RAS and PP2A activities converge on epigenetic gene regulation

RAS-mediated human cell transformation requires inhibition of the tumor suppressor Protein Phosphatase 2A (PP2A). Both RAS and PP2A mediate their effects by phosphoregulation, but phosphoprotein targets and cellular processes in which RAS and PP2A activities converge in human cancers have not been systematically analyzed. Here, based on mass spectrometry phosphoproteome data we discover that phosphosites co-regulated by RAS and PP2A are enriched on proteins involved in epigenetic gene regulation. As examples, RAS and PP2A co-regulate the same phosphorylation sites on HDAC1/2, KDM1A, MTA1/2, RNF168 and TP53BP1. Mechanistically, we validate co-regulation of NuRD chromatin repressor complex by RAS and PP2A. Consistent with their known synergistic effects in cancer, RAS activation and PP2A inhibition resulted in epigenetic reporter de-repression and activation of oncogenic transcription. Notably, transcriptional de-repression by PP2A inhibition was associated with increased euchromatin and decrease in global DNA methylation. Further, targeting of RAS- and PP2A-regulated epigenetic proteins decreased viability of KRAS-mutant human lung cancer cells. Collectively the results indicate that epigenetic protein complexes involved in oncogenic gene expression constitute a significant point of convergence for RAS hyperactivity and PP2A inhibition in cancer. Further, the results provide a rich source for future understanding of phosphorylation as a previously unappreciated layer of regulation of epigenetic gene regulation in cancer, and in other RAS/PP2A-regulated cellular processes.

worldwide (Prior, Hood et al., 2020). RAS-mediated human cell transformation is 66 preceded by cell immortalization which is caused by loss of tumor suppressors such 67 as TP53, RB1 or CDKN2 (Minna, Roth et al., 2002). Hyperactivation of RAS 68 signaling is however not alone sufficient for malignant transformation of immortalized 69 human cells, but requires simultaneous inhibition of the phosphatase activity of the 70 tumor suppressor Protein Phosphatase 2A (PP2A) (Hahn, Dessain et al., 2002, 71 Rangarajan, Hong et al., 2004, Sablina, Hector et al., 2010, Sato, Larsen et al., 72 2013, Tian, Doerig et al., 2018, Yu, Boyapati et al., 2001. PP2A inhibition and RAS 73 mutations also significantly synergize in predicting poor overall survival of cancer 74 patients across the TCGA pan-cancer data (Kauko, Laajala et al., 2015). Moreover, 75 reactivation of the tumor suppressor activity of PP2A efficiently inhibits RAS-driven 76 tumorigenesis (Liu, Gu et al., 2015, Saddoughi, Gencer et al., 2013, Sangodkar, Perl 77 et al., 2017, and synergize with pharmaceutical targeting of RAS downstream 78 effector MEK (Kauko, O'Connor et al., 2018). Thus, understanding the mechanistic 79 basis of the synergism between PP2A inhibition and RAS activity could provide 80 significant novel opportunities for targeting RAS-dependent cancers. 81 82 PP2A comprises a family of trimeric protein complexes that counter-balance kinase-83 mediated phosphorylation throughout cell signaling networks (Fowle, Zhao et al., 84 2019). The PP2A trimers are composed of a scaffolding PP2A-A subunit, a catalytic 85 C subunit, and one of the alternative substrate determining B-subunits. In about 10% 86 of human cancers PP2A is inhibited by genomic mutations, but the most prevalent 87 mechanism for PP2A inhibition in cancer is overexpression of one of the numerous 88 oncogenic PP2A inhibitor proteins such as CIP2A, PME1 or SET (Kauko & 89 Westermarck, 2018) (Fig. 1A). Several downstream effectors/kinases of RAS are 90 identified as PP2A targets, but it is not yet fully elucidated what the cancer-relevant 91 cellular processes in which RAS and PP2A activities converge. Indeed, PP2A has 92 been shown to regulate for example RAF-MEK-ERK and PI3K-AKT pathways (Fowle 93 et al., 2019, Kauko & Westermarck, 2018, Sablina et al., 2010, but whether there 94 are processes beyond kinase signaling that are relevant to RAS/PP2A co-operation 95 in cancer is poorly understood. Interestingly, a recent phosphoproteome analysis 96 revealed that PP2A inhibition and RAS activity regulate highly overlapping 97 phosphoproteins (Kauko et al., 2015). However, the functional relevance of these 98 findings has not been studied yet. 99 100 Gene expression in multicellular organisms is regulated through various epigenetic 101 processes involving for example insertion or removal of chemical tags on the 102 nucleotides and histones (Miller & Grant, 2013). Importantly, both epigenetic gene 103 silencing of tumor suppressors, as well as increased transcription of oncogenic 104 genes contribute to cancer initiation, progression, and therapy resistance (Baylin & 105 Jones, 2016, Quagliano, Gopalakrishnapillai et al., 2020. DNA methylation is the 106 best characterized epigenetic mechanism mediating gene repression. On the other 107 hand, mechanisms that impact nucleosomes via covalent histone modifications 108 leading to open chromatin state (euchromatin) are well established for oncogenic 109 transcription. Strategies to impact epigenetic gene regulation might therefore be 110 Here, we have addressed the open question of convergence of RAS-and PP2A-135 mediated phosphoregulation in cancer by using previously published 136 phosphoproteome datasets in which RAS proteins and PP2A complexes were 137 targeted by siRNAs (Kauko, Imanishi et al., 2020, Kauko et al., 2015 (Fig. 1A). The 138 results demonstrate that epigenetic gene regulation is particularly enriched among 139 the cellular processes co-regulated by  This is due to numerous, but previously unidentified, RAS-and PP2A-regulated  150  151  152  153  154  155  156  157  158  159  160  161  162  163  164  165  166  167  168  169  RESULTS  170   171   Systematic analysis of phosphosites co-regulated by RAS-and PP2A  172   173 To comprehensively map the phosphoproteins co-regulated by PP2A and RAS, we 174 combined data from two recent phosphoproteome studies, in which either all three 175 forms of RAS (HRAS, KRAS and NRAS) (Kauko et al., 2015), or the PP2A scaffold 176 protein PP2A-A, or the PP2A inhibitor proteins CIP2A, PME1, and SET (Kauko et al., 177 2020), were depleted by siRNAs (Fig. 1A, Table S1). For RAS and the PP2A 178 inhibitory proteins we included the phosphosites that were dephosphorylated in the 179 siRNA transfected cells, whereas for PP2A-A we included the phosphosites that had  (Table S2). As a clear indication for 185 convergence of RAS and PP2A activities on phosphoproteome regulation, altogether 186 270 distinct phosphorylation sites on 237 proteins were found to be co-regulated 187 both by RAS and PP2A targeting (Fig. 1B). Interestingly, when assessing the overlap 188 of the regulated phosphosites between RAS and PP2A modulations, sites 189 dephosphorylated by RAS inhibition overlapped more frequently with sites 190 dephosphorylated by PP2A inhibitor protein inhibition, than with sites regulated by 191 PP2A inhibition (Fig. S1). This can be explained by the notion that both RAS 192 inhibition and PP2A reactivation inhibit phosphorylation of sites that are constitutively 193 phosphorylated in cancer cells, and based on the recent model that the majority of 194 cellular phosphosites are exclusively dominated by either phosphatase activation or 195 inhibition (Kauko et al., 2020). The overlap between RAS and PP2A inhibitor protein 196 SET was particularly notable (Fig. S1A,B). This could explain very potent antitumor 197 effects of SET inhibition in RAS-driven tumorigenesis (Liu et al., 2015, Saddoughi et 198 al., 2013. In addition to 237 proteins in which at least one phosphorylation site was 199 co-regulated by both RAS and PP2A (Fig. 1B), RAS and PP2A co-regulated 200 phosphorylation of 57 overlapping proteins but in these proteins the RAS and PP2A 201 regulated sites were not identical (Table S2). Collectively these analyses 202 demonstrate a clear convergence of RAS and PP2A-mediated phosphoregulation, 203 both at the level of individual phosphorylation sites, but also at the level of proteins. 204

205
To identify cellular processes that would be governed by the convergence of RAS-206 and PP2A-mediated phosphoregulation, we analyzed enriched gene ontologies (GO) 207 based on the 237 proteins in which there was at least one phosphosite regulated by 208 both RAS and PP2A (Fig. 1B). The STRING database (Szklarczyk,Gable et al.,209 2021) analysis revealed a clear enrichment of GOs related to epigenetic and 210 transcriptional gene regulation (Fig. 1B, GO terms 0070087, 0003916, and 211 0070577). To increase the resolution of the analysis, we performed an analysis in 212 which the shared RAS/PP2A phosphotargets were divided into four clusters (Fig.  213 1B). Whereas cluster 1 was mostly associated with cytoskeleton and cell adhesion, 214 and cluster 2 with nucleic acid binding, clusters 3, and especially 4, revealed a very 215 strong association of the target proteins with histone modifications and chromatin 216 remodeling (Fig. 1B). While recent data validate the critical role for PP2A in 217 regulating transcriptional elongation (Huang, Jee et al., 2020, Vervoort, Welsh et al., 218 2021, and epigenetic gene regulation has important role in RAS-mediated 219 oncogenesis (Vaz, Hwang et al., 2017), the role of PP2A and RAS in 220 phosphorylation-dependent regulation of epigenetic complexes is very poorly -221 understood. Based on these notions, we focused our downstream analysis on RAS 222 and PP2A convergence on epigenetic gene regulation and transcription.  Interestingly, many epigenetic RAS/PP2A phosphotargets were found to constitute 234 protein complexes with each other (Fig. 2A). The most apparent examples were the 235 NuRD, DNMT1 and DOT1L complexes ( Fig. 2A). We postulate that RAS/PP2A 236 signaling can potentially regulate DNA methylation, histone methylation, and histone 237 deacetylation via phosphorylation of these complexes (Trevino et al., 2015) (Fig. 2A). 238 Consistent with the convergence model, most protein members of these epigenetic 239 complexes were regulated by both RAS and PP2A ( Fig. 2A), both at the level of 240 individual phosphosites, but also at the level of proteins (Fig. 2B, S2). Naturally some 241 epigenetic factors were also found to be regulated by either RAS or PP2A only (Fig.  242 2B, S2). In the case of PP2A-regulated targets, the evidence for direct PP2A-243 mediated dephosphorylation was strengthened by the identification of putative 244 binding motifs for PP2A B-subunit B56 in many of these proteins (Table S3). Based 245 on previous evidence, that B56 binding motif containing proteins can act as scaffolds 246 for the recruitment of the other complex proteins for PP2A-mediated 247 dephosphorylation (Hertz, Kruse et al., 2016), the majority of the individual PP2A 248 target proteins from our data could become accessible for PP2A-mediated provided clues about their potential functional importance. Using the RNF168 253 paralog, RNF169-Histone 2 complex, as the model structure (PDB), the RAS/PP2A-254 regulated RNF168 serine 481 was found adjacent to arginine 466 that is critical for 255 histone binding (Fig. 2C). On the other hand, PP2A-regulated serine 714 on DNMT1 256 is located at the base of the loop that must relocate for DNA binding (Fig. 2C). 257

258
We further characterized the oncogenic potential of selected PP2A/RAS regulated 259 epigenetic target proteins. To this end, we targeted the selected proteins with three 260 siRNAs per gene, in three KRAS mutant non small cell lung cancer (NSCLC) cell 261 lines, A549, H358 and H460 (Fig. S3A). In parallel, H358 cells were treated with 262 chemical inhibitors for DNMT1, bromodomain proteins, and HDAC1/2 (Fig. S3B). 263 Collectively the results show a significant role for most of the epigenetic RAS/PP2A 264 target proteins for NSCLC cell viability.   Based on the phosphoproteome data ( Fig. 2), we proceeded to validate the impact of 294 PP2A and RAS on selected epigenetic factors, and especially on chromatin 295 recruitment of the NuRD complex as it showed very high level of PP2A/RAS-296 mediated phosphorylation regulation ( Fig. 2A). 297 298 CHD3 is a member of the NuRD complex, and it supports viability of KRAS-mutant 299 lung cancer cell lines (Fig. S3). Based on Phosphosite Plus database 300 (www.phosphosite.org) CHD3 is phosphorylated on 54 distinct serines or threonines 301 but there are no reports about the functional relevance of any of these phosphosites. 302 In our data, RAS inhibition resulted in dephosphorylation of S713 ( Fig. 3A and S2), 303 whereas PP2A reactivation by SET inhibition caused dephosphorylation of S1601 304 and S1605 (Fig. S2, Table S1). Using yeast CHD1 crystal structure (PDB 3MWY) as 305 a model, the RAS target site S713 was located on the unstructured region in the 306 vicinity of the nucleotide binding mediating region between amino acids 761-768 in 307 human CHD3 protein (Fig. 3A). This structural organization was supported by 308 AlphaFold analysis of human CHD3 (Fig. S4A). S713 phosphorylation in KRAS 309 mutant A549 lung cancer cells seems to be critical for the stability of CHD3 protein, 310 as S713A mutation dramatically inhibited CHD3 protein expression, whereas the 311 phosphorylation mimicking mutation S173D resulted in increased protein expression 312 To test the impact of RAS/PP2A activities on SFRP1 promoter activity, we used the 379 same siRNA treatments as were used for generating the phosphoproteome data 380 (Kauko et al., 2020, Kauko et al., 2015. Notably, inhibition of PP2A either by 381 siPP2A-A, or by chemical serine/threonine phosphatase inhibitor Okadaic acid, 382 resulted in increased promoter activity (Fig. 4B, C). Related to the endogenous 383 PP2A inhibitory mechanisms, PME1 depletion resulted in further repression of 384 SFRP1-GFP reporter activity, while SET depletion did not have an effect (Fig. 4B, 385 C). The role for PME1-mediated PP2A inhibition in promoting oncogenic transcription 386 was further supported by the increased reporter activity upon transient PME-1 387 overexpression (Fig. 4D, E). On the other hand, RAS inhibition resulted in significant 388 decrease in SFRP1-GFP reporter activity ( Fig. F and G). Downstream of RAS, the 389 effect on reporter activity appears to be at least partly mediated by MEK-ERK MAPK 390 pathway, as MEK inhibitor AZD6244 treatment also resulted in significant reporter 391 activity inhibition (Fig. 4E). 392

393
These data indicate that like their opposing roles in oncogenesis (Rangarajan et al., 394 2004, Sangodkar et al., 2017, Sato et al., 2013, Zhou, Updegraff et al., 2017  Other significant GO terms indicated the role for PP2A in regulation of cellular 459 adhesion, migration and motility, all highly relevant for malignant cancers (Fig. S6). 460 GSEA of genes downregulated by RAS inhibition, revealed an overlap with PP2A-461 regulated GSEA signatures, such as inflammatory response, and mitotic spindle 462 (Fig. 5C, D). In addition, RAS was found to regulate genes related to IL6-JAK-STAT 463 signaling and G2M checkpoint (Fig. 5D). On the other hand, the most significantly 464 enriched GO biological process upon RAS silencing was "negative regulation of cell 465 proliferation activity" (Fig. S6). PP2A and RAS targeting (Fig. 5F, in bold). This is consistent with recent results that 472 PP2A complexes containing Striatins stimulate YAP1 activity leading to cellular 473 transformation (Kurppa & Westermarck, 2020). On the other hand, YAP1 drives 474 resistance to KRAS and EGFR inhibition (Kurppa, Liu et al., 2020, Shao, Xue et al., 475 2014. Further, KRAS and YAP1 synergistically activate FOS transcription factors, 476 leading to EMT (Shao et al., 2014). In addition, AR target genes were also enriched 477 among PP2A and RAS regulated targets (Fig. 5F, in bold). This is consistent with 478 previous results indicating role for both RAS activity and PP2A inhibition in promoting 479 malignant growth of AR-positive prostate cancers (Khanna, Rane et al., 2015, Weber 480 & Gioeli, 2004. Since both the PP2A-regulated phosphoproteome targets ( Fig. 2A), and increased 499 activity of the methylation-sensitive reporter assay (Fig. 4), indicated that PP2A could 500 regulate DNA methylation, we analyzed global DNA methylation in PP2A inhibited 501 cells by Reduced Representation Bisulfite Sequencing (RRBS). Consistent with the 502 transcriptional de-repression ( Fig. 4 and 5), the siRNA mediated inhibition of PP2A-A 503 predominantly resulted in DNA demethylation (Fig. 6A). Of the total 211 differentially 504 methylated regions, 143 regions showed a decrease in methylation marks 505 (hypomethylated), while 68 showed an increase in methylation (hypermethylated) 506 (Fig. 6B). The occupancy of the differentially regulated methylation marks was lowest 507 at the exons (7%), while almost symmetrically distributed between introns (36%), 508 intergenic areas (30%), and promoter regions (27%) (Fig. 6C). Inhibition of PP2A 509 had an overall maximum impact on methylation of chromosome 11 (Fig. 6D). On the 510 other hand, when the ratio between hypomethylation and hypermethylation was 511 considered, the highest degree of hypomethylation was seen in X chromosome, 512 which was exclusively hypomethylated, followed by the chromosome 13 in which six 513 regions were hypomethylated and only one region was hypermethylated in response 514 to PP2A inhibition (Fig. 6D). 515

516
To interrogate PP2A function in oncogenic transcription via its impact on DNA 517 methylation, we performed a GO enrichment analysis based on differentially 518 methylated regions (DMRs), and by using the Enrichr analysis tool (Kuleshov, Jones 519 et al., 2016) (Fig. S7). Importantly, overlapping with the GSEA analysis of PP2A-520 regulated gene expression (Fig. 5C), both epithelial to mesenchymal transition, and 521 several gene sets related to membrane associated GTPase activity (corresponding 522 to KRAS activity in GSEA), were enriched in PP2A inhibited cells (Fig. S7).

589
Epigenetic gene regulation has an established role in cancer initiation and 590 progression (Baylin & Jones, 2016, Cheng et al., 2019, Laugesen & Helin, 2014, 591 Quagliano et al., 2020. Several loss-of-function studies across different species 592 have also demonstrated an integral role for epigenetic gene regulation in signal 593 transduction, development, and malignant progression downstream of RAS proteins 594 (Vaz et al., 2017). However, it has been surprisingly poorly known how RAS impacts 595 phosphorylation of epigenetic proteins, and whether RAS activity towards epigenetic 596 About 20 years ago, inhibition of serine/threonine phosphatase activity of PP2A was 607 established by several studies as a prerequisite for RAS-mediated malignant 608 transformation of human, but not of mouse, cells (Hahn et al., 2002, Rangarajan et 609 al., 2004, Yu et al., 2001. Mechanistically the requirement of PP2A inhibition for 610 RAS-mediated human cell transformation was attributed to the role of PP2A as an 611 inhibitor of the activities of several downstream mediators of RAS activity such as 612 MEK/ERK and AKT kinases, or transcription factor MYC (Sablina et al., 2010, Yeh, 613 Cunningham et al., 2004. In this prevailing model, PP2A inhibition is seen merely as 614 a mechanism to boost signal transduction initiated by RAS. However, it has not been 615 previously systematically addressed whether their activities would converge on 616 particular cellular mechanisms or processes. This is a critical unanswered question, 617 as understanding why RAS activity and PP2A inhibition are mutually required for 618 human cell transformation could lead to fundamental novel understanding of the 619 basis of human cancer development and facilitate novel therapeutic approaches that 620 target the roots of human malignancies. In this context, PP2A inhibition has been 621 demonstrated to drive resistance of KRAS mutant cells towards a wide array of 622 kinase inhibitors (Kauko et al., 2018). Further, we and others have shown that PP2A 623 inhibition drives cancer cell resistance to epigenetic therapies ( Fig. S5D-F) (Kauko et 624 al., 2020, Shu, Lin et al., 2016. Thereby the presented results may provide 625 important clues for understanding the role of PP2A-mediated phosphoregulation on 626 responses of RAS-driven cancers to epigenetic therapies that has thus far been 627 clinically disappointing. On the other hand, the results encourage testing of the 628 emerging PP2A reactivating therapies (Vainonen, Momeny et al., 2021) for their 629 impact on KRAS-mutant cancers in combination with epigenetic therapies. 630

631
Our data provides strong evidence that PP2A inhibition drives oncogenic 632 transcription and globally impacts both DNA methylation and chromatin accessibility. 633 While the role of PP2A inhibition in oncogenic transcription is fully consistent with its 634 role in the CDK9-mediated (RNAPII-driven) transcriptional elongation (Huang et al., 635 2020, Vervoort et al., 2021, the role of PP2A in DNA methylation and chromatin 636 accessibility has been largely uncharacterized. While DNA methylation of the 637 promoters is the most studied epigenetic mechanism regulating gene expression, 638 recent reports indicate that cancer cells harbor hypomethylated regions at the 639 intergenic regions (Lee & Wiemels, 2016). Thereby the observed hypomethylation at 640 intergenic regions by PP2A inhibition indicate novel possible regulatory mechanisms 641 for cancer progression. On the other hand, our results demonstrating global 642 chromatin opening by PP2A inhibition are consistent with recent results from mouse 643 T cells in which deletion of regulatory B-subunit of PP2A (PPP2R2D) resulted in 644 chromatin opening (Pan, Sharabi et al., 2020). 645

646
Our results reveal dozens of RAS-and PP2A-regulated phosphorylation sites in 647 epigenetic proteins previously implicated in transcription and cancer (Fig. 2, and S2). 648 Structurally we provide evidence that at least some of these phosphosites are 649 located on functionally important regions of the epigenetic proteins involved in 650 oncogenic transcription (Fig. 2C). Mechanistically we validate the impact of RAS and 651 PP2A on selected NuRD complex members. While the mechanism by which RAS 652 and PP2A regulate CHD3 protein stability remains speculative, loss of CHD3 653 expression upon RAS inhibition (Fig. 3B) most likely contributes to the observed 654 chromatin remodeling phenotype in RAS/PP2A-modulated cells (Fig. 7). On the 655 other hand, previous studies have shown that HDAC2 phosphorylation is required for 656 its interactions with epigenetic multiprotein complexes such as Sin3, NuRD or 657 CoREST (Delcuve, Khan et al., 2012). In our data both RAS and PP2A regulate C-658 terminal HDAC1 and HDAC2 phosphorylation on largely overlapping sites (Fig. 2). 659 Functionally we validate that RAS and PP2A modulation regulate chromatin binding 660 of HDAC1/2 and that this correlates with transcriptional activity of highly HDAC 661 responsive SFRP1 promoter system. Naturally both phenotypes can also be 662 attributed to RAS/PP2A-mediated phosphorylation regulation of other NuRD complex 663 components such as MTA2. Additionally, we cannot exclude that RAS and PP2A 664 regulates gene expression and chromatin accessibility by directly regulating histones 665 or transcription factors (Gil & Vagnarelli, 2019). Interestingly, in addition to 666 RAS/PP2A-mediated phosphoregulation of epigenetic proteins reported here, PP2A 667 has been shown to dephosphorylate BRD4, HDAC 4/5/7, PRMT1/5 and TET2 that 668 can contribute to chromatin structure regulation (Tinsley & Allen-Petersen, 2022) 669 Therefore, these data collectively indicate that precise control of gene expression 670 relies on the finely tuned balance between kinase and phosphatase activities. The details of the phosphoproteomics pipeline and data analyses related to analyses 722 of RAS and PP2A-regulated phosphosites are described in previous publications 723 (Kauko et al., 2020, Kauko et al., 2015. The raw data can be accessed via the 724 PRIDE partner repository with the dataset identifier PXD001374 (for RAS regulated 725 phosphosites) and PXD016102 (for PP2A-regulated phosphosites). For identification 726 of overlapping phosphosites regulated by RAS, or any of the PP2A conditions, the 727 following filtering criteria was used when assessing the phosphoproteome data 728 normalized as described in (Kauko et al., 2020, Kauko et al., 2015. For inhibition of 729 phosphorylation by RAS targeting: fold change -0.5 log2 and FDR<0.1%. For PP2A-730 regulated proteins fold change 0.5 log2 (for increased phosphorylation by PPP2R1A 731 targeting) and -0.5 log2 (for dephosphorylation by PME-1, CIP2A and SET targeting) 732 and FDR<0.05%. The lower FDR criteria used for RAS data is due to notion that in 733 general these earlier experiments (Kauko et al., 2015) had more variation between 734 the replicate samples due to less sensitive mass spectrometry and inexperience in 735 sample handling.

Pull-down assays 791
To confirm the interaction between PP2A-B56α and HDAC1 pull down assay was 792 performed as described before (PMID: 34145035). Briefly, the cells were transfected 793 with the respective plasmids using the jetPRIME® transfection reagent. 48 h later 794 cells were harvested on ice by scraping and lysed in a buffer containing 100 mM 795 NaCl, 1 mM MgCl2, 10 % glycerol, 0.2 % protease inhibitor tablet (Roche) and 25 796 units/ml Benzonase (Millipore). Cells were rotated at 4°C on a roller and fifteen min 797 later the final concentration of NaCl and EDTA was increased to 200mM and 2mM, 798 respectively. After further rotation of 10 min cells were centrifuged at 16,000 rpm for 799 20 min. 10% of the lysate was stored as input and the remaining was incubated with 800 the 20µl of prewashed GFP trap magnetic beads (ChromoTek GFP-Trap®) at 4°C 801 on a roller for two hours. Post incubation the beads were washed three times using 802 the lysis buffer and eluted by adding 20 µl of 2X SDS loading buffer and boiling at 803 95°C for 10 min. Inputs and the CO-IP samples were further loaded on a 4-20% 804 gradient gel to access the interactions. 805 806

Western blots 807
Cells were lysed in RIPA buffer (50 mM Tris-HCl pH 7.5, 0.5 % DOC, 0.1 % SDS, 808 1% NP-40, and 150 mM NaCl) with protease and phosphatase inhibitors 809 (#4693159001 and #4906837001, Roche) followed by sonication at highest setting 810 with a pulse of ± 30 sec. After centrifugation at 16,000g for 30 min lysates were 811 collected in a fresh tube and protein concentration was determined using BCA assay 812 (Pierce). 6X loading buffer was added to lysates and they were boiled at 95°C for 10 813 min. Equal amounts of lysates were load in 4-20% precast gradient gels (Bio-Rad) 814 and separated at 80-100 Volts. Proteins were blotted using PVDF membrane (Bio-815 Rad) and blocked for 1 h at RT. Membranes were incubated overnight with primary 816 antibody followed by washing. For detection, HRP-labelled secondary antibodies 817 (DAKO) followed by incubation with Pierce™ECL Western Blotting Substrate 818 (Thermo Fisher Scientific) was used, or LI-COR Biosciences secondary antibodies 819 (IRDye 680 or IRDye 800) was used followed by detection by Odyssey® Imaging 820 Systems or Bio-Rad Laboratories ChemiDoc Imaging Systems. 821 822 Antibodies 823 The following antibodies, at the indicated dilutions, were used: To determine the synergy between PP2A activation and HDAC inhibition drug 833 synergy screen was done. H460 cells were seeded in 96 well plate (3000 cells/well) 834 and next day treated with respective drugs for 48 h. Cell viability was measured 835 using the CellTiter-Glo® cell viability end-point assay (Promega), and the synergy 836 was determined using the synergy finder tool 837 (https://synergyfinder.fimm.fi/synergy/20220404143330175356/). HCT116 reporter 838 cells were treated with the respective drugs at their IC50 concentrations, and 48 h 839 later imaged for fluorescence using the IncuCyte ZOOM and/or S3 live cell imaging, 840 and then harvested using RIPA buffer. Fluorescence signal was analyzed using the 841 ImageJ tool while the GFP signal was determined using Western blotting. 842 843

Anchorage-independent colony formation assay 844
For the anchorage-independent colony formation assay, which typically correlates 845 with in vivo tumorigenicity, 2x10 4 cells were resuspended in 1.5ml growth medium 846 containing 0.4% agarose (4% Agarose Gel, Termo Fisher Scientific Gibco; top layer) 847 and plated on 1ml bottom layer containing growth medium and 1.2% agarose in a 848 12-well plate. After 14 days of growth, colonies were stained over night with 1mg/ml 849 Nitro blue tetrazolium chloride (NBT; Molecular Probes) in PBS. Colonies were 850 imaged using a Zeiss SteREO Lumar V12 stereomicroscope. Analysis was done 851 using the ImageJ software. First, the background was subtracted using the rolling 852 ball function with a radius of 50μm, then auto-thresholding was applied to separate 853 the colonies. Area percentage was calculated using the ImageJ built-in function 854 'Analyze Particles' with exclusion of particles smaller than 500μm 2 that are not 855 paired-end --comprehensive --gzip --bedGraph --remove_spaces --buffer_size 80% -911 -cytosine_report --ignore_r2 2. Data were aligned to the Ensembl Homo Sapiens 912 GRCh38 v95 genome and corresponding annotations were used. Incomplete 913 conversions were filtered out with the filter_non_conversion tools from Bismark. 914 Finally, differential methylation analysis was run using Bioconductor (Huber, Carey et 915 al., 2015) • methylKit v1.12.0 (Akalin, Kormaksson et al., 2012) • packages 916 running on R v3.6.1 (R Core Team, 2019 retrieved from https://www.r-project.org/)•. 917 For the methylKit analysis two scenarios were considered: CpG context and tiled 918 standardized peaks was computed from bam files using the bedtools coverage 944 command v2.27.1. Standardized peaks read count was normalized using median of 945 ratios normalization, and then converted to RPM.