selleckchem, 0.1 μM (SPC212) / 0.25 μM (MSTO-211H)), the FAK inhibitor BI 853520 (Boehringer Ingelheim, Ingelheim, Germany, 5 μM (SPC212) / 2 μM (MSTO-211H)), the NF-κB inhibitor BAY 11–7082 (Biomol, Hamburg, Germany, 5 μM (SPC212) / 2 μM (MSTO-211H)), the FGF receptor inhibitor erdafitinib (MedChem Express, 0.5 μM), the triple angiokinase inhibitor nintedanib (MedChem Express, 0.5 μM (SPC212) / 2 μM (MSTO-211H)), the MMP inhibitor ilomastat/GM6001 (MedChem Express, 1 μM (SPC212) / 10 μM (MSTO-211H)), the TEAD inhibitor K-975 (MedChem Express, 1 μM (SPC212) / 3 μM (MSTO-211H)), and the PORCN inhibitor WNT-C59 (MedChem Express, 5 μM). DMSO was used as vehicle control in the experiments.

Analysis of cisplatin and pemetrexed treatment effects

2D and 3D cultures were treated with the chemotherapeutic alkylating agent cisplatin (Sigma-Aldrich, St. Louis, MO, USA) and cultures were imaged at 0 h, 48 h and 72 h. Three different concentrations were used in the 2D approach (1 μM, 3 μM, 10 μM), whereas the lowest concentration was omitted in the 3D models due to limited effects. Additionally, pemetrexed (MedChem Express) was used in 2D cultures at four different concentrations (0.3 μM, 1 μM, 3 μM, 10 μM). Effects of cisplatin and pemetrexed on tumor cell growth were evaluated as described above.

Statistical analysis

All cell behavior experiments were performed in at least three independent replicates and are shown as means and SEM, unless stated otherwise. Statistical analyses were conducted using GraphPad Prism 8.0. Differences were evaluated by ANOVA and were considered statistically significant at a p-value < 0.05.