Protein phosphatase 2 phosphatase activator (PTPA) promotes oncogene-induced senescence and carboplatin response in human malignant pleural mesothelioma cells Authors: Eleonora Nigliato, Yan Ma, Rita Derua, Maria Butyrin, Sandra P. Nunes, Léa Guilmot, Antonio De Gianni, Rüveyda Dok, Andrew P. Feld, Michael Ohlmeyer, Brian E. Wadzinski, Sandra Nuyts, Ilaria Elia, Anna Sablina, Veerle Janssens Published: 2026 ABSTRACT PurposeMalignant pleural mesothelioma (MPM) is a rare cancer of the pleura, frequently related to asbestos exposure. It is characterized by a grim prognosis, as few therapeutic options are available. Typically, MPM exhibits high molecular heterogeneity, providing opportunities for exploiting therapeutic vulnerabilities in well-defined subgroups of patients. Nevertheless, these options have been insufficiently explored.Methods & resultsWe found that inactivation of Protein Phosphatase 2A (PP2A) is a recurrent event in MPM, affecting the cancer cell phenotype and chemotherapy response. Specifically, in silico analysis showed that heterozygous loss of PTPA (PPP2R4), encoding the PP2A activator PTPA, occurred in 26% of MPM patients (TCGA, n = 87), correlating with significantly decreased overall survival. Re-expression of PTPA in PTPA-low human MPM cells decreased 2D colony formation and anchorage-independent cell growth, without affecting cell migration. At enzyme level, PTPA restoration increased PP2A catalytic subunit expression and activity (methylation), without favoring specific PP2A holoenzymes. Comparative RNAseq and Gene Set Enrichment Analysis revealed increased RAS pathway activation, activated p53 signaling and induction of senescence (SA-β-Galactosidase, SASP, NF-κB activation) upon PTPA restoration, presumably by increased PP2A-dependent dephosphorylation of Kinase Suppressor of Ras 1. Suppressing Ras signaling by Trametinib (MEK-inhibitor) prevented senescence induction, and pretreatment with senolytic agents restored impaired growth. ATUX-792, a pharmacologic PP2A activator, did not phenocopy senescence induction. Finally, PTPA restoration sensitized MPM cells to carboplatin as the likely consequence of lethal stress induction.ConclusionsPTPA promotes oncogene-induced senescence (OIS) in MPM. By preventing OIS, heterozygous PTPA loss may contribute to mesothelial transformation and carboplatin resistance.Supplementary InformationThe online version contains supplementary material available at 10.1007/s13402-026-01230-5. Introduction Malignant mesothelioma is a rare cancer with a unique biology that originates in the mesothelium, a protective cell layer covering body cavities, such as the pleura, the peritoneum, the pericardium, and the tunica vaginalis [1]. Among its different subtypes, malignant pleural mesothelioma (MPM) is the most common, accounting for 90% of cases [2]. The main risk factor for MPM is asbestos fibre inhalation, with a long latency period of 20–40 years between initial exposure and disease presentation [2, 3]. Despite asbestos being banned in more than 55 countries, global MPM incidence is still increasing, with approximately 2400 new cases reported each year [3, 4]. According to molecular genomics studies, MPM is a highly heterogeneous cancer that is largely defined by loss-of-function mutations in tumor suppressor genes, most frequently in CDKN2A, BAP1, NF2, LATS2, and TP53 [5, 6]. Hence, besides aberrant cell cycle checkpoint regulation, inactivated Hippo signaling, characterized by increased oncogenic YAP activity, is one of the major molecularly altered pathways [1]. Rarely, activating oncogenic driver mutations have been found (e.g. in KRAS, EGFR, HGFR and FGFR) [7–9] and signaling pathways such as mammalian Target of Rapamycin (mTOR) or Focal Adhesion Kinase (FAK) are only sporadically activated [1]. Despite this comprehensive molecular profiling, limited and ineffective treatment options exist for MPM. For decades, chemotherapy based on cisplatin (or carboplatin) and pemetrexed has been the standard of care; targeted treatments, although tested to some extent in clinical trials, have not shown any improved outcomes. Recently, the emergence of immune checkpoint inhibitors has shown promising results, though only in a small group of responder patients [2, 10]. Consequently, MPM prognosis remains poor, with an overall survival rate of < 1 year following diagnosis [11]. This highlights the critical need to better understand the molecular mechanisms behind MPM and to identify new biomarkers to guide targeted therapy [12]. Although the etiology of MPM is most often linked to asbestos exposure, other causes include radiation and infection with the small DNA tumor virus, Simian virus 40 (SV40) [13, 14]. SV40 infection in humans is indeed associated with development of very specific tumor types, such as mesothelioma, lymphoma, brain and bone cancers [15]. Although the detection rates of SV40 in human mesothelioma show considerable variability [14], SV40 induces mesothelioma in rodents following intrapleural or intracardiac injection [16]. In addition, when combined with asbestos, it acts as a co-carcinogen in the development of MPM [16]. The SV40 early region encodes three tumor antigens, large T (LT), small T (ST) and 17KT, which contribute to cellular transformation [17]. While LT primarily binds to the tumor suppressors pRb and p53 to achieve cell immortalization, ST exclusively targets protein phosphatase 2A (PP2A) [17], a large family of Ser/Thr-specific phosphatases with predominant tumor suppressive functions [18–20]. Active PP2A holoenzymes are composed of a catalytic C subunit, a scaffolding A subunit and a regulatory B-type subunit, which belongs to one of four different families (B/B55/PR55, B’/B56/PR61, B” or B”’/striatins) and directs PP2A functions to different substrates and pathways [18, 20]. Interaction of ST with PP2A occurs through the scaffolding Aα subunit [21–24], and subverts the function of several tumor suppressive PP2A trimers [25, 26], while promoting the function of oncogenic PP2A complexes [27]. Consequently, PP2A targeting by ST has proven a crucial factor to achieve full transformation of normal human cells [28, 29], including mesothelial cells [30]. In accordance, targeted genetic inactivation of specific PP2A complexes results in increased spontaneous and chemical-induced tumorigenesis in vivo [31–33], and in most human carcinomas and hematologic cancers, PP2A inactivation is a common event, albeit largely occurring by non-genomic mechanisms [18]. Nevertheless, the role of PP2A dysfunctions in the biology of MPM has remained largely unexplored [34, 35]. Here, we have comprehensively characterized genetic PP2A alterations in MPM patient samples using publicly available datasets and investigated their putative clinical relevance. Based on the alterations found, we focused on PTPA, the gene encoding the PP2A activating chaperone, PTPA [36, 37]. ATP-dependent PTPA activity is essential for PP2A holoenzyme biogenesis, promoting activation and carboxymethylation of the C subunit, while generating catalytically competent trimeric PP2A holoenzymes [36]. Previous cancer biology studies have shown that PTPA (previously named PPP2R4) is a haploinsufficient tumor suppressor gene, and heterozygous loss of PTPA is associated with decreased patient survival in several human carcinoma types [31], including in KRAS-mutant non-small cell lung cancer [38]. In the latter cancer type, PTPA heterozygous loss increases cancer cell-specific growth and differentially affects the therapeutic response of the cancer cells to MEK and TOR kinase inhibitors [38]. We now reveal an inhibitory role for PTPA in regulating MPM cancer cell-specific growth by promoting oncogene-induced senescence, while increased PTPA expression sensitizes MPM cells to carboplatin treatment. Collectively, our data provide the first insights into the role of PP2A and PTPA in MPM biology and chemotherapy response. Materials and methods Cancer databases and survival analysis The Pleural Mesothelioma (TCGA, PanCancer Atlas, 87 patients) study in the cBioPortal database [39–41] was queried for genomic alterations (i.e. GAIN, HETLOSS, MUT) in all PP2A subunit and regulator encoding genes (gene names starting with PPP2, striatins, carboxymethyl transferase and esterase) and in known cellular PP2A inhibitor genes (CIP2A, SET, ENSA, ARPP19, etc.) (Suppl. Fig. S1). Copy number alteration (derived from GISTIC or RAE analysis) and mutation analyses were performed on 82/87 samples (for which these genomic data were available) and analyzed in relation to overall patient survival. RNA expression data (available for all 87 samples) were examined by setting the thresholds of Z < -1 and Z > +1 (Suppl. Fig. S2), and these values were also correlated with overall survival. Overall survival analyses were performed in cBioPortal (Kaplan-Meier curves with log-rank tests) comparing PTPA hetloss and PTPA unaltered groups, and independently, using Cox proportional hazards models, including both univariate and multivariate analyses to identify independent prognostic factors. Statistical analyses and visualization were conducted in R using the packages survival, broom, dplyr, and ggplot2. Further analyses were performed in cBioPortal to compare differences in distribution of specific histological subtypes, tumor stages, genomic instability scores, age at diagnosis and treatment (only radiation therapy data were available) between the PP2A altered and unaltered patient groups, and to determine co-occurrence or mutual-exclusivity of defined PP2A alterations with each other, or with established MPM marker genes (BAP1, TP53, CDKN2A, CDKN2B, LATS2 and NF2). Cell lines A panel of cell lines representative of different histological MPM subtypes was used, including IST-MES1, IST-MES2, NCI-H2052, MSTO-211H, NCI-H2452, ZL34, and NCI-H226, along with an untransformed pleural mesothelial cell line MeT-5A. All cell lines were cultured at 37 °C and 5% CO₂ in RPMI-1640 medium (Gibco, #A1049101) supplemented with 10% fetal bovine serum (FBS; Biowest, #S1600) and penicillin/streptomycin (100 units/mL and 10 mg/mL, respectively; Sigma, #P0781). All cell lines were obtained from the American Type Culture Collection (ATCC), except for ZL34, which was a generous gift of Prof. L. Willems (ULg, Liège, Belgium) and NCI-H226, which was kindly provided by Prof. G. Halder (KU Leuven, Leuven, Belgium) (Table 1). IST-MES1 and IST-MES2 were purchased from IRCCS Ospedale Policlinico San Martino. Mycoplasma-free status was regularly tested with Venor™ GeM (Minerva Biolabs, #11–1050). Table 1MPM gene marker alterations in the three MPM cell lines used for phenotypic studiesZL34NCI-H226NCI-H2452BAP1/Hom del [42]Het loss [42] CDKN2A Hom del (this study)Hom del (CVCL_1544)Hom del (CVCL_1553) NF2 Hom del (this study)Hetloss [43]WT [43] TP53 WT (this study, [44])WT [45]Truncation [45] Table 1MPM gene marker alterations in the three MPM cell lines used for phenotypic studiesZL34NCI-H226NCI-H2452BAP1/Hom del [42]Het loss [42] CDKN2A Hom del (this study)Hom del (CVCL_1544)Hom del (CVCL_1553) NF2 Hom del (this study)Hetloss [43]WT [43] TP53 WT (this study, [44])WT [45]Truncation [45] MPM gene marker alterations in the three MPM cell lines used for phenotypic studies Lentiviral transduction HEK293T cells (ATCC) were cultured in DMEM (Sigma, #D6546) supplemented with 10% FBS, 2 mM L-glutamine (Sigma, #G7513), and penicillin/streptomycin (100 units/mL and 10 mg/mL, respectively). For lentivirus production, cells in a 6-well plate were co-transfected with 0.9 µg of pCMV-deltaR8.9, 0.1 µg of pMD.G-VSVG, and 1 µg of either a pLenti6.4(3xFLAG)- or pLKO-based expression plasmid using TurboFect™ (Thermo Fisher, #R0531). To generate EGFP pLenti6.4(3xFLAG) and PTPA pLenti6.4(3xFLAG), the EGFP and PTPA coding regions were cloned into pLenti6.4(3×FLAG) without removing the start codons, resulting in expression of both tagged and untagged GFP or PTPA. For stable knockdown experiments, two independent PTPA-targeting short hairpins (shPTPA1 and shPTPA2) and one short hairpin targeting EGFP (shGPP) were cloned into pLKO (sequences in Suppl. Table S1). The viral supernatant was collected 72 h post-transfection, passed through a 0.45 μm filter, and quantified using the Lentivirus qPCR Quantification Kit (Abm, #LV900). Subsequently, MPM or MeT-5A cells were seeded in a 24-well plate and transduced with a multiplicity of infection (MOI) of 10 for 24 h in the presence of Polybrene (Sigma, #TR-1003-G), and then selected with Blasticidin (Gibco, #A1113903) (pLenti6.4) or puromycin (Gibco, #A1113803) (pLKO). Protein lysates were prepared, and immunoblotting for PTPA was performed to confirm overexpression or knockdown. All subsequent experiments were executed with polyclonal populations of each transduced cell line. Protein extraction, nuclear/cytoplasmic fractionation, and western blotting Cells were collected and washed twice with ice-cold phosphate-buffered saline (PBS) (Sigma, #D8537) before being lysed in RIPA buffer (Thermo Fisher, #89900) supplemented with cOmplete™ protease inhibitors (Roche, #4693116001) and PhosSTOP™ phosphatase inhibitors (Roche, #4906845001) to obtain total cell lysates. Protein concentration of the lysates was determined using Pierce™ BCA (Invitrogen, #23225) to ensure equal loading. To obtain nuclear and cytoplasmic extracts, subcellular fractionation was performed with the NE-PER Nuclear and Cytoplasmic Extraction kit (Thermo Fisher, #7833), according to manufacturer’s instructions. For Western blotting, lysates or IPs were diluted in sample buffer (4X NuPAGE™ LBS,#NP0008 and Invitrogen, #NP007) and boiled (10 min, 96 °C). Total lysates (15–20 µg) were separated by SDS-PAGE on NuPAGE™ 4–12% Bis–Tris Midi Protein Gels (Invitrogen,#WG1402) at 150 V for 90 min. Proteins were transferred to a nitrocellulose membrane (Cytiva, #10600002) by wet blotting (60 min, 100 V). Membranes were stained with Ponceau (Thermo Fisher, #BP103) and blocked for 1 h at room temperature using either 5% non-fat dry milk (PanReac AppliChem,#A0830) or 0.1% bovine serum albumin (BSA, PanReac AppliChem, #A6588) in Tris-buffered saline (TBS)-Tween®20 (TBS-T). Membranes were incubated overnight at 4 °C with primary antibodies (Suppl. Table S2), followed by incubation with the corresponding HRP-conjugated secondary antibodies (Suppl. Table S2) for 1 h at room temperature. Protein bands were visualized with WesternBright® HRP ECL substrate (Advansta, #K-12045-D50) using Amersham™ ImageQuant™ 800 imager (Cytiva). Signals were quantified using Image Studio Lite (5.2) and ImageQuantTL (v10.2–499). Quantification of phospho-specific and total protein signals was performed on different blots by normalization to the vinculin signal (loading control), followed by calculation of the ratios between the normalized phospho-specific and normalized total protein signals. PP2A C subunit immunoprecipitation-mass spectrometry (IP-MS) ZL34 cells were seeded in 10 cm dishes at a density of 7.5-8 × 105 cells/dish. Following protein extraction, equal amounts of lysates of parental and PTPA-overexpressing cells were incubated with a GFP-tagged PP2A C nanobody (10 µg nanobody per 1 mg of lysate protein) for 1 h at 4 °C, followed by incubation with 20 µl of GFP-trapping beads (ChromoTek) pre-washed in NENT 100 buffer (20 mM Tris-HCl pH 7.4, 1 mM EDTA, 0.1% NP-40, 25% glycerol, and 100 mM NaCl) containing 1 mg/mL BSA, for an additional hour at 4 °C. Beads were then extensively washed: twice with 500 µL NENT 150 (NENT 100 with 150 mM NaCl instead); twice with NENT 100; twice with 50 mM Tris-HCl (pH 8.5), twice with Tris-HCl (pH 8.5) containing 0.0025% ProteaseMAX (Promega, #V2072), and subjected to overnight on-bead digestion in the presence of 50 mM Tris-HCl (pH 8.5), 0.01% ProteaseMAX and 0.5 µg trypsin (Thermo Fisher, #90058). After digestion, 100 µl of water was added to the beads. The supernatant was brought to 0.5% TFA and allowed to stand at room temperature for 15 min to degrade the ProteaseMAX. After centrifugation, the supernatant was desalted on Pierce C18 Spin Columns (Thermo Fisher Scientific). The eluate was vacuum-dried and resuspended in 20 µl of 5% CH3CN, 0.1% formic acid of which 5 µl was injected on a Xevo G3 QTOF (Waters) LC-MS/MS instrument operated in MSe mode. Data analysis was executed using PLGS™ software (Waters) using Uniprot Homo sapiens (reviewed) to which the GFP protein was added as a database. Oxidation (M) was included as a variable modification. Two missed trypsin cleavages were allowed. Semi-quantification of proteins was executed using the Hi3 method essentially as described in Silva et al. [46] Only ‘pass2’ peptides were taken into account. Samples were normalized based on the GFP tag of the nanobody. Missing values were imputed (average). Peptide abundances that were common to different PP2A isoforms were attributed to the individual isoforms pro rata of their Hi3 values only taking unique peptides into account. 2D colony formation MPM or MeT-5A cells were seeded at a density of 3000–5000 cells/dish in 10 cm dishes. After 5 days of incubation, the medium was carefully removed and replaced with fresh complete medium. After 12–14 days of growth, cell colonies were washed with PBS and stained with crystal violet (2 g/L in H2O; Sigma, #V5265) prepared in 2.5% glutaraldehyde solution (Sigma, #49629). Colony number was quantified using the GelCount software. 3D colony formation in soft agar 24-well plates were coated with culture medium containing 0.8-1% low-gelling temperature agarose (Sigma-Aldrich, #A9045). MPM cell lines were seeded at a density of 1–3 × 103 cells per well. Cells were resuspended in RPMI containing 20% FBS and 0.4–0.5% agarose. After solidification, medium was added on top, and plates were incubated at 37 °C, 5% CO2. After 21 days, cells were stained with Thiazolyl Blue Tetrazolium bromide (MTT, 3 mg/mL in PBS; Thermo Fisher, #L11939.06) at 37 °C for 4 h. Images were acquired using HP Scanjet G2710 Photo Scanner, and colonies were quantified using Fiji (ImageJ). Wound healing assay ZL34 cells were seeded at a density of 4 × 104 cells/well in a 96-well plate (Sartorius ImageLock, #4856) and incubated overnight. The wound was made using the WoundMaker™ pin block, followed by two washes with PBS. Cells were cultured in RPMI containing 2% FBS to prevent growth. Cell migration was monitored and imaged using an IncuCyte® S3 device (Sartorius) for 24 h. Relative wound density (%) was quantified using the IncuCyte 2024 B Software, defined as the ratio of the cell-occupied area of the initial wound area to the total wound area. Confluency assay in real-time MeT-5A cells were seeded at a concentration of 3000 cells per well in a 96-well plate. After 24 h, the cells were transfected with one of three predesigned DsiRNAs targeting human PTPA or a negative DsiRNA control (TriFECTa® DsiRNA kit (IDT), sequences in Suppl. Table S1) at a final concentration of 10 nM, using Lipofectamine™ RNAiMAX (Invitrogen, #13778075). Live-cell imaging was performed for 96 h in the Incucyte (Sartorius). Confluency was determined by using the AI confluency mask provided by the Incucyte software. The same setup was used for stably transduced shGFP and shPTPA MeT-5A cells. Cells were seeded at 3000 cells per well (96-well-plate) and imaged for 96 h in the Incucyte device. Bulk RNA-sequencing ZL34 cells with or without stable PTPA overexpression were seeded in a 6-well plate at a density of 2 × 105 cells/well. After 48 h, RNA extraction was performed using the RNeasy Mini Kit (Qiagen, #74104). The samples were subjected to DNase (Qiagen, #79254) digestion to remove any residual DNA. Samples with a 260 nm/280 nm ratio between 1.8 and 2.1 were further processed for RNAseq, which was outsourced to Novagen. Raw FASTQ reads were aligned to the human reference genome (GRCh38) using STAR (v2.7.3a) with GENCODE v42 annotation. Gene-level counts were generated with featureCounts based on the same annotation. Differential expression analysis was performed in R using DESeq2 following standard procedures. All protein-coding genes were ranked by Log2FoldChange and used to perform pre-ranked Gene Set Enrichment Analysis (GSEA) to identify enriched pathways. Raw and processed data are available in Gene Expression Omnibus (GEO) through accession number GSE312009. Compound treatments Carboplatin (Sanbio, #HY-17393), Trametinib (MedChemExpress, #HY-10999), Dasatinib (Merck, #SML2589) and Quercetin (MedChemExpress, #HY-18085) were prepared according to the manufacturer’s instructions. ATUX-792 (Atux Iskay LLC) was prepared and stored as described previously [47]. To obtain dose-response curves, MPM cells were seeded in 96-well plates at a density of 3,000 cells per well and treated with increasing concentrations of ATUX-792 (0.39, 0.78, 1.56, 3.13, 6.25, 12.5, 25 and 50 µM) or carboplatin (2.5, 5, 15, 50, 100, 120, 180, 200 and 300 µM) for 72 h. Cell viability was assessed using Thiazolyl Blue Tetrazolium bromide (MTT) (Alfa Aesar, #L11939.06). Briefly, 20 µL of MTT solution (2 mg/mL) was added to each well and incubated at 37 °C for 4 h. Following incubation, the medium was removed, and formazan crystals were solubilized in 200 µL DMSO. Absorbance was measured at 550 nm, and EC50 values were calculated using logistic regression in GraphPad Prism 8.0.1 (GraphPad Software; RRID: SCR_002798). For short-term treatment assays, MPM cells were seeded in 6-well plates at a density of 80,000 cells per well. After 24 h, cells were treated with Trametinib at 25 µM (NCI-H2452) or 15 µM (ZL34 cells), or with ATUX-792 at 6 µM (ZL34) or 8 µM (NCI-H2452). These concentrations were pre-determined in MTT assays. After 48 h of treatment, protein extraction was performed. The same experimental setup was used for 48 h of treatment with dasatinib (10 nM) and quercetin (1 µM), after which the surviving cells were re-seeded at 10,000 cells per well in 24-well plates and allowed to grow for one week. At the end of the experiment, cells were stained with crystal violet, and images were acquired using an HP Scanjet G2710 Photo Scanner. Cell confluence was quantified using Fiji (ImageJ). For long-term treatment, ZL34 cells were seeded in 6-well and 24-well plates at a density of 100,000 and 7500 cells per well, respectively. After 24 h, cells were treated with the indicated concentration of carboplatin for a duration of 12 days (25, 50, 75 µM). The treatment medium was refreshed every three days. At the end point, cells were fixed using crystal violet (2 g/L in H2O) prepared in 2.5% glutaraldehyde solution. Images were acquired using HP Scanjet G2710 Photo Scanner, and confluence was quantified using Fiji (ImageJ). The same experimental setup was used for ZL34 cells pretreated with dasatinib/quercetin for 48 h, and which were then further treated with a single concentration of carboplatin (50 µM) for 12 days. Flowcytometric cell cycle analysis ZL34 cells were treated with 50 µM carboplatin for 12 days, as described above. Following treatment, cells were fixed in 70% ethanol (ThermoFisher Scientific, #397690010) and stained with 10 µg/mL propidium iodide (PI) (Sigma-Aldrich, #P4170) supplemented with 100 µg/mL RNase A (Invitrogen, #12091021). Cell cycle profiles were analyzed by flow cytometry using a FACSVerse instrument (BD Biosciences). Senescence associated (SA)-β-galactosidase activity assays For colorimetric SA-β-Gal activity quantification, GFP- or PTPA-overexpressing ZL34 cells were seeded in a 6-well plate at a density of 15,000 cells per well and cultured for 5 days. Cellular senescence was assessed using the Senescence β-Galactosidase Staining Kit (Cell Signaling Technology, #9860) according to the manufacturer’s instructions. Briefly, cells were fixed and incubated with β-galactosidase staining solution containing X-gal at pH 6. Plates were then incubated overnight at 37 °C in the absence of CO₂. Image analysis was performed using QuPath v0.6.0 [48] with the QuPath extension Cellpose0.11.3 [49] installed. Regions corresponding to 50–70% confluent cell areas were manually annotated, and cell detection was carried out using Cellpose-SAM [50] on the optical density sum of the images. Subsequently, a training dataset was generated from representative images to train an object classifier distinguishing senescent from non-senescent cells, enabling quantification of the relative percentage of senescent cells per condition (all channels were included except DAB). For flow cytometric SA-β-Gal activity quantification, parental and PTPA OE cells were seeded in 10 cm culture dishes at a density of 1.5 × 10⁵ cells per dish and cultured for 5 days, after which they were harvested and fixed in 2% paraformaldehyde in PBS for 10 min. SA-β-Gal activity was then assessed by flow cytometry using the CellEvent™ Senescence Green Flow Cytometry Assay Kit (Invitrogen, #C10850) according to the manufacturer’s instructions. Cells were analysed in a BD FACSymphony™ A1 Cell Analyzer (Waters Biosciences). FlowjoTM 10.10.0 software (Waters Biosciences) was used for data analysis. Results In silico analysis of PP2A alterations in MPM patient dataset To obtain a comprehensive image of potential PP2A alterations in MPM, we analyzed a publicly available Pleural Mesothelioma dataset (TCGA, PanCancer Atlas, 87 patients) in silico, using cBioPortal [39–41]. We evaluated genomic alterations in PP2A-related genes (canonical PP2A subunits and several PP2A regulators) by determining the frequency of gene amplification/gain, homozygous or heterozygous loss, and mutations (Suppl. Fig. S1). Additionally, RNA expression levels of these genes were analyzed (Suppl. Fig. S2). All identified alterations were then correlated with overall survival (OS) to identify potential clinical relevance. In accordance with findings in most other cancer types [18], PP2A gene mutations were overall rare (< 1.5%), as were gene amplifications (< 4%) and homozygous losses (< 1.5%) (Suppl. Fig. S1). The bulk of changes appeared either as shallow deletions (HETloss) or shallow gains (GAIN), with some genes predominantly affected by shallow deletions (e.g. PPP2R2C, PPP2R2D, PPP2R5C, PPP2R5E, PPP2R3C, STRN3, PPP2R4, SET, MASTL and PABIR1), and others mainly by shallow gains (e.g. PPP2R5A, PPP2R5B, PPP2R3A, PPME1, TIPRL, CIP2A, ENSA and ANP32E) (Suppl. Fig. S1). Three genes, PPP2R2A (encoding the regulatory B55α subunit), PPP2R4 (encoding PTPA), and CIP2A (encoding Cellular Inhibitor of PP2A), were not only commonly altered among MPM patients (Fig. 1A and 1B), but were also associated with significant alterations in overall survival of patients (Fig. 1C–G), suggesting clinical relevance. Specifically, shallow PPP2R4 deletion (26%) was linked to lower OS (Fig. 1C), correlating with significantly decreased mRNA expression compared to tumors with normal PPP2R4 copy number (Fig. 1D). 23–24% of MPM tumors presented with CIP2A copy number gain (Fig. 1A) or increased mRNA expression (Fig. 1B), significantly correlating with worse OS (Fig. 1F) - in accordance with an earlier in silico study [34]. Lastly, PPP2R2A showed 20% heterozygous loss and 15% copy number gain (Fig. 1A), correlating well with alterations in mRNA expression (Fig. 1B). When comparing PPP2R2A gain to heterozygous loss, a trend toward decreased OS (p = 0.136) was observed when B55α expression was high (Fig. 1G). As internal control, significant differences in overall survival in this MPM cohort were also found for patients with loss or mutation of NF2, or with loss of CDKN2A/2B (Suppl. Fig. S3), previously described as markers with prognostic potential in independent MPM cohort studies [6, 51]. Fig. 1Genomic and transcriptional alterations of PPP2R2A, PPP2R4 and CIP2A in MPM TCGA, PanCancer Atlas dataset (n = 87). (A) OncoPrint diagram showing the copy number alterations (CNA) of PPP2R2A, PPP2R4, and CIP2A in MPM sample set (n = 82) obtained through cBioPortal (TCGA, PanCancer Atlas). Each column represents a patient sample, and each row represents a gene. Alteration types include mutations (MUT), deep deletions (–2: homozygous loss, HOM), shallow deletions (–1: heterozygous loss, HET), gains (+ 1: low-level increase, GAIN) and amplifications (+ 2: high-level increase, AMP). (B) OncoPrint diagram showing the mRNA expression profiles of PPP2R2A, PPP2R4, and CIP2A obtained through cBioPortal (TCGA, PanCancer Atlas, n = 87). Each column represents a patient sample, and each row represents a gene. mRNA expression levels are represented as Z-scores, with high expression defined as Z ≥ 1 and low expression as Z ≤ − 1. (C) Correlation of PPP2R4 copy number alterations (CNA) with overall survival. Significance is determined by log-rank test (p < 0.05). (D) PTPA mRNA expression positively correlates with copy number status (unpaired t-test, **: p < 0.01). (E) Overall survival analyses were independently performed using Cox proportional hazards models, including both univariate and multivariate analyses to identify independent prognostic factors in the malignant pleural mesothelioma TCGA PanCancer Atlas dataset. The plots show hazard ratios (HR, 95% confidence interval) on a logarithmic scale for overall survival according to PTPA status (PTPA hetloss versus unaltered), age at diagnosis, tumor stage, tumor histology (epithelioid versus non-epithelioid) and radiotherapy treatment (yes versus no). P-values < 0.05 can be considered significant. (F) Correlation of CIP2A mRNA expression with overall survival. Z-score cutoffs of + 1 and − 1 define high and low expression groups. Significance is determined by log-rank test (p < 0.05). (G) Correlation of PPP2R2A copy number alterations with overall survival, comparing patients grouped by gain (+ 1) with patients grouped by heterozygous loss (–1) of PPP2R2A. Significance is determined by log-rank test (p < 0.05) Genomic and transcriptional alterations of PPP2R2A, PPP2R4 and CIP2A in MPM TCGA, PanCancer Atlas dataset (n = 87). (A) OncoPrint diagram showing the copy number alterations (CNA) of PPP2R2A, PPP2R4, and CIP2A in MPM sample set (n = 82) obtained through cBioPortal (TCGA, PanCancer Atlas). Each column represents a patient sample, and each row represents a gene. Alteration types include mutations (MUT), deep deletions (–2: homozygous loss, HOM), shallow deletions (–1: heterozygous loss, HET), gains (+ 1: low-level increase, GAIN) and amplifications (+ 2: high-level increase, AMP). (B) OncoPrint diagram showing the mRNA expression profiles of PPP2R2A, PPP2R4, and CIP2A obtained through cBioPortal (TCGA, PanCancer Atlas, n = 87). Each column represents a patient sample, and each row represents a gene. mRNA expression levels are represented as Z-scores, with high expression defined as Z ≥ 1 and low expression as Z ≤ − 1. (C) Correlation of PPP2R4 copy number alterations (CNA) with overall survival. Significance is determined by log-rank test (p < 0.05). (D) PTPA mRNA expression positively correlates with copy number status (unpaired t-test, **: p < 0.01). (E) Overall survival analyses were independently performed using Cox proportional hazards models, including both univariate and multivariate analyses to identify independent prognostic factors in the malignant pleural mesothelioma TCGA PanCancer Atlas dataset. The plots show hazard ratios (HR, 95% confidence interval) on a logarithmic scale for overall survival according to PTPA status (PTPA hetloss versus unaltered), age at diagnosis, tumor stage, tumor histology (epithelioid versus non-epithelioid) and radiotherapy treatment (yes versus no). P-values < 0.05 can be considered significant. (F) Correlation of CIP2A mRNA expression with overall survival. Z-score cutoffs of + 1 and − 1 define high and low expression groups. Significance is determined by log-rank test (p < 0.05). (G) Correlation of PPP2R2A copy number alterations with overall survival, comparing patients grouped by gain (+ 1) with patients grouped by heterozygous loss (–1) of PPP2R2A. Significance is determined by log-rank test (p < 0.05) To provide further insights into the overall survival differences between PP2A altered and unaltered groups, putative significant associations with known genetic MPM alterations, such as loss or mutation of BAP1, TP53, NF2, LATS2, CDKN2A and CDKN2B, or with clinical parameters, such as tumor histology, stage, aneuploidy, age at diagnosis and received treatment, were additionally assessed. Regarding treatment, only data on radiotherapy were available. For the patient group with PTPA heterozygous loss, the decrease in overall survival could not be attributed to significant differences in tumor histology, stage, radiotherapy treatment or age at diagnosis (Suppl. Fig. S4A-S4D), while there was a significant correlation with tumor aneuploidy (Suppl. Fig. S4E). This is suggestive for increased genomic instability in tumors with PTPA hetloss, a feature that is often linked with more aggressive tumor behavior. Accordingly, univariate as well as multivariate analysis considering age at diagnosis, tumor stage, histology, radiotherapy and PTPA status, showed that PTPA hetloss is strongly predictive for worse prognosis (Fig. 1E), as well as non-epithelioid histology (i.e. an increased sarcomatoid tumor component) – an already well-known independent prognostic factor [6]. PTPA hetloss did not further significantly correlate or was neither mutually exclusive with any of the commonly altered MPM marker genes (Suppl. Fig. S4F). Similar analyses for CIP2A (Suppl. Fig. S5) and PPP2R2A stratified groups (Suppl. Fig. S6) neither revealed any significant associations with clinical parameters or recurrent MPM genetic alterations, except for the tumors with high CIP2A expression, which significantly correlated with increased tumor break load, another feature of increased genomic instability (Suppl. Fig. S5F). Finally, increased CIP2A expression significantly co-occurred with PTPA hetloss (Suppl. Fig. S7A), and survival analysis taking any of the three relevant PP2A alterations into account (PTPA hetloss, PPP2R2A gain and high CIP2A expression) increased the statistical significance of the difference in prognosis between ‘PP2A’ altered and unaltered groups (Suppl. Fig. S7B). PTPA expression is decreased in human MPM cell lines Since alterations in PTPA (PPP2R4) expression had not been previously linked to MPM, we focused on this gene to clarify its role in MPM biology. To identify suitable MPM cell models for further studies, PTPA protein levels were checked in a panel of seven human MPM cell lines (3 epithelioid, 1 sarcomatoid, 3 biphasic subtypes) and one normal human pleural mesothelium cell line (MeT-5A) via western blotting. In accordance with cBioPortal data (i.e. 26% hetloss vs. 4% of PPP2R4 gain), we predominantly found significantly decreased PTPA expression across MPM cells, with all MPM cell lines tested showing much lower quantified PTPA protein levels than untransformed mesothelial MeT-5A cells (Fig. 2A). PTPA expression suppresses pleural mesothelioma growth We next restored PTPA expression in three MPM cell lines (ZL34, NCI-H2452, and NCI-H226) by stably overexpressing FLAG-tagged PTPA or FLAG-tagged GFP as a control, using lentiviral transduction, and confirmed PTPA overexpression by immunoblotting (Fig. 2B). We then assessed colony formation ability in 2D and anchorage-independent growth in 3D. Compared to GFP overexpression, we found a significant decrease in the number of colonies in all MPM cell lines overexpressing PTPA (Fig. 2C), which was further validated in 3D colony formation assays in soft agar (Fig. 2D; NCI-H2452 cells were unable to grow as colonies in this assay). These findings suggest that decreased PTPA expression promotes cancer cell-specific growth, in line with the clinical survival data (Fig. 1C) and published findings on PTPA-regulated growth effects in other human cancer types [26, 31, 38]. We also assessed the effects of PTPA re-expression on ZL34 cell migration using scratch wound assays, but found no significant changes (Suppl. Fig. S8). Fig. 2PTPA re-expression inhibits MPM (anchorage-independent) cell growth. (A) PTPA protein expression is decreased in MPM cells as opposed to normal mesothelial cells. Western blot analysis and quantification of PTPA expression across different mesothelial and MPM cell lines: MeT-5A (normal pleural mesothelium), IST-MES1, IST-MES2, and NCI-H226 (epithelioid MPM), NCI-H2052 (sarcomatoid MPM), MSTO-211H, NCI-H2452 and ZL34 (biphasic MPM). Expression levels were normalized to β-actin in each cell line, and compared to MeT-5A (n = 3, one-way ANOVA, **: p < 0.01, ***: p < 0.001). (B) Stable re-expression of FLAG-tagged PTPA in three MPM cell lines. Cells were transduced with lentiviral vectors expressing FLAG-GFP (GFP, control) or FLAG-PTPA (PTPA OE). The upper band in the anti-PTPA immunoblot (IB) represents FLAG-tagged PTPA, while the lower band corresponds to untagged PTPA. The dotted line indicates lanes originally not loaded adjacently. (C) 2D colony formation assays and (D) 3D soft agar growth assays in MPM cell lines overexpressing FLAG-tagged GFP (GFP) or FLAG-tagged PTPA (PTPA OE). Colony numbers were quantified, and statistical analysis was performed using an unpaired t-test (n = 3–4; *p < 0.05; **p < 0.01; ***p < 0.001), except for NCI-H226, which was analyzed with a paired t-test for the 2D growth due to unusual variation in absolute colony number between independent replicates (n = 4; *p < 0.05; **p < 0.01; ***p < 0.001) PTPA re-expression inhibits MPM (anchorage-independent) cell growth. (A) PTPA protein expression is decreased in MPM cells as opposed to normal mesothelial cells. Western blot analysis and quantification of PTPA expression across different mesothelial and MPM cell lines: MeT-5A (normal pleural mesothelium), IST-MES1, IST-MES2, and NCI-H226 (epithelioid MPM), NCI-H2052 (sarcomatoid MPM), MSTO-211H, NCI-H2452 and ZL34 (biphasic MPM). Expression levels were normalized to β-actin in each cell line, and compared to MeT-5A (n = 3, one-way ANOVA, **: p < 0.01, ***: p < 0.001). (B) Stable re-expression of FLAG-tagged PTPA in three MPM cell lines. Cells were transduced with lentiviral vectors expressing FLAG-GFP (GFP, control) or FLAG-PTPA (PTPA OE). The upper band in the anti-PTPA immunoblot (IB) represents FLAG-tagged PTPA, while the lower band corresponds to untagged PTPA. The dotted line indicates lanes originally not loaded adjacently. (C) 2D colony formation assays and (D) 3D soft agar growth assays in MPM cell lines overexpressing FLAG-tagged GFP (GFP) or FLAG-tagged PTPA (PTPA OE). Colony numbers were quantified, and statistical analysis was performed using an unpaired t-test (n = 3–4; *p < 0.05; **p < 0.01; ***p < 0.001), except for NCI-H226, which was analyzed with a paired t-test for the 2D growth due to unusual variation in absolute colony number between independent replicates (n = 4; *p < 0.05; **p < 0.01; ***p < 0.001) For completion, we also performed the reciprocal experiment, lowering normal PTPA expression in untransformed mesothelial Met-5A cells by an shRNA or siRNA approach (Suppl. Fig. S9). Cells with a stably integrated PTPA-targeted short hairpin (shPTPA1 or shPTPA2) showed pronounced reduction in PTPA expression correlating with an obvious growth inhibition, as measured in a confluency and an independent colony growth assay (Suppl. Fig. S9A-S9C). Likewise, three independent PTPA siRNAs led to a pronounced reduction in PTPA expression and growth inhibition, further culminating in a complete growth arrest at 96 h post-transfection (Suppl. Fig. S9D-S9E). These results do not contradict the PTPA reconstitution data in PTPA-low MPM cells (Fig. 2), but probably just reflect the known essentiality of PTPA - with complete loss of PTPA being lethal in mice [31] and yeast [52, 53]. To achieve a growth promoting effect, reduced PTPA expression likely needs to occur in a real cancer-context in the presence of intrinsically activated oncogenic signaling (as demonstrated before in KRAS mutant A549 cells [38]), or be combined with expression of an activated oncogene (as demonstrated before in non-transformed HEK293-TER cells expressing an oncogenic mutant of HRAS [26]). Biochemical effects of PTPA overexpression on PP2A C expression, carboxy-terminal methylation and holoenzyme assembly PTPA plays a crucial role in promoting activation and carboxymethylation of the PP2A C subunit, thereby regulating PP2A holoenzyme biogenesis [36]. Interestingly, we found a marked increase in overall PP2A C levels in PTPA overexpressing ZL34 cells, along with a slight but statistically significant reduction in C subunit demethylation (Fig. 3A). These findings were independently confirmed in PTPA-overexpressing NCI-H226 cells (Suppl. Fig. S10A). To determine whether this also led to enhanced formation of specific PP2A holoenzymes, some of which are known to be highly methylation-dependent [54], we performed PP2A C immunoprecipitation using a GFP-tagged PP2A C nanobody, followed by mass spectrometry to quantify the interacting subunits relative to the C subunit in normal and PTPA overexpressing conditions (Fig. 3B). While we retrieved the PP2A A subunit (sum of α and β isoforms), seven B-type subunits (B55α, B55δ, B56α, β, γ, δ, ε) and one PP2A regulator (PABIR1), we did not observe a selective increase in any particular of these PP2A C interactors upon PTPA overexpression (Fig. 3B) - further confirmed by immunoblotting analysis of the PP2A C IPs for pan-B55, B56δ, B56ε and PABIR1 (also called FAM122A) (Suppl. Fig. S10B). Immunoblot analysis of total ZL34 cell lysates neither showed any changes in pan-B55, B56δ, B56ε or PABIR1 expression upon PTPA overexpression (Suppl. Fig. S10C). Normalization of C subunit levels (sum of α and β isoforms) to the GFP-tag of the nanobody independently confirmed the increase in PP2A C expression upon PTPA overexpression (Suppl. Fig. S10D). Thus, although PTPA overexpression led to increased C subunit expression and methylation, the overall stoichiometry of the detected PP2A subunit complexes remained unchanged, with PPP2R2A/B55α-containing PP2A complexes consistently being the most abundant PP2A holoenzymes found in MPM cells in both conditions (Fig. 3B). Fig. 3Biochemical effects of PTPA overexpression on PP2A C subunit expression, carboxymethylation and holoenzyme assembly. (A) Representative immunoblots showing levels of total PP2A C and demethylated PP2A C, with vinculin used as a loading control. Quantification of total PP2A C and demethylated PP2A C was performed by measuring band intensities. The ratio of demethylated PP2A C to total PP2A C was calculated for each condition. Statistical analysis was performed using an unpaired t-test (n = 3; *p < 0.05; **p < 0.01). (B) PP2A C immunoprecipitation followed by mass spectrometry (IP-MS) was performed to assess PP2A holoenzyme composition in parental or PTPA overexpressing (PTPA OE) conditions. Semi-quantification of displayed PP2A subunits was performed using the Hi3 peptide quantification method, and data were expressed relative to PP2A C (α and β isoforms together) in each sample (n = 3) Biochemical effects of PTPA overexpression on PP2A C subunit expression, carboxymethylation and holoenzyme assembly. (A) Representative immunoblots showing levels of total PP2A C and demethylated PP2A C, with vinculin used as a loading control. Quantification of total PP2A C and demethylated PP2A C was performed by measuring band intensities. The ratio of demethylated PP2A C to total PP2A C was calculated for each condition. Statistical analysis was performed using an unpaired t-test (n = 3; *p < 0.05; **p < 0.01). (B) PP2A C immunoprecipitation followed by mass spectrometry (IP-MS) was performed to assess PP2A holoenzyme composition in parental or PTPA overexpressing (PTPA OE) conditions. Semi-quantification of displayed PP2A subunits was performed using the Hi3 peptide quantification method, and data were expressed relative to PP2A C (α and β isoforms together) in each sample (n = 3) PTPA overexpression results in activation of the KSR1/MEK/ERK pathway, increased p53 tumor suppressor activity and p21 expression, and induction of cellular senescence To gain mechanistic insights into the signaling pathways underlying decreased growth of PTPA overexpressing MPM cells, we performed bulk RNAseq in both GFP control (n = 3) and PTPA OE (n = 3) ZL34 cells, followed by pathway and gene set enrichment analysis. Analysis of all PP2A genes within the obtained gene sets confirmed PPP2R4/PTPA overexpression at RNA level (as expected), while no other PP2A genes, including PPP2CA and PPP2CB, showed significant changes in expression (Suppl. Fig. S11). This suggests that the increase in PP2A C subunit protein observed upon PTPA overexpression (Fig. 3A) likely occurred through increased translation or increased protein stabilization. Looking at all whole transcriptome changes in an unbiased manner, volcano plot filtering identified differentially expressed genes between control and PTPA overexpressing cells: a total of 384 genes were significantly upregulated (log2fold-change > 1, and padj < 0.05), while 178 genes were significantly downregulated (log2fold-change < -1, and padj < 0.05) (Fig. 4A). Subsequent HALLMARK pathway enrichment analysis unexpectedly revealed increased RAS pathway activation in PTPA OE cells (Fig. 4B–C), validated by increased phospho-MEK and phospho-ERK, and decreased phospho-KSR1 levels (Fig. 4D). As KSR1 (Kinase Suppressor of RAS 1) phospho-Ser392 is an established direct PP2A substrate site inhibiting RAS pathway activation [55–57], PTPA re-expression and subsequent PP2A activation may thus directly stimulate KSR1 dephosphorylation, thereby promoting activated Ras signaling. Fig. 4PTPA re-expression results in activation of the KSR1/RAF/MEK/ERK pathway. (A) Volcano plot showing differential gene expression between GFP control and PTPA-overexpressing ZL34 cells (n = 3). Red: upregulated in PTPA OE (log2fold-change > 1, padj < 0.05); blue: downregulated in PTPA OE (log2fold-change < -1, padj < 0.05); grey: -1 < log2fold-change < 1 or padj > 0.05. (B) Gene set enrichment analysis (GSEA) bubble plot showing significantly enriched signaling pathways in PTPA-overexpressing versus GFP control ZL34 cells, based on RNAseq analysis of both stable cell lines (n = 3). (C) GSEA plot showing enrichment of the ‘HALLMARK KRAS SIGNALING UP’ pathway, with normalized enrichment score (NES) and false discovery rate (FDR) indicated. (D) Increased KRAS signaling in PTPA OE cells, underscored by decreased phospho-KSR1(Ser392)/total KSR1, increased phospho-MEK1/2(Ser217/221)/total MEK1/2, and increased phospho-ERK1/2(Thr202/Tyr204)/total ERK1/2, as determined by immunoblotting with indicated antibodies, followed by quantification of the bands. Statistical analysis was performed using an unpaired t-test (n = 3, *p < 0.05) As activated Ras signaling coincided with decreased growth (Fig. 2), we hypothesized PTPA re-expression might actually be promoting oncogene-induced senescence (OIS), a cellular response to oncogene-induced replicative, oxidative or metabolic stress that prevents the active proliferation of genomically unstable cells. In support of this hypothesis, GSEA analysis revealed increased p53 pathway activation (Figs. 4B and 5A), further validated by increased p21 expression (CDKN1A: p53 target gene) at the protein level (Fig. 5B) and increased markers of DNA-damage (phospho-ATR/total ATR and γH2AX) (Fig. 5B). Consistent with the results observed in ZL34, a pronounced increase in p21 expression was also detected in PTPA re-expressing NCI-H226 and NCI-H2452 cells (Suppl. Fig. S12). Activation of p53, either by the DNA Damage Response (DDR) pathway or DDR-independent mechanism [58, 59], is indeed recognized as the major driver of OIS in response to activated Ras signaling [60–62]. In accordance, GSEA analysis using the SenMayo gene set revealed significant positive enrichment in PTPA OE cells (Fig. 5C), a finding that was further corroborated by increased SA-β-Galactosidase activity (Fig. 5D–E , Suppl. Fig. S13) and the upregulation of genes associated with the Senescence-Associated Secretory Phenotype (SASP), including IL6, MMP1, MMP3, and IL1A (Fig. 5F). Finally, our RNAseq data revealed enrichment of TNFα signaling via the NF-κB gene set (Figs. 4B and 5G): NF-κB activation is also associated with cellular senescence as the master regulator of gene transcription of SASP [63, 64]. Consistently, subcellular fractionation followed by immunoblotting revealed elevated levels of NF-κB p65 in the nuclear compartment of PTPA OE cells (Fig. 5H), supporting activation of NF-κB-mediated transcription. Thus, we concluded that PTPA re-expression in PTPA-low ZL34 cells promotes RAS pathway activation and oncogene-induced senescence, thereby likely explaining the observed decrease in MPM cell growth (Fig. 2). PTPA re-expression results in activation of the KSR1/RAF/MEK/ERK pathway. (A) Volcano plot showing differential gene expression between GFP control and PTPA-overexpressing ZL34 cells (n = 3). Red: upregulated in PTPA OE (log2fold-change > 1, padj < 0.05); blue: downregulated in PTPA OE (log2fold-change < -1, padj < 0.05); grey: -1 < log2fold-change < 1 or padj > 0.05. (B) Gene set enrichment analysis (GSEA) bubble plot showing significantly enriched signaling pathways in PTPA-overexpressing versus GFP control ZL34 cells, based on RNAseq analysis of both stable cell lines (n = 3). (C) GSEA plot showing enrichment of the ‘HALLMARK KRAS SIGNALING UP’ pathway, with normalized enrichment score (NES) and false discovery rate (FDR) indicated. (D) Increased KRAS signaling in PTPA OE cells, underscored by decreased phospho-KSR1(Ser392)/total KSR1, increased phospho-MEK1/2(Ser217/221)/total MEK1/2, and increased phospho-ERK1/2(Thr202/Tyr204)/total ERK1/2, as determined by immunoblotting with indicated antibodies, followed by quantification of the bands. Statistical analysis was performed using an unpaired t-test (n = 3, *p < 0.05) Fig. 5PTPA re-expression results in increased DNA-damage, induction of p53 tumor suppressor activity and cellular senescence. (A) Enrichment plot for the ‘HALLMARK p53 pathway’ gene set. (B) Increased DNA-damage marker (phospho-ATR/total ATR and γH2AX) and CDKN1A (p53 target gene, encoding p21 CIP/KIP) expression in PTPA OE cells was confirmed by immunoblot analysis. Quantification of p21 and γH2AX signal intensity was normalized to vinculin (loading control), with GFP control condition set to 1 (unpaired t-test, n = 3–4, *p < 0.05, **p < 0.01). Phospho-ATR signals were normalized to total ATR protein, and set to 1 in the GFP control condition (unpaired t-test, n = 3). (C) GSEA enrichment plot for the ‘SenMayo’ gene set, underscoring increased senescence in PTPA OE cells. (D) Colorimetric SA-β-Gal activity assay. Quantification of blue, senescent cells in GFP control versus PTPA OE ZL34 cells was executed in QuPath (n = 6, unpaired t-test, **p < 0.01). Representative pictures can be found in Suppl. Fig. S13. (E) Quantification of SA-β-Gal-positive cells by flow cytometry in parental versus PTPA OE ZL34 cells. SA-β-Gal-positive cells were labelled with a green fluorescent marker and quantified in the flow cytometer. Remark that the proper lentivirally transduced GFP cell line could therefore not be used as a control in this experiment (n = 6, unpaired t-test). (F) Heat map representing the expression levels of genes associated with cellular senescence and the Senescence-Associated Secretory Phenotype (SASP) throughout all RNAseq replicates. (G) Enrichment plot for the ‘HALLMARK TNFα signaling via NF-κB’ gene set. (H) Western blot analysis of NF-κB p65 in cytoplasmic and nuclear fractions. Sp1 was used as a nuclear marker and loading control. On the right, quantification of NF-κB levels in the nuclear compartment is shown (n = 2), normalized to Sp1 PTPA re-expression results in increased DNA-damage, induction of p53 tumor suppressor activity and cellular senescence. (A) Enrichment plot for the ‘HALLMARK p53 pathway’ gene set. (B) Increased DNA-damage marker (phospho-ATR/total ATR and γH2AX) and CDKN1A (p53 target gene, encoding p21 CIP/KIP) expression in PTPA OE cells was confirmed by immunoblot analysis. Quantification of p21 and γH2AX signal intensity was normalized to vinculin (loading control), with GFP control condition set to 1 (unpaired t-test, n = 3–4, *p < 0.05, **p < 0.01). Phospho-ATR signals were normalized to total ATR protein, and set to 1 in the GFP control condition (unpaired t-test, n = 3). (C) GSEA enrichment plot for the ‘SenMayo’ gene set, underscoring increased senescence in PTPA OE cells. (D) Colorimetric SA-β-Gal activity assay. Quantification of blue, senescent cells in GFP control versus PTPA OE ZL34 cells was executed in QuPath (n = 6, unpaired t-test, **p < 0.01). Representative pictures can be found in Suppl. Fig. S13. (E) Quantification of SA-β-Gal-positive cells by flow cytometry in parental versus PTPA OE ZL34 cells. SA-β-Gal-positive cells were labelled with a green fluorescent marker and quantified in the flow cytometer. Remark that the proper lentivirally transduced GFP cell line could therefore not be used as a control in this experiment (n = 6, unpaired t-test). (F) Heat map representing the expression levels of genes associated with cellular senescence and the Senescence-Associated Secretory Phenotype (SASP) throughout all RNAseq replicates. (G) Enrichment plot for the ‘HALLMARK TNFα signaling via NF-κB’ gene set. (H) Western blot analysis of NF-κB p65 in cytoplasmic and nuclear fractions. Sp1 was used as a nuclear marker and loading control. On the right, quantification of NF-κB levels in the nuclear compartment is shown (n = 2), normalized to Sp1 Activated Ras signaling, increased p21 expression and induced senescence cause the observed growth decrease in PTPA overexpressing MPM cells To determine whether increased Ras signaling is indeed the cause of senescence induction in PTPA re-expressing cells, we used the MEK inhibitor Trametinib to suppress RAF/MEK/ERK signaling. Trametinib treatment abolished ERK activation (phospho-ERK/total ERK) as expected, while also completely reversing induced p21 expression in PTPA overexpressing ZL34 (Fig. 6A) and NCI-2452 cells (Fig. 6B). Further, pretreatment of ZL34 cells with senolytic agents dasatinib/quercetin for 48 h, enabled a full rescue of the observed growth defect of the surviving cells in a confluency assay (Fig. 6C), underscoring that PTPA re-expression-induced growth inhibition is indeed caused by oncogene-induced senescence. Fig. 6Activated Ras signaling causes senescence induction, and OIS causes decreased growth in PTPA OE MPM cells. (A-B) ZL34 and NCI-H2452 control and PTPA-overexpressing cells were treated for 48 h with 15 µM and 25 µM Trametinib, respectively, and harvested for protein extraction. Immunoblot analysis of the lysates demonstrated decreased ERK phosphorylation (phospho-ERK/total ERK), alongside decreased p21 levels (senescent marker) in ZL34 (A) and NCI-H2452 cells (B) (n = 3 (ZL34) and n = 4 (NCI-H2452); two-way ANOVA, *p < 0.05, ***p < 0.001). (C) GFP control and PTPA OE ZL34 cells were pre-treated with 10nM dasatinib (Das) and 1µM quercetin (Que) for 48 h. Surviving cells were harvested, counted, and an equal amount of cells was reseeded. After 7 days of growth, cells were stained with crystal violet and quantified to assess confluency (n = 3; unpaired t-test, *p < 0.05). Dasatinib/quercetin-untreated cells (No pre-T) were used as a control Activated Ras signaling causes senescence induction, and OIS causes decreased growth in PTPA OE MPM cells. (A-B) ZL34 and NCI-H2452 control and PTPA-overexpressing cells were treated for 48 h with 15 µM and 25 µM Trametinib, respectively, and harvested for protein extraction. Immunoblot analysis of the lysates demonstrated decreased ERK phosphorylation (phospho-ERK/total ERK), alongside decreased p21 levels (senescent marker) in ZL34 (A) and NCI-H2452 cells (B) (n = 3 (ZL34) and n = 4 (NCI-H2452); two-way ANOVA, *p < 0.05, ***p < 0.001). (C) GFP control and PTPA OE ZL34 cells were pre-treated with 10nM dasatinib (Das) and 1µM quercetin (Que) for 48 h. Surviving cells were harvested, counted, and an equal amount of cells was reseeded. After 7 days of growth, cells were stained with crystal violet and quantified to assess confluency (n = 3; unpaired t-test, *p < 0.05). Dasatinib/quercetin-untreated cells (No pre-T) were used as a control The pharmacologic PP2A activator ATUX-792 does not phenocopy the effects of PTPA restoration in MPM cells Lately, there has been much excitement about the development of pharmacologic PP2A activators (SMAPs, Small Molecule Activators of PP2A) and their potential application in cancer treatment [65, 66]. To test whether the effects of PTPA restoration in PTPA-low MPM cells could be phenocopied by such a SMAP, we used the ATUX-792 compound [47] to evaluate its putative senescence-inducing potential in our cell models (Fig. 7A–D). Although ATUX-792 clearly exerted a therapeutic effect, decreasing cell viability in both control and PTPA OE cells in two independent MPM cell lines (Fig. 7C and D), immunoblot analysis of p21 expression demonstrated this did not correlate with senescence induction (Fig. 7A and B). In fact, ATUX-792 even reduced increased p21 levels in PTPA OE cells (Fig. 7A and B). This unexpected, seemingly senolytic effect of ATUX-792 might explain why PTPA OE NCI-H2452 cells appeared more sensitive to this compound than the control cells (Fig. 7C), although this was not well reproduced in the other MPM cell line, ZL34 (Fig. 7D). Overall, these data suggest that the therapeutic effect of ATUX-792 in MPM cells does not involve induction of senescence, but is likely caused by other growth-inhibiting or cell-killing mechanisms, one of which could be the promotion of senolytic death. Fig. 7Effects of the pharmacologic PP2A activator ATUX-792 on cell viability and senescence induction (p21 expression). (A-B) Control and PTPA OE MPM cells were treated with 8 µM ATUX-792 in NCI-H2452 cells (A) or 6 µM ATUX-792 in ZL34 cells (B) for 48 h and harvested for protein extraction. Immunoblot analysis of the lysates was performed for the senescence marker p21 (n = 3, two-way ANOVA, *p < 0.05, ***p < 0.001). (C-D) ATUX-792 dose-response curves in control (GFP) and PTPA re-expressing NCI-H2452 (C) and ZL34 cells (D). Cells were plated, treated and viability assessed by MTT at 72 h (n = 1, each datapoint is the average of three technical replicates). Curves were fitted by nonlinear regression, and EC50 values were calculated Effects of the pharmacologic PP2A activator ATUX-792 on cell viability and senescence induction (p21 expression). (A-B) Control and PTPA OE MPM cells were treated with 8 µM ATUX-792 in NCI-H2452 cells (A) or 6 µM ATUX-792 in ZL34 cells (B) for 48 h and harvested for protein extraction. Immunoblot analysis of the lysates was performed for the senescence marker p21 (n = 3, two-way ANOVA, *p < 0.05, ***p < 0.001). (C-D) ATUX-792 dose-response curves in control (GFP) and PTPA re-expressing NCI-H2452 (C) and ZL34 cells (D). Cells were plated, treated and viability assessed by MTT at 72 h (n = 1, each datapoint is the average of three technical replicates). Curves were fitted by nonlinear regression, and EC50 values were calculated PTPA overexpression sensitizes MPM cells to carboplatin As induction of senescence is often associated with chemosensitization as a result of lethal cell stress [63, 64], we next hypothesized that MPM cells with high vs. low PTPA expression might respond differently to anti-cancer drugs. This was tested for chemotherapeutics carboplatin and pemetrexed, which are both in clinical use for MPM treatment [1, 2]. While no PTPA-dependent effects were observed for pemetrexed treatment, PTPA re-expression sensitized MPM cells for carboplatin following short-term (72 h) treatment in two independent cell lines (Fig. 8A and B). Since the effect was rather modest in ZL34 cells (Fig. 8A), we further subjected these cells to a long-term treatment of 12 days and found a clearly reduced total colony area compared to GFP control cells at a dose of 50 µM (Fig. 8C). Notably, pretreatment with senolytic agents (combo dasatinib/quercetin) abolished the difference in carboplatin sensitivity between control and PTPA OE cells in this assay (Fig. 8C). Assessment of p21 expression further suggested that carboplatin exerts its therapeutic effect, at least in part, by inducing (DNA-damage-induced) senescence, as p21 levels were significantly further increased in both PTPA-high and PTPA-low cells (Fig. 8D). Since the difference in carboplatin sensitivity between control and PTPA OE cells was abolished by pre-removal of the senescent cells in the PTPA OE condition, we concluded that their difference in carboplatin sensitivity is likely to be explained by a lethal combination of PTPA-dependent oncogene-induced senescence and carboplatin-dependent DNA-damage-induced senescence. This is supported by flow cytometry-based cell cycle analysis following 12 days of treatment, showing a trend toward increased cell death in PTPA-overexpressing ZL34 cells (Fig. 8E, Suppl. Fig. S14). Increased carboplatin sensitivity was further accompanied by trends for increased γH2AX (a marker of DNA damage) and increased ATR phosphorylation following 12 days of treatment (Fig. 8F). Overall, these results underscore the clinical survival data (Fig. 1C), hinting towards heterozygous loss of PTPA as a potential predictive marker for adverse carboplatin outcome, and hence, associated with a worse prognosis. Fig. 8PTPA re-expression leads to increased sensitivity to carboplatin treatment by a lethal combination of oncogene-induced senescence and carboplatin-related DNA-damage-induced senescence. (A) Representative dose-response curves of 72h-carboplatin treatment in control (GFP) and PTPA re-expressing (PTPA OE) ZL34 cells. EC50 values were determined for three independent assays (paired t-test, *p < 0.05) (B) Representative dose-response curves of 72 h-carboplatin treatment in control (GFP) and PTPA re-expressing (PTPA OE) NCI-H2452 cells (n = 3; EC50 value could not be determined for GFP cells). (C) Crystal violet staining of GFP control and PTPA-overexpressing ZL34 cells, following treatment with the indicated concentrations of carboplatin (25 µM, 50 µM, 75 µM) for 12 days. In one condition, cells were pre-treated with 10 nM dasatinib / 1 µM quercetin for 48 hours. Colony area was quantified using Fiji software, and statistical significance was assessed by two-way ANOVA (n = 3; *p < 0.05, **p < 0.01). (D) p21 immunoblot analysis of control (GFP) and PTPA OE ZL34 cells following long treatment with 50 µM carboplatin. Band intensities were quantified and related to those of vinculin, which was used as a loading control (n = 3, two-way ANOVA). (E) Percentage of cells in the indicated phase of the cell cycle after long-term (12 days) treatment with 50 µM carboplatin, as determined by flow cytometry. (F) Immunoblot analysis of the DNA damage marker γH2AX, total ATR, and phospho-ATR (Ser428) in ZL34 control and PTPA-overexpressing cells following carboplatin (50 µM) treatment. Quantification of signal intensity is shown relative to control cells (GFP). All signals were normalized for the vinculin signal determined on the same blot. Statistical significance was determined using a two-way ANOVA test (n = 2-3; *p < 0.05) PTPA re-expression leads to increased sensitivity to carboplatin treatment by a lethal combination of oncogene-induced senescence and carboplatin-related DNA-damage-induced senescence. (A) Representative dose-response curves of 72h-carboplatin treatment in control (GFP) and PTPA re-expressing (PTPA OE) ZL34 cells. EC50 values were determined for three independent assays (paired t-test, *p < 0.05) (B) Representative dose-response curves of 72 h-carboplatin treatment in control (GFP) and PTPA re-expressing (PTPA OE) NCI-H2452 cells (n = 3; EC50 value could not be determined for GFP cells). (C) Crystal violet staining of GFP control and PTPA-overexpressing ZL34 cells, following treatment with the indicated concentrations of carboplatin (25 µM, 50 µM, 75 µM) for 12 days. In one condition, cells were pre-treated with 10 nM dasatinib / 1 µM quercetin for 48 hours. Colony area was quantified using Fiji software, and statistical significance was assessed by two-way ANOVA (n = 3; *p < 0.05, **p < 0.01). (D) p21 immunoblot analysis of control (GFP) and PTPA OE ZL34 cells following long treatment with 50 µM carboplatin. Band intensities were quantified and related to those of vinculin, which was used as a loading control (n = 3, two-way ANOVA). (E) Percentage of cells in the indicated phase of the cell cycle after long-term (12 days) treatment with 50 µM carboplatin, as determined by flow cytometry. (F) Immunoblot analysis of the DNA damage marker γH2AX, total ATR, and phospho-ATR (Ser428) in ZL34 control and PTPA-overexpressing cells following carboplatin (50 µM) treatment. Quantification of signal intensity is shown relative to control cells (GFP). All signals were normalized for the vinculin signal determined on the same blot. Statistical significance was determined using a two-way ANOVA test (n = 2-3; *p < 0.05) Discussion In the present study, we investigated the potential role of PP2A alterations in MPM, a highly aggressive cancer with limited therapeutic options [2]. Analysis of the cBioPortal dataset revealed that 26% of MPM patients present with heterozygous loss of PTPA (PPP2R4), correlating with reduced overall survival. Uni- and multivariate analysis taking into account clinical parameters such as tumor histology, stage, age at diagnosis and irradiation treatment confirmed the prognostic role of PTPA hetloss in this patient cohort, although the low number of patients (n = 82) remains a limitation of our study. Decreased PTPA expression was further confirmed in seven arbitrarily chosen human MPM cell lines, underscoring it is a recurrent event in MPM. Re-expression of PTPA in MPM cells decreased cancer cell-specific growth, which was causally linked with increased oncogene-induced senescence and activated RAS/ERK signaling. Moreover, PTPA re-expression enhanced sensitivity to carboplatin, likely due to a lethal combination of oncogene- and therapy-induced cell stress. As PTPA is one of the major cellular regulators of PP2A, promoting C subunit activation, methylation and holoenzyme assembly with most regulatory B-type subunits [36], PTPA dysfunction, together with the overexpression of endogenous PP2A inhibitors [18], constitutes one of the clinically important mechanisms leading to PP2A inactivation in cancer - a phenomenon commonly found in carcinomas, brain tumors and hematologic cancers, and here now extended to mesothelioma. Because most PP2A holoenzymes counteract specific oncogenic signaling pathways [18], decreased PP2A activity is expected to promote tumor initiation and/or progression [67]. Our group has previously demonstrated in vivo evidence that heterozygous loss of PTPA promotes spontaneous tumor formation in mice, predominantly leading to hematologic malignancies and hepatocellular carcinoma [31]. This phenotype correlated with increased activation of several established PP2A-B56 substrates known to act as oncogenes when hyperphosphorylated. Furthermore, PTPA deficiency specifically impaired the catalytic activity of B56-containing PP2A holoenzymes, while B55-containing complexes remained unaffected [31]. In a mouse model of mutant Kras-induced lung adenocarcinoma, reduced PTPA expression led to elevated c-MYC levels, an established RAS pathway target, thereby also promoting tumor growth [38]. In the present study, we took a completely different approach by examining the impact of PTPA re-expression on both signaling and phenotypic characteristics of PTPA-low MPM cells. Phenotypically, we found a reduction in 2D and 3D colony formation, consistent with previous data showing that PTPA depletion in transformed (A549, A375) or immortalized epithelial cells expressing an activated oncogene (HEK-TER) had the opposite effect [26, 31, 38]. In contrast, near-depletion of PTPA in normal mesothelial cells (MeT-5A), i.e. in the absence of any oncogenic activation, resulted in a strong growth arrest, likely reflecting the essential nature of PTPA for cell survival [31, 52, 53]. At the enzyme level, PTPA-overexpressing MPM cells displayed elevated levels of the PP2A catalytic C subunit with increased C methylation, while the overall stoichiometry of PP2A C-containing holoenzymes that could be detected in the PP2A-C nanobody IPs by our MS approach (i.e. B55α/δ, all B56 isoforms and PABIR-containing complexes), remained unchanged, suggestive of an overall activation of ‘PP2A’ without favoring any specific holoenzymes. This contrasts with the PP2A-B56-specific effect observed upon partial PTPA depletion [31], which would likely only be extended to PP2A-B55 complexes upon complete PTPA knockout (as explained, this is, however, lethal [31]). RNAseq analysis underscored that PTPA re-expression in MPM cells resulted in activation of the Ras signaling pathway, an observation that is presumably attributable to increased dephosphorylation of KSR1 at Ser392 (Mus musculus numbering), further leading to MEK and ERK activation. Such a mechanism aligns with previous studies showing that PP2A not only regulates KSR1 activity by dephosphorylating the 14-3-3 binding site, but also controls its subcellular localization [57], and both PP2A-B55 and PP2A-B56 complexes have been shown to mediate direct KSR1 Ser392 dephosphorylation depending on the cell context [55, 57]. Hyperactivated RAS/ERK signaling in PTPA-reconstituted MPM cells correlated with increased cellular senescence - underscored by enrichment of the SenMayo gene set, increased SA-β-Gal activity and increased p21 expression. Since the latter was completely rescued by the MEK inhibitor Trametinib, we concluded that activated RAS/ERK signaling in fact caused the cellular senescence. OIS was further underscored by induction of the Senescence-Associated Secretory Phenotype (SASP) and activation of NF-κB signaling. The latter is likely part of the induction of the senescence program [63, 64], rather than a direct consequence of PP2A activation, since PP2A is known to regulate NF-κB signaling by dephosphorylating IKKβ, IκBα and RelA/p65, exerting an inhibitory effect on this pathway [68]. Nevertheless, as the senolytic agents dasatinib/quercetin restored decreased growth of PTPA overexpressing cells, a strong causal relationship was established between senescence induction and the observed growth decrease of PTPA-high MPM cells. Although the tumor suppressive role of PP2A is well-documented [18, 19], far less studies have pointed toward its role in regulating oncogene-induced senescence (OIS), stress-induced premature senescence (SIPS, e.g. by oxidative stress, DNA-damage) or senescence upon aging in general (replicative senescence) [69, 70]. Interesting to mention in this respect is a study reporting inhibition of Ras-induced senescence by SV40 ST antigen, essentially through an identical mechanism (inhibition of KSR1 Ser392 dephosphorylation) as the one we found here [71]. As explained, SV40 ST is considered a human carcinogen, and its oncogenic activity is strongly linked to PP2A inhibition and the development of cancers such as mesothelioma. By inhibiting PP2A, ST effectively prevents the Ras-induced senescence response, a phenomenon that may be mimicked in MPM by heterozygous loss of PTPA. Likewise, CIP2A depletion has repeatedly been shown to induce senescence in cancer cells, in vitro and in vivo [72–74]. Remarkably, in MPM, our cBioPortal analysis and that of others [34] demonstrated increased CIP2A expression in up to 24% of patients, correlating with decreased OS, suggesting that here too, inhibition of OIS may be involved in the mechanisms underlying CIP2A oncogenicity. The long latency of MPM development from the time of exposure to asbestos suggests that multiple genetic alterations are required for tumorigenic conversion of mesothelial cells. Although KRAS and TP53 mutations have been suggested to not play a critical role in MPM induction either in humans or in rats [9, 75], a recent study has questioned this for KRAS [9], and activated Ras signaling in MPM may additionally be the indirect result of much more common loss-of-function mutations in NF2 [7]. Thus, inhibition of Ras-induced senescence by PP2A dysfunctions, such as heterozygous loss of PPP2R4 (PTPA), overexpression of CIP2A or expression of SV40 ST, may actually play a major, but not previously recognized, clinically-relevant role in MPM development. Besides induction of senescence, our RNAseq analysis also identified activation of p53 signaling in PTPA-reconstituted MPM cells, further confirmed by increased expression of p21. Activation of p53 in response to activated Ras signaling is one of the best understood pathways involved in OIS, and may involve activation of the DNA Damage Response (DDR) as well as DDR-independent mechanisms [59, 60]. In our model, γH2AX and phospho-ATR analysis in PTPA overexpressing cells demonstrated an obvious increase in DNA-damage or activation of the DDR under basal conditions, suggestive of a DDR-dependent p53 activation mechanism. Alternatively, PP2A activation upon PTPA overexpression may also directly contribute to p53 activation, as both PP2A-B55 and PP2A-B56 complexes have been shown to positively regulate p53, either through direct dephosphorylation of p53 or through dephosphorylation and destabilization of Mdm2 [76–78]. Further, we found p21 induction in both wild-type (ZL34, NCI-H226) as well as mutant TP53 MPM cell lines (NCI-H2452) upon PTPA re-expression, suggesting that increased p21 expression could also occur independently from TP53. Likewise, neither BAP1 nor NF2 status influenced PTPA-dependent senescence induction in our models, while it could be relevant that all our cell models did not express CDKN2A (p16) (Table 1). Another interesting finding of this study pertains to the mechanism-of-action of ATUX-792, an alleged PP2A activating compound based on an enzymatic assay [47]. Although ATUX-792 clearly exerted a growth inhibiting and cytotoxic effect in MPM cells, it did not phenocopy the senescence-inducing effect of PTPA restoration. On the contrary, in fact, it reduced p21 expression in PTPA-high MPM cells, consistent with senescence inhibition. In a model of lung emphysema, ATUX-792 reduced ERK phosphorylation/activation [79], which might mechanistically explain how this apparent senolytic effect is achieved (cfr. similar effect of MEK inhibitor Trametinib). By all means, these data demonstrate caution is warranted about the therapeutic use of RAS/ERK pathway inhibitors to inhibit cell growth in MPM. On the other hand, in an in vivo setting, senescence may not always be beneficial, as senescent cells can promote tumor phenotypes via their SASP and effects on the tumor microenvironment, as well as by presenting a pool of dormant, oftentimes therapy-resistant tumor cells that may resume proliferation or further progress and metastasize under the right conditions [80]. Hence, it would certainly be worthwhile to follow-up on the effects of PP2A activators such as ATUX-792, and their future use as potential senolytic agents in cancer therapy. In the final part of our work, we found that PTPA re-expression in ZL34 cells increases cellular sensitivity to carboplatin, a drug used to treat many MPM patients, despite the frequent occurrence of primary or secondary resistance. We executed typical 72 h-dose-response curves, but also mimicked chronic exposure of the cancer cells to the drug (as in patients), by longer-term treatment for 12 days. In both cases, we observed a strong induction of p21 expression, elevated DNA damage and increased ATR phosphorylation in both control and PTPA overexpressing cells. As pretreatment with senolytic agents abolished cell sensitization to carboplatin, the enhanced therapeutic response of PTPA-high MPM cells can thus likely be explained by induction of lethal cell stress (i.e. a combination of therapy-induced DNA-damage in already oncogene-induced replicatively stressed cells). In accordance with such a predictive marker potential for PTPA, previous studies reported increased HeLa cell resistance to diverse chemotherapeutics, including cisplatin, etoposide and taxol, upon siRNA-mediated knockdown of PTPA [81], while in KRAS-mutant lung adenocarcinoma cells, PTPA knockdown induced resistance to MEK inhibitors, while sensitizing for mTOR inhibitors [38]. In conclusion, our study provides new insights into the role of PTPA in MPM, revealing a novel mechanism whereby PTPA is identified as an essential contributor to Ras-induced senescence, a physiologic anti-proliferative program that limits propagation of cells at risk for transformation following oncogenic activation. Combined with an additional therapy-induced stress such as DNA-damage, PTPA expression promotes an increased therapeutic response, as seen upon carboplatin treatment in PTPA-high versus PTPA-low MPM cells. Therefore, heterozygous loss of PTPA may not only contribute to asbestos-related MPM development by circumventing OIS, but also to a decreased therapeutic response to carboplatin, coinciding with decreased overall survival of patients. Although these are important results, in vivo validation of our findings remains essential, especially given the double-edged role of senescence in cancer biology and treatment. This will be a challenging task, given many MPM cell lines exhibit poor subcutaneous xenograft growth in mice, and few reliable orthotopic in vivo MPM models have currently been established [82]. Finally, our work suggests the potential therapeutic utility of PP2A-activating drugs that could mimic the effects of PTPA overexpression in MPM, either by the induction of senescence or when combined with carboplatin or other potent DNA-damaging treatments, by the induction of lethal cell stress. Given our finding that ATUX-792 appears not to be such a PTPA-mimicking compound, this only pleads for more and profound studies into the precise mechanisms-of-action of the still growing list of currently described PP2A activating drugs. Electronic Supplementary Material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2