Databases, data processing, and expression analysis

The RNA-seq dataset from the European Genome-phenome Archive EGAD00001001915 (https://ega-archive.org/datasets/EGAD00001001915; 17/12/2021), including 141 epithelioid (EMM), 62 biphasic (BMM), 7 sarcomatoid (SMM), and 1 desmoplastic (DMM) mesotheliomas, was used for validation. The DepMap database was accessed via https://depmap.org/portal/ in 09/2022. RNA-seq samples of EGA and Gene Expression Omnibus (GEO) were processed/reprocessed using the same pipeline described in the Genomic Data Commons Data Portal–The Cancer Genome Atlas (GDC–TCGA) (https://portal.gdc.cancer.gov/; 25/03/2020). In brief, the downloaded data were analyzed with FASTQC [9], then the processed reads were mapped to the human genome (GRCh38.d1.vd1) using STAR [10]. GEO raw counts were calculated using HTSeq [11], while EGA raw counts were obtained using STAR. The gene IDs were annotated using the biomaRt package [12]. The raw counts for the 211 MPM and 3 normal lung specimens were used as inputs for DESeq2 [13]. Within the outputs, we verified the 15 DEGs previously identified [8]. All data were handled using R (https://www.R-project.org ; 15/03/2020).

Survival analysis

Patients were stratified as high/low groups based on the median of each gene expression. Kaplan–Meier survival estimators were used to analyze each patient group’s prognosis, and log-rank tests were used to compare the two groups’ survival results. The R packages survival and ggplot2 [14] were used to calculate the log-rank and plot the curves. Average and SD are displayed in the box plots. After confirming that the assumptions were met (p > 0.05), two-way analysis of variance (ANOVA) and the Tukey multiple comparison test were employed to compare means across the groups (***p value < 0.001; **p value < 0.01; *p value < 0.05, ns p value > 0.05).

Cell culture

MeT-5A (RRID:CVCL_3749) and MSTO-211H (RRID:CVCL_1430) cells were purchased from American Type Culture Collection (ATCC, Manassas, VA, USA), authenticated within the last 3 years with short tandem repeat profiling, and cultured in mycoplasma-free conditions with 100 U/mL penicillin and 100 U/mL streptomycin (Euroclone S.p.A., Milan, Italy) at 37 °C, 5% CO2 atmosphere. For Met-5A the Medium 199 was supplemented with 10% FBS, 3.3 nM EGF, 400 nM hydrocortisone, and 870 nM zinc-free bovine insulin (Gibco, Carlsbad, CA, USA). For MSTO the 1:1 mixture DMEM/Ham’s F-12 was supplemented with 15% FBS and 2 mM l-glutamine (Euroclone S.p.A., Milan, Italy).

Western blot (WB) analysis

Cells were collected at confluence, washed twice in PBS, and homogenized in Mammalian Protein Extraction Reagent (M-PER; ThermoFisher Scientific, Waltham, MA, USA) containing inhibitors of proteases and phosphatases (Roche Diagnostics GmbH, Rotkreuz, Switzerland). The extracted proteins were quantified by BCA test (Carlsbad, CA, USA) and 7 μg were denatured and separated by electrophoresis using precast Novex 8–16% or 4–12% Wedge Wells Tris-Glycine Gels (Invitrogen-Life Technologies, Carlsbad, CA, USA) and electroblotted onto PVDF membranes (Bio-Rad Laboratories Inc., Hercules, CA, USA). The membranes were blocked with 5% milk in TBST and probed overnight at 4 °C with the specific rabbit polyclonal antibodies (Abs) against BAG2 (cod A304-751A-T, 1:1000; Bethyl Laboratories Inc., Montgomery, TX, USA), MAD2L1 (cod 10337-1-AP, 1:800), Midkine (cod 28546-1-AP, 1:500), β-tubulin (cod 10068-1-AP, 1:8000), as well as with the mouse monoclonal Abs against epidermal growth factor receptor (EGFR) (cod 18986-1-A, P, 1:7000) and GAPDH (cod 60004-1-Ig, 1:10,000) (all from Proteintech, Rosemont, IL, USA). β-tubulin was used as the index of protein loading in the lanes. The secondary Abs, anti-rabbit IgG-HRP (cod 111-035-144, 1:10,000; Jackson ImmunoResearch Laboratories, West Grove, PA, USA) and anti-mouse IgG-HRP (cod SA00001-1, 1:20,000; Proteintech, Rosemont, IL, USA) were added for 1 h at room temperature and used for signal detection. Reactive bands were detected using Clarity MaxTM Western ECL Substrate (Bio-Rad Laboratories Inc., Hercules, CA, USA). Visualization was performed using a ChemiDoc Imaging System (Bio-Rad Laboratories Inc., Hercules, CA, USA). Densitometry of bands was carried out with ImageLab 6.0 software (Bio-Rad Laboratories Inc., Hercules, CA, USA).

Gene silencing or drug treatments

MSTO and Met-5A cells were seeded at optimized densities in 6-well or 96-well plates. At 24 h cells reached 60–80% confluence, the medium was substituted and enriched with the transfection mix (Lipofectamine RNAiMax, Invitrogen-Life Technologies, Carlsbad, CA, USA; reduced serum medium Opti-MEM, Gibco, Carlsbad, CA, USA; the chosen siRNA, 10 µM). All siRNAs were SMARTpool ON-TARGETplus siRNA (Dharmacon Inc., Lafayette, CO, USA). A non-targeting siRNA-negative-control for non-specific siRNA delivery effects (a baseline for target gene silencing), and the positive-control-siRNA targeting GAPDH were employed. Cells in 96-well plates were placed in IncuCyte system (Sartorius AG, Goettingen, Germany). This is the t0 time for the live imaging analysis, which will last for a further 72 h. Cells in 6-well plates were placed in the incubator and collected after 72 h for protein extraction, WB analysis, and verification of gene silencing that was considered successful when the target protein showed >60% of reduced expression. For drug treatments, MSTO and MeT-5A cells were seeded in 96-well plates at appropriate densities and after 24 h (30–50% confluence) treated with iMDK (7.5, 15, and 50 µM) or Neratinib (10, 20, and 40 µM) (Tocris Bioscience, Bristol, UK), dissolved in DMSO, or with the vehicle alone. The treatments lasted until the end of the assays.

Cell proliferation assay and cell viability

Silenced/treated cells were monitored with IncuCyte® live-cell imaging for 72 h. Cell growth was measured by analyzing the area of confluence, normalized to t0. At 72 h after transfection cells were rinsed twice with serum-free media HBSS (Gibco, Carlsbad, CA, USA). Calcein AM (Invitrogen-Life Technologies, Carlsbad, CA, USA), 2 µM in HBSS, was added to the cells and incubated for 2 h. Plates were placed in the IncuCyte for 2 h and the images were analyzed using IncuCyte Base Analysis Software (Sartorius AG, Goettingen, Germany) where fluorescence intensity (FI) was proportional to the number of living cells. At the end, the overall FI was also read with FLUOstar Optima (BMG Labtech, Ortenberg, Germany) plate reader (at 494/517 nm).

The Shapiro–Wilk test was used to verify the normal distribution of the data, from at least three repeats. If there was a statistically significant difference between groups, contrasts were evaluated with ANOVA or multiple Mann–Whitney tests. The statistical analysis was carried out by using GraphPad Prism 9 software (GraphPad Software, Inc., San Diego, CA, USA).

Caspase 3/7 and cytotoxicity assay

Cultures were enriched with IncuCyte® Caspase 3/7 Green Dye at 5 µM or the IncuCyte® Cytotox Red Dye at 25 nM (all from Sartorius AG, Goettingen, Germany). Then, 96-well plates were placed in IncuCyte instrument for 72 h and the collected images were analyzed using IncuCyte Base Analysis Software (Sartorius AG, Goettingen, Germany). To compare the median values of the different treatment groups, we performed multiple Mann–Whitney tests.

Wound healing assay

MSTO and MeT-5A cells were seeded in 96-well IncuCyte ImageLock Plates and grown to confluence. After 24 h a cell-free gap (scratch) in each well was made with IncuCyte Wound Maker 96-Tool (Sartorius AG, Goettingen, Germany). Then, cells were washed with PBS and treated with different combinations of siRNAs or drugs, using low serum concentrations in cell medium (2% FBS) to suppress cell proliferation. The scratch areas were monitored for 24 h, using a phase-contrast setting in the IncuCyte instrument. Images were captured every 3 h. The percentage of wound closure (WC) was calculated using the following formula:

$$}=\frac}_-}_}}}_}\times 100$$

We employed a simple linear regression to examine the relationship between WC over time. Subsequently, statistical analysis, including ANOVA followed by Sidak’s multiple comparison tests, was conducted. Prior to analysis, we ensured the fulfillment of assumptions.

Formalin-fixed paraffin-embedded (FFPE) tissue specimens

Forty FFPE archival MPM tissue specimens were collected (in 2017–2020) from 28 patients at the Department of Thoracic Surgery and analyzed at the Department of Pathology of Rigshospitalet, Copenhagen University Hospital, Denmark [15]. Signed informed consent was obtained from all patients. Twelve patients presented matched biopsies + resections, eight (inoperable) had only biopsies, and eight were only resections. The histologic subtypes were 12 EMM, 17 BMM, 6 SMM, and 5 DMM. All patients who underwent pleurectomy/decortication had EMM or BMM and had received three cycles of standard neoadjuvant chemotherapy with cisplatin-pemetrexed doublet. Five patients diagnosed with BMM showed a transformation to DMM in the corresponding P/D. For each specimen, only tissue sections with a tumor cell content >50% were used. As non-neoplastic controls, we used 13 FFPE tissue samples from 12 patients with reactive mesothelial proliferation (RMP), who had been operated for benign intrathoracic pathologies not related to MPM (6 spontaneous pneumothorax, 4 lung fibrosis, 1 post-empyema pleural fibrosis, and 1 aorta aneurism).

Immunohistochemistry (IHC)

IHC was carried out on 3.5 μm paraffin sections by using recombinant Anti-BAG2 Rabbit mAb [EPR3567] (1:400), Anti-Midkine Rabbit mAb [EP1143Y] (1:100; cod ab79406, ab52637; Abcam, Cambridge, MA, USA), and Anti-MAD2 (MAD2L1) Rabbit pAb (1:250; cod TA308923; OriGene, Rockville, MD, USA). The sections were heated for 60 min and deparaffinized using Tissue Clear (Sakura Finetek, Alphen aan den Rijn, Netherlands) and an ethanol gradient. Antigen retrieval treatment was carried out for 15 min at 98 °C with citrate buffer (10 mM Tris, 0.5 mM EGTA, pH 6.0) for the anti-MDK Ab, and TEG buffer (10 mM Tris, 0.5 mM EGTA, pH 9.0) for the anti-MAD2L1 and anti-BAG2 Abs. Then, sections were washed and incubated for 15 min in a solution of Milli-Q water and H2O2 at 1% (Merck KGaA, Darmstadt, Hesse, Germany) followed by incubation with the specific primary Ab diluted in Antibody Diluent, Background Reducing (Dako, Agilent Technologies, Inc, Santa Clara, California, USA) overnight at 4 °C. After 24 h, sections were washed in TBST (50 mM Tris, 150 mM NaCl, 0.5% Triton X-100, pH 7.6), incubated for 45 min with the secondary Ab EnVision Rabbit, (Dako, Agilent Technologies, Inc, Santa Clara, CA, USA), washed with TBST, and incubated for 5 min with Liquid DAB+ Substrate Kit (2 components for IHC; OriGene, Rockville, MD, USA). Sections were then counterstained for 30 s using Mayer’s hematoxylin (Sakura Finetek, Alphen aan den Rijn, Netherlands). Sections not incubated with primary Abs served as negative controls. Sections were scanned using NDP.view2 Plus software and a NanoZoomer-XR Digital slide scanner C12000-01 (Hamamatsu Photonics, Hamamatsu, Japan). A blinded evaluation of the scans was made by an experienced pathologist (ESR). For each section, H-score was calculated, measuring the percentage of the area exhibiting the different intensities (weak = A; moderate = B; strong = C) at 200× magnification (final score from 0 to 300), as follows [15,16,17]:

$$}=1}\left[ \%\, }\,}\right]+2}\left[ \%\, \; B}\right]+3}\left[ \% \,\; C}\right]$$

For statistical purposes, DMMs were merged with SMM patients and Welch’s ANOVA tests corrected for Dunnett T3 were performed. For the evaluation of the IHC scores as biomarkers, the sensitivity, specificity, Yuden index, and the area under the curve (AUC) were calculated with the online tool of the Johns Hopkins University (http://www.rad.jhmi.edu/jeng/javarad/roc/JROCFITi.html).

For IHC of HER2, MSTO cells were FFPE. IHC from breast ductal carcinoma in situ tissue was used as a positive control. For antigen retrieval, heat induction with EDTA buffer (pH 7.4) was used. An automated system (Ventana Benchmark Ultra, Roche Diagnostics, Basel, Switzerland) was used to apply Ventana anti-HER2/neu antibody (clone 4B5, rabbit monoclonal) in accordance with the manufacturer’s instructions.