Recurrent PPP2R1A Mutations in Uterine Cancer Act Through A

Published OnlineFirst August 2, 2016; DOI: 10.1158/0008-5472.CAN-15-3342 Cancer Molecular and Cellular Pathobiology Research

Recurrent PPP2R1A Mutations in Uterine Cancer Act through a Dominant-Negative Mechanism to Promote Malignant Cell Growth Dorien Haesen1, Layka Abbasi Asbagh2, Rita Derua1, Antoine Hubert1, Stefanie Schrauwen3, Yana Hoorne1,Fred eric Amant3, Etienne Waelkens1, Anna Sablina2, and Veerle Janssens1

Abstract

Somatic missense mutations in the Ser/Thr protein phospha- inhibitor TIPRL1. Dominant-negative Aa mutants retain bind- tase 2A (PP2A) Aa scaffold subunit gene PPP2R1A are among ing to specific subunits of the B56/B0 family and form substrate the few genomic alterations that occur frequently in serous trapping complexes with impaired phosphatase activity via endometrial carcinoma (EC) and carcinosarcoma, two clinically increased recruitment of TIPRL1. Accordingly, overexpression aggressive subtypes of uterine cancer with few therapeutic of the Aa mutants in EC cells harboring wild-type PPP2R1A options. Previous studies reported that cancer-associated Aa increased anchorage-independent growth and tumor formation, mutants exhibit defects in binding to other PP2A subunits and and triggered hyperphosphorylation of oncogenic PP2A-B56/B0 contribute to cancer development by a mechanism of haploin- substrates in the GSK3b, Akt, and mTOR/p70S6K signaling sufficiency. Here we report on the functional significance of the pathways. TIPRL1 silencing restored GSK3b phosphorylation most recurrent PPP2R1A mutations in human EC, which cluster and rescued the EC cell growth advantage. Our results reveal in Aa HEAT repeats 5 and 7. Beyond predicted loss-of-function how PPP2R1A mutations affect PP2A function and oncogenic effects on the formation of a subset of PP2A holoenzymes, we signaling, illuminating the genetic basis for serous EC develop- discovered that Aa mutants behave in a dominant-negative ment and its potential control by rationally targeted therapies. manner due to gain-of-function interactions with the PP2A Cancer Res; 76(19); 1–13. 2016 AACR.

Introduction refs. 2–9). PPP2R1A aberrations occur early during progression in the precursor lesions (3), are distinctive for the serous Althoughnotascommonasendometrioid carcinoma (type subtype (6), and clearly distinguish uterine serous carcinomas I), serous uterine carcinoma (type II) is a highly aggressive from the clinicopathological similar ovarian high-grade serous disease characterized by high mortality due to a tendency for carcinomas (1, 7, 8). In the highly aggressive uterine carcino- early metastasis and resistance to chemotherapy (1). While sarcomas and undifferentiated carcinomas, PPP2R1A mutation genome-wide molecular changes in low grade endometrioid is also frequent (Supplementary Table S1). PPP2R1A encodes carcinomas have been revealed through The Cancer Genome the Aa subunit of type 2A protein phosphatases (PP2A), Atlas (2), the molecular changes associated with serous endo- suggesting a prominent, but so far unexplored, role for PP2A metrial cancer (EC) pathogenesis have only recently started to in the etiology of these cancers. emerge. Besides TP53 mutation (in 80%–95% of cases), rela- PP2A phosphatases are well-recognized human tumor sup- tively few additional molecular genetic aberrations can be pressors (10). They consist of a catalytic C subunit (PP2Ac), a found in these cancers, the most prevalent being alterations structural A subunit and one of multiple regulatory B-type in PPP2R1A, PIK3CA, FBXW7, CCNE1, and CHD4 (3–5). subunits defining substrate specificity of the holoenzyme Specifically, PPP2R1A is mutated in 18.4% to 43.2% of (11). Because different isoforms exist of A, B, and C subunits, all cases, depending on the study (Supplementary Table S1; their combination results in numerous PP2A holoenzymes, each with different signaling functions in a wide variety of physiological processes (12). The A subunit forms the flexible 1Laboratory of Protein Phosphorylation and Proteomics, Department of Cellular and Molecular Medicine, KU Leuven, Leuven, Belgium. 2VIB scaffold between the C and B-type subunits, and is composed of Center for the Biology of Disease, Department of Human Genetics, KU 15 HEAT (Huntington, elongation factor 3, A, TOR) motifs 3 Leuven, Leuven, Belgium. Laboratory of Gynaecological Oncology, (13). Biochemical and structural studies have demonstrated Department of Oncology, KU Leuven, Leuven, Belgium. that interaction with a regulatory B-type subunit is mediated Note: Supplementary data for this article are available at Cancer Research by HEAT repeats (HR) 1–8, while interaction with C occurs Online (http://cancerres.aacrjournals.org/). through HR 11–15 (14–18). Despite these different interaction Corresponding Author: Veerle Janssens, University of Leuven (KU Leuven), domains, C and B subunit binding to the A subunit does Gasthuisberg O&N1, Herestraat 49, PO-box 901, B-3000 Leuven, Belgium. Phone: not occur independently from each other (19). For instance, 32-16-330-684; Fax: 32-16-330-735; E-mail: [email protected] N-terminal deletions of Aa inhibit C binding (14), and B55/Ba doi: 10.1158/0008-5472.CAN-15-3342 and B56/B0g3 do not bind to a C-terminally truncated Aa 2016 American Association for Cancer Research. (15, 20), while PR72/B" does (20). This suggests cooperativity

www.aacrjournals.org OF1

Haesen et al.

between specific B-type subunits and C in binding the A Cell lines subunit, while other B-type subunits may bind A independently HEK293, HEK293T, and HEC-1-A (ATCC), characterized by from C (21). Particularly the conserved C-terminal tail of the C Short Tandem Repeat profiling, were used at low passage number subunit provides additional, stabilizing contacts with B55/B (<15) immediately after receipt or after resuscitation from early and most B56/B0 subunits, but not B56/B0d,PR72/B"orstriatin/ made stocks. B"', to promote holoenzyme assembly (22). Thus, the deter- minants governing PP2A trimer assembly are significantly Lentiviral transduction dependent on the B-type subunit that is incorporated. HEK293T cells were transfected with WT/mutant Aa cloned While some PP2A complexes may be proto-oncogenic into lentiviral pLA CMV N-Flag, or with empty vector alone, in (23–25), most PP2A trimers suppress oncogenic signaling the presence of pCMV-deltaR8.91 and pMD.G-VSVG, using events (12, 26, 27). Hence, inhibition of PP2A—in the pres- Turbofect (Thermo Scientific). Lentiviral TIPRL1 shRNAs or GFP ence of oncogenic RasV12—is an absolute requirement to shRNA control (Supplementary Table S2), originally cloned in achieve full transformation of human epithelial cells immor- pLKO.1-puro (TRC 1/1.5 human shRNA library), were generated talized by expression of telomerase and SV40 large T (26–29). similarly. Twenty-four hours later, the supernatant was used to There is increasing evidence of PP2A deregulation in human transduce HEC-1-A EC cells. solid cancers and hematologic malignancies (30–32). Partic- ularly, cancer-associated missense mutations have been re- PP2A subunit binding assays ported in PPP2R1A and PPP2R1B, encoding the nonredundant HEK293 cells were transfected with appropriate PP2A sub- Aa and Ab subunits. Inactivating Ab mutations occur with unit expression vectors using PEI. 48 hours post-transfection, high frequency (up to 15%) in lung and colon cancers (33), lysates were prepared in NET buffer, and GST pulldowns, FLAG while Aa mutations have less frequently (up to 7%) been pulldowns, HA pulldowns or GFP-trapping were executed as reported in melanoma, lung and breast carcinoma (34). Most described (40). Bound proteins were boiled in 2x NuPage sample Ab and all Aa mutations affect PP2A holoenzyme formation, buffer (Invitrogen), and analyzed by immunoblotting on 4% either by interrupting interaction between A and C subunits, to 12% gels (BioRad). Membranes were blocked in 5% milk in between A and all or specific B-type subunits, or both (26, 28, TBS/0.1% Tween-20, developed with primary antibodies (Sup- 35, 36). For the characterized Aa mutants E64D/G and plementary Table S3), horseradish peroxidase-coupled secondary R418W, these binding defects result in functional haploinsuf- antibodies (DAKO) and chemiluminescence (Westernbright ficiency that promotes transformation via activation of ECL-HRP, Isogen Life Science). Densitometric quantification was the PI3K/Akt pathway (28). Accordingly, heterozygous Aa done with ImageJ. For statistics, we applied one-way multiple E64D/G knock-in mice show increased susceptibility to ben- comparisons ANOVA to the average values of all quantified bands zopyrene-induced lung cancerogenesis (37). of a given condition on a given gel. Here, we characterized 11 PPP2R1A missense mutants, commonly observed in serous uterine carcinoma. All these "IP-on-IP" approach Aa mutations are heterozygous, cluster in HEAT repeats 5 or HEK293 cells were cotransfected with expression vectors 7, and many of them are recurrent (Supplementary Table S1), for EGFP-TEV-B56d (or B56g1) and HA-Aa (WT or mutant). also in other cancer types (38). In concordance with the Forty-eight hours later, GFP-trapping was performed, and observed genetic data, our biochemical and functional assays the trapped complexes were incubated overnight at 4Cwith revealed a dominant mechanism-of-action of these mutants 0.2 mg/mL recombinant Tobacco Etch Virus (TEV) protease in in EC. cleavage buffer (TBS, 1 mmol/L DTT, 0.5 mmol/L EDTA). After the addition of EDTA, PMSF, and TLCK (all 1 mmol/L), TEV eluates were subjected to HA pulldown, and the washed pre- Materials and Methods cipitates were analyzed by immunoblotting. Site-directed mutagenesis Aa was cloned into HA-tag eukaryotic expression vector PP2A activity assays pMB001. PCR-based site-directed mutagenesis (Stratagene) to HA-agarose beads from IP-on-IP, or GFP beads from GFP- generate Aa point mutants (P179R, R182W, R183G, R183Q, trapping experiments, were washed with 20 mmol/L Tris.HCl R183W, R249H, S256F, S256Y, W257C, W257G, R258H, pH7.4, 1 mmol/L DTT, and resuspended in 60 to 80 mL enzyme R418W) and the AA1-411 deletion mutant was performed directly dilution buffer (Millipore). Twenty microliters of this "phos- in pMB001 using Pwo polymerase (Roche) and complementary phatase suspension" or 20 mL of a 1/80 dilution of purified DNA primers (Sigma Genosys) containing the desired mutations PP2A dimer (0.1 Unit/mL, Millipore) was incubated with 9 mLof (Supplementary Table S2). WT and mutant (P179R, R182W, 2 mmol/L stock of R-R-A-pT-V-A phosphopeptide for 10 to 60 R183G, R183Q, and S256F) Aa coding regions were subcloned minutes at 30C (in the linear range of the assay). If applicable, in pLA CMV N-Flag via In-Fusion HD (Clontech). Eukaryotic 0.5 to 2 mg of recombinant TIPRL1, diluted in 20 mL enzyme expression vectors for GST-tagged PP2A B-type subunits were as dilution buffer, was preincubated with the "PP2A" suspension described (39). B55a, B56g1, and B56d (40) were cloned in for 15 minutes at 30C, before the reaction was started by pEGFP-TEV, and B56e in pEGFP-C1 (Clontech). STRN3 cDNA addition of the substrate. Released phosphate was determined (clone HsCD00623024) was obtained from DNASU repository by malachite green (40). (The Biodesign Institute; ref. 41) and subcloned into pEGFP-TEV. TIPRL cDNA (isoform 1) was obtained by RT-PCR on mRNA Recombinant TIPRL1 production isolated from HEC-1-A and cloned into pLA CMV N-Flag, pEGFP- His-tagged TIPRL1 (pET15b) was expressed in E. coli BL21 TEV, and pET15b. cells at 37Candpurified from the soluble bacterial lysate

OF2 Cancer Res; 76(19) October 1, 2016 Cancer Research

Mechanism of Action of PP2A Aa Subunit Cancer Mutants

through metal affinity chromatography (Ni-NTA beads, Affiland). Results After five washes in 50 mmol/L Tris–HCl pH 8.0, 300 mmol/L Cancer-associated Aa mutations dramatically affect PP2A NaCl and 13 mmol/L imidazole, and elution in 250 mmol/L holoenzyme formation imidazole, the purified protein was dialyzed against 20 mmol/L To determine how recurrent EC-associated Aa mutants affect Tris–HCl pH 7.4/ PEG 10,000, and stored in 60% glycerol at PP2A holoenzyme assembly, we evaluated their binding to 80 C. Concentration was determined by absorbance at 280 nm PP2Ac. We assessed interaction of endogenous PP2Ac with (NanoDrop 2000; Thermo Scientific). HA-tagged WT Aa, the PP2Ac-binding deficient R418W Aa mutant (28, 35) and 11 EC-associated Aa mutants expressed PPP2R1A exome sequencing in HEK293 cells using HA pulldown. All mutants, but R249H, RNA was isolated from HEC-1-A cells using TRIzol (Life showed reduced C binding, when compared with WT Aa Technologies), reverse-transcribed with MuLV RT (New Eng- (Fig. 1A and B). land BioLabs), and amplified with Pwo polymerase (Roche) Using GST pulldown, we next determined the binding prop- 0 0 0 and primers 5 -ATGGCGGCGGCCGACGGCGACG-3 and 5 - erties of HA-tagged Aa mutants to GST-tagged B-type subunits of 0 TCAGGCGAGAGACAGAACAGTCAG-3 .Thepurified fragment the B55/B, B56/B0 and PR72/B" families. Importantly, in our (Illustra GFX; GE Healthcare) was subjected to Sanger sequenc- assays, the melanoma-linked R418W mutant behaved as previ- ingusingthesameprimers. ously reported (Supplementary Fig. S1; refs. 28, 35). We found that only Aa-R249H mutation, which did not affect C binding, Mass spectrometry also did not affect the interaction with all tested B-type subunits FLAG-tagged WT or mutant Aa were isolated from trans- (Fig. 1C and D), while all other Aa mutations led to altered – duced HEC-1-A cells by FLAG pulldown. HEC-1-A cells trans- interaction patterns (Fig. 1C E; Supplementary Fig. S1). Most Aa duced with the pLA vector were used as negative control. mutants showed a dramatic decrease in binding to the B55/B – subunits, Ba and Bb. On the one hand, the binding defects to Proteins were eluted in 50 mmol/L Tris HCl pH 8.0, contain- 0 00 ing 50 mmol/L NaCl and 1.5 mg/mLFLAGpeptide(MDYKDH- B56/B and PR72/B family members were different for different DGDYKDHDIDYKDDDDK). Eluates were trypsin digested Aa mutants. For example, P179R, R183G/W, and W257C fi 00 and desalted on C18 ZipTips (Millipore) before analysis on mutants ef ciently bound to PR72/B , whereas R182W, a hybrid quadrupole-orbitrap mass spectrometer (Q Exactive; R183Q, S256F/Y, W257G, and R258H lost the ability to interact Thermo Fisher Scientific). Relative quantification of the inter- with this subunit (Fig. 1E; Supplementary Fig. S1), indicating that a both the position of the mutation and the type of substitution actomes of mutant versus WT A was executed with Progenesis 00 0 (Nonlinear Dynamics) incorporating protein identifications impact PR72/B binding. For the B56/B subunits, Aa mutants obtained with MASCOT (Matrix Science) using uniprot_sw- overall showed the least binding defects to B56d and the most prot (release 07/15/2015; Homo sapiens; 20,279 entries). Pro- pronounced defects in binding to B56b and B56g1 (Fig. 1E; fi teins with an abundance that was at least 3 times higher in any Supplementary Fig. S1), revealing remarkable isoform-speci c differences in binding behavior. of the experimental conditions than in the "FLAG-only" con- 000 trol were considered as true interaction partners (96 proteins). Finally, binding of HA-tagged mutant Aa to striatin/B Only abundances of peptides with MASCOT scores 25 (sig- subunits was monitored in GFP-trapping assays with GFP- nificance threshold) were taken into account. Resulting pro- tagged STRN3 (Fig. 2). Most of the Aa mutants showed tein abundances of interactors in each experimental condition increased binding to GFP-STRN3 as compared with WT Aa were subsequently normalized according to Aa input abun- (Fig. 2A). However, upon STRN3 coexpression, several Aa dance. From those values, relative quantification to WT Aa was mutants were expressed at higher levels (Fig. 2A and B), while this was not consistently observed upon coexpression of any of determined. Heat maps of differential interactions were made 0 00 with GENE-E. the B55/B, B56/B , or PR72/B subunits (Fig. 1C, Supplemen- tary Fig. S1). After we normalized the increase in binding to the expression levels, we found that two Aa mutants, S256F and Soft agar colony assay R258H, still bound significantly better to STRN3 as compared 4 A total of 1 10 cells/well were seeded in triplicate into with WT Aa, while all others bound at least as well as WT Aa 0.35% top agar/0.5% base agar (Sigma Aldrich) supplemented (Fig. 2C). with 2 mg/mL Fungizone (Invitrogen) in 6-well plates. Every 2 to To provide further insights into the different binding behav- þ 3 days, fresh medium (DMEM 10% fetal calf serum) was ior of the Aa mutants to different B-type subunits, we coex- dropped on top. After four weeks, colonies were counted pressed HA-tagged WT or mutant Aa (R182W, R183G/Q, and fi (Motic AE31 microscope, 4magni cation). S256F) with GFP-tagged B-type subunits (B55a,B56g1, B56d, B56e, and STRN3) and evaluated the presence of endogenous Xenografting PP2Ac and the HA-tagged Aa subunits in the GFP-trapped A total of 5 106 cells in 200 mL PBS were subcutaneously complexes (Supplementary Fig. S2). Unfortunately, anti-A injected in 6-week-old female NMRI-nu (nu/nu) mice (Jackson immunoblotting could not unambiguously reveal the presence Laboratory). Xylazine (8 mg/kg)/ketamine (90 mg/kg) was of endogenous Aa because of insufficient resolution of the HA- used as local anaesthetic. Per condition, three mice were inject- tagged and endogenous A bands. Nevertheless, we found that ed in both flanks (n ¼ 6). All procedures were approved by KU the PP2Ac binding patterns in all cases completely mimicked Leuven ECD (P104/2012). Tumor number was determined those of the GFP-B-type subunits' inputs, while the anti-HA one month after injection. Tumor size was calculated as: tumor signals were consistent with our binding assays described above width2 (tumor length/2). (Figs. 1 and 2). Because PP2Ac binding patterns completely

www.aacrjournals.org Cancer Res; 76(19) October 1, 2016 OF3

Haesen et al.

Figure 1. Binding of PP2Ac, B55/B, B56/B0, and PR72/B00 subunits to EC-associated Aa mutants. A, WT Aa, the melanoma-associated R418W mutant and 11 EC-associated Aa mutants (all HA-tagged), or an empty HA vector () were transfected into HEK293 cells. Following HA pulldown, interaction with endogenous PP2Ac was examined by immunoblotting (IB). B, after quantification with ImageJ, ratios between HA and C signals were determined and calculated relative to Aa WT. Mean values and a representative image (A) of four independent experiments are shown (, P < 0.05; , P < 0.01). C, GST-tagged B-type subunits or empty GST vector () were coexpressed in HEK293 cells with HA-tagged WT or mutant Aa (R249H and R183Q). HA signals in the lysates and GST pulldowns were determined by immunoblotting. D, ImageJ-quantified ratios were determined between GST and HA signals and calculated relative to Aa WT (set to 100% for each B- type subunit pulldown). Mean values and a representative image (C) of three independent experiments are shown (, P < 0.05; , P < 0.01). E, summary of mutant Aa binding defects. Numbers indicate percentage of residual Aa mutant binding compared with WT Aa (set at 100%). Color code: <30%, orange, Aa mutant binding is significantly decreased; >30%, yellow, modestly decreased; green, does not significantly differ from WT.

OF4 Cancer Res; 76(19) October 1, 2016 Cancer Research

Mechanism of Action of PP2A Aa Subunit Cancer Mutants

Figure 2. Binding of B000 subunit STRN3 to EC-related Aa mutants. A, GFP-tagged STRN3 or GFP alone () was coexpressed with HA-tagged WT or mutant Aa in HEK293 cells. HA signals in lysates and GFP-trapped complexes were determined by immunoblotting (IB). After quantification with ImageJ, ratios were determined between GFP and HA signals in the GFP-trapped complexes, and between vinculin and HA in the lysates. B, mean values of HA/vinculin ratios in the lysates, calculated relative to Aa WT (set to 100%), from three independent experiments (, P < 0.05; , P < 0.01). C, mean values from three independent experiments of HA/GFP ratios in the GFP-trapped complexes, calculated relative to Aa WT, and further corrected for differences in expression with Aa WT (cf. B; , P < 0.05). mimicked those of the GFP-B-type subunits' inputs, it is likely (MS) in a semiquantitative way as outlined in Supplementary that these B-type subunits bind a mixture of HA-tagged and Table S4. This approach allowed us to assess the complete Aa endogenous A subunits, and that, despite the overall decrease interactome in endometrial epithelial cells in an unbiased, in PP2Ac binding to the Aa mutants (Fig. 1A), PP2Ac retrieval nontargeted way. We identified 96 cellular proteins as specific into maintained mutant Aa-B-type subunit complexes may not coeluting interaction partners of Aa:49oftheseinteractorshad be majorly affected. Although some B-type subunits (B56g, previously been reported in PP2A interactome studies from B56d, PR72, and STRN3) were intrinsically capable of forming HEK293 (42–44) and Hela cells (Fig. 4A; ref. 45), the 47 others A-B complexes without PP2Ac (Supplementary Fig. S3), we were novel (Supplementary Fig. S4). The known interaction found normal PP2Ac binding in mutant Aa-B56d,Aa-B56g1, partners include 12 PP2A subunits, 2 PP4 subunits, a cellular and Aa-PR72 complexes isolated by an IP-on-IP approach PP2A inhibitor (TIPRL1; ref. 46), several well-established (Fig. 3 and data not shown). Taken together, these results B-type subunit interactors (9 components of the STRIPAK indicate that the observed decrease in PP2Ac binding to the complex,refs.42and43;PPFIA1/2,ref.47),andanumber Aa mutants is indirectly due to loss of Aa binding to B-type of poorer characterized PP2A interactome constituents, such as subunits that show mutual cooperativity with PP2Ac to bind A the Integrator subunits (44). (B55a/b,B56a/b/e). Overall, the R183Q and S256F mutants show the most pronounced binding deficiencies to several of the identified inter- Interactome of WT and mutant Aa in endometrial cancer cells actors, while the P179R mutant shows the least compared with To validate the alterations in the interaction pattern of Aa WT Aa. Consistent with our binding assays in HEK293 cells mutants, we expressed WT Aa and 5 FLAG-tagged Aa mutants (Fig. 1), the MS data reveal decreased binding for PP2A C, (P179R, R182W, R183G, R183Q, and S256F) in the EC cell line B55/B, and some B56/B0 subunits to the Aa mutants, while HEC-1-A using a lentiviral approach. P179R and S256F binding to other B56/B0 subunits is sustained (Fig. 4A). We mutants have the highest rate of recurrence in uterine cancer confirmed these results by immunoblotting of the FLAG pull- (Supplementary Table S1), while the three other mutants downs with (few) available isoform-specific PP2A subunit represent groups with different PP2A subunit binding profiles antibodies (Fig. 4B). Altogether, the data suggest that these (Fig. 1E). Following anti-FLAG pulldown and subsequent elu- mutations are loss-of-function. tion by the FLAG peptide, we analyzed the trapped complexes In contrast, the MS data suggested largely unaltered or and their putative interaction partners by mass spectrometry increased binding of Aa mutants to the B000/STRN subunits,

www.aacrjournals.org Cancer Res; 76(19) October 1, 2016 OF5

Haesen et al.

A B B56δ - GFP GFP WT Aα R183G R183Q S256F – –

IB: B56d

IB: HA (Aα)

IB: C HA Pull down Pull HA

Ratio C/HA 1.0 0.81.01.2

IB: B56d

IB: HA (Aα)

C GFP-TEV Supernatant IB: C

IB: B56d

IB: HA (Aα) Lysates IB: C

Figure 3. Binding of PP2Ac is retained in mutant Aa-B complexes. A, schematic representation of the IP-on-IP principle. This approach enabled isolation of mutant A-B complexes and distinguish the amount of PP2Ac in these complexes from the amount of PP2Ac binding to the B-type subunit through endogenous (endog.) A subunit. B, PP2Ac binding is unaffected in mutant Aa–B56d complexes. EGFP-TEV-B56d was coexpressed with HA-tagged WT or mutant Aa. Following serial pulldown (IP-on-IP), i.e., GFP-trapping, followed by TEV cleavage and HA pulldown on the GFP-TEV supernatants, retrieval of PP2Ac in the Aa-B56d complexes was analyzed by immunoblotting (IB). After quantiﬁcation, ratios between C and HA signals were determined. C, average C/HA ratios from three independent experiments.

indicating gain-of-function effects of these mutations (Fig. 4A). Taken together, we observed differential effects of Aa muta- This result was validated by immunoblotting (Fig. 4C). The tions on the formation of specific PP2A complexes, making it reliability of the MS data is further underscored by the obser- challenging to define their mode of action as loss-of-function or vation that interactors belonging to the same established multi- gain-of-function. protein complex (e.g., STRIPAK and Integrator) showed similar binding patterns to the different Aa mutants (Fig. 4A). Finally, Cancer-associated Aa mutants increase anchorage- the MS data revealed the most obvious and strongest gain-of- independent cell and xenograft growth of EC cells function interaction of the mutants with the PP2A inhibitor We next analyzed whether ectopic expression of the Aa TIPRL1 (Fig. 4A). We validated the enhanced interaction of Aa mutants in HEC-1-A cells affects oncogenic properties of these mutants to TIPRL1 by immunoprecipitation and TIPRL1 immu- cells. We confirmed that both PPP2R1A alleles in HEC-1-A are noblotting (Fig. 4C). TIPRL1–Aa interaction occurred indepen- WT by Sanger sequencing. All tested Aa mutants showed an dently from Aa-C binding through Aa HEAT-repeats 1–10 increased tumorigenic potential versus WT Aa upon their (Supplementary Fig. S5A). PP2Ac retrieval within this complex expression in HEC-1-A cells. The effect was particularly pro- was not affected by Aa mutation, but correlated with WT/ nounced for R183G, R183Q, and S256F mutants as both mutant Aa binding (Supplementary Fig. S5B). Further bio- anchorage-independent and xenograft growth (Fig. 5A and chemical characterization of PP2A–TIPRL1 complexes showed B) were significantly increased compared with cells expressing that TIPRL1 could be retrieved with B56d (Supplementary Fig. WT Aa or empty vector. S6) or other B-type subunits tested (Supplementary Fig. S7), To determine the effects of Aa mutants on oncogenic sig- despite the observation that high TIPRL1 overexpression naling, we performed immunoblotting analysis of the phos- markedly decreased overall expression of A, B, and C subunits phorylation status of cancer-related PP2A substrates and com- (Supplementary Figs. S6 and S7). ponents of cancer-associated signaling pathways (Fig. 5C–E;

OF6 Cancer Res; 76(19) October 1, 2016 Cancer Research

Mechanism of Action of PP2A Aa Subunit Cancer Mutants

A B FLAG Pulldowns Total cell lysates P179R R182W R183G R183Q S256F WT - P179R R182W R183G R183Q S256F WT -

IB:Flag (A a ) * IB:A

IB:C PP2A 0.8 0.7 0.5 0.5 1.3 1.9 IB:B55 a

IB:B56a PP4

0.5 0.2 0.6 0.2 0.5 3.1

IB:B56γ2

STRIPAK 0.3 0.2 0.8 1.2 0.5 1.9

IB:B56γ3

0.1 1.2 0.8 1.2 0.8 1

IB:B56d

0.8 1.1 0.5 0.8 0.8 1.8

IB:B56ε Integrator 0 0.5 0 0 1.8 3.1

C FLAG Pulldowns Total cell lysates P179R R182W R183G R183Q S256F WT - P179R R182W R183G R183Q S256F WT -

IB:Flag (Aa ) * IB:A

IB:TIPRL

0.4 0.9 0.4 2.1 1.7 0

IB:STRN3

1 0.9 0.5 1.5 1.6 0.5

Figure 4. Mass spectrometry–based identification of WT and mutant Aa interactomes in HEC-1-A cells. A, semiquantitative heat maps of WT and mutant Aa interactomes. Forty-nine established Aa interacting proteins were identified by MS in FLAG peptide eluates of FLAG pulldowns, performed on lysates of HEC-1-A cells stably expressing N-terminally FLAG-tagged Aa variants, or FLAG-tag alone. Heat maps display the relative abundance of each interaction partner in the pulldowns; regardless of absolute abundances, the interactor occurring in the highest abundance is colored dark red, and the least abundant partner is colored dark blue. Data interpretation, thus, needs to be done with caution, preferably with the list of absolute abundances at hand (Supplementary Table S4), as small differences in abundance may be identically displayed as marked changes. B, validation of endogenous PP2Ac, B55/ B, and B56/B0 subunit binding to mutant Aa. The presence of these subunits in FLAG pulldowns (left) and cell lysates (right) of transduced HEC-1-A cells was determined by immunoblotting (IB) with available isoform-specific antibodies. , FLAG-tagged Aa in the lysates. Quantifications represent calculated ratios between C or B signals and FLAG signals for a given Aa mutant in the FLAG pulldowns. C, validation of gain of TIPRL1 and STRN3 binding to mutant Aa. Same approach as in B to determine endogenous TIPRL1 and STRN3 in FLAG pulldowns.

Supplementary Fig. S8). We found signiﬁcantly increased onco- tial induced by Aa mutants could be triggered by upregulation genic signaling in cells expressing the strong tumorigenic of the Akt, GSK3b, and p70S6K pathways. mutants, R183G/Q and S256F, as demonstrated by hyperphosphorylation of p70S6K Thr389 and S6 Ser235/236 (Fig. 5C), Mutant Aa proteins act in a dominant manner through GSK3b Ser9, and Akt Thr308 (Fig. 5D). On the other hand, we formation of substrate-trapping complexes observed only a mild effect for the P179R and R182W mutants Increased tumorigenic potential of HEC-1-A cells upon on phospho-GSK3b, and on phospho-S6 for R182W that is expression of Aa mutants without any reduction of endoge- consistent with their milder tumorigenic phenotype (Fig. 5A nous Aa levels suggests that the Aa mutants behave in a and B). Remarkably, ERK1/2 Thr202/Tyr204 phosphorylation dominant-negative manner. To further underscore dominant- was decreased upon expression of all mutants (Fig. 5E). Taken negative effects of the mutant Aa alleles, we determined the together, these data suggest that increased tumorigenic poten- phenotypic effects of gradually increased expression of the least

www.aacrjournals.org Cancer Res; 76(19) October 1, 2016 OF7

Haesen et al.

A P179R R182W R183G R183Q S256F WT Aα PLA

B C ) 3 Tumor volume (mm Tumor

Time (days) D E F R183Q R183G P179R - P179R R182W R182W WT S256F WT R183G S256F R183Q - S256F - R182W R183G R183Q P179R WT Flag (Aα) Flag (Aα)αgalF ( A )

P-p70S6k P-Akt P-ERK1/2 Vinculin Vinculin Vinculin

p70S6k Akt ERK1/2 Vinculin Vinculin Vinculin

P-S6 P-GSK-3b Vinculin Vinculin

S6 GSK-3b Vinculin Vinculin

Figure 5. Tumorigenic phenotypes of EC cells expressing Aa mutants. A and B, anchorage-independent growth of HEC-1-A cells, ectopically expressing mutant or WT Aa. The pictures give an idea of the number and size of colonies obtained for each condition at a given time point after seeding (A). The graph (B) represents the mean (SD) colony number from three replicates and six measurements (ANOVA; , P < 0.01; , P < 0.001). C, tumor growth of xenografted HEC-1-A cells. Size (graph) and number of tumors (n) are indicated for each condition (n ¼ 6). Differences in tumor size were analyzed via ANOVA (, P < 0.05; , P < 0.01). D–F, effects of mutant Aa expression on oncogenic signaling. Lysates of WT/mutant Aa expressing HEC-1-A cells were analyzed with indicated antibodies, and immunoblot signals quantiﬁed (Supplementary Fig. S8). "Total" and "Phospho" signals were determined on different blots, which were both redeveloped for vinculin as a loading control. Results from one of three independent experiments are shown.

oncogenic mutant, P179R. Increasing expression of FLAG- phosphorylation was not affected (Supplementary Fig. S9A). tagged Aa P179R mutant, but not WT Aa,indeedcorrelated Given that Aa mutants with highest gain-of-binding to with increasing GSK3b Ser9 phosphorylation (Supplementary STRN3 and TIPRL1 showed the more pronounced onco- Fig. S9A and S9B), while S6 Ser235/236 and Akt Thr308 genic behavior (Figs. 2,4C,5) and a decrease was observed in

OF8 Cancer Res; 76(19) October 1, 2016 Cancer Research

Mechanism of Action of PP2A Aa Subunit Cancer Mutants

A B C

0 0.5 2 0 0.5 2 TIPRL1 (μg) TIPRL1 (μg)

D E

0 0.5 2 0 0.5 2 TIPRL1 (μg) TIPRL1 (μg)

Figure 6. Effects of recombinant TIPRL1 on activity of different PP2A complexes. A, Coomassie Blue staining of His-TIPRL1, purified from E. coli. B and C, TIPRL1 specifically inhibits mutant Aa-containing trimers. PP2A-B56g1(B)orPP2A-B56d (C) trimers harboring HA-tagged WT or mutant Aa,isolatedby IP-on-IP from GFP-TEV-B56g1– or GFP-TEV-B56d–expressing HEK293 cells, were preincubated for 150 at 30Cwithbufferor0.5–2 mg recombinant TIPRL1. PP2A activity was determined on R-R-A-pT-V-A and calculated relative to the buffer condition (set at 100%). D, same experiment with PP2A-B56g1 and PP2A-B56d trimers isolated by GFP-trapping from GFP-B56g1– and GFP-B56d–expressing HEK293 cells. E, same experiment with de novo–purified PP2A dimer. dephosphorylation of GSK3b,Akt,andp70S6K(Fig.5Cand To further underscore TIPRL1 contribution to the dominant- D), which are substrates of PP2A-B56g-and/orB56d-specific negative effects of the Aa mutants on anchorage-independent complexes (27, 28, 48–50), Aa mutants might trigger tumor- growth in EC cells, we suppressed TIPRL1 expression in Aa igenic transformation through gain-of-function interaction S256F-expressing HEC-1-A cells using lentiviral shRNAs tar- with TIPRL1. Anti-FLAG immunoprecipitations confirmed geted against TIPRL1. TIPRL1 silencing rescued the increase in increasing binding of Aa P179R to PP2Ac, TIPRL1, and B56d anchorage-independent growth (Fig. 7A and B) and GSK3b (Supplementary Fig. S9C), the only B-type subunit that signif- Ser9 hyperphosphorylation (Fig. 7C) that were triggered by this icantly retained binding to this Aa mutant in HEC-1-A cells Aa mutant. However, we did not observe any decrease in S6 (Fig.4B),andforwhichGSK3b is an established substrate (48). Ser235/236 phosphorylation (Fig. 7C). These results confirm Thus, by retaining binding to certain B-type subunits, such as that tumorigenic transformation triggered by Aa mutants is B56g and B56d,Aa mutants might form substrate trapping TIPRL1 dependent. complexes that are capable of competing with active, WT Aa- containing PP2A–B56g/d trimers for substrate binding. Such dominant-negative complexes might be catalytically impaired Discussion through increased recruitment of TIPRL1. Consistently, addi- Heterozygous missense mutation of PPP2R1A is a recurrent tion of recombinant TIPRL1 to mutant Aa-containing B56g or event in human cancer. There is a prevailing view that PPP2R1A B56d trimers inhibited PP2A activity in vitro (Fig. 6A–C), while mutations are loss-of-function due to impairment of PP2A holo- this was not the case in WT Aa-containing trimers, regardless of enzyme formation (28, 35, 37). This results in haploinsufficiency whether they were obtained by the IP-on-IP approach (Fig. 6B of Aa, primarily affecting holoenzyme formation with B-type and C) or by direct GFP-trapping of GFP-tagged B56g/d subunits subunits that no longer bind to mutant Aa and show the least (Fig. 6D). Moreover, TIPRL1 markedly inhibited the PP2A core binding affinity for WT Aa. When tested in HEK293 cells, forma- dimer in vitro (Fig. 6E). tion of PP2A–B56g trimers becomes significantly impaired, like

www.aacrjournals.org Cancer Res; 76(19) October 1, 2016 OF9

Haesen et al.

A B * shTIPRL1_1

*** ** S256F A a

WT A a

Figure 7. shTIPRL1_2 TIPRL1 knockdown rescues increased anchorage-independent growth and GSK3b phosphorylation in S256F C S256F WT A a mutant Aa-expressing HEC-1-A cells. S256F-mutant Aa or WT Aa- expressing HEC-1-A cells were lentivirally transduced with two S256F A a different TIPRL1-targeted shRNAs (Supplementary Table S2), and analyzed for effects on anchorage- shGFP shGFP shTIPRL1_1 shTIPRL1_1 shTIPRL1_2 shTIPRL1_2

independent growth in soft agar assays IB:Vinculin (A and B) and for oncogenic signaling by immunoblotting with indicated

WT A a antibodies (C). The pictures (A) IB:TIPRL1 illustrate colony size, while the 0.9 0.81.4 0.9 1.20.6 Ratio TIP/vinc. graph (B) represents colony number shGFP (, P < 0.05; , P < 0.01; , P < 0.001). IB:Flag (Aα)

1.5 1.51.0 1.3 1.32.0 Ratio Flag/vinc.

IB:P-GSK3β S256F A a IB:GSK3β

0.6 0.7 2.3 1.8 1.0 0.4 Ratio Phos./Tot.

IB:P-S6

WT A a IB:S6

2.4 1.0 1.1 1.8 0.2 0.6 Ratio Phos./Tot.

that resulting in Akt hyperphosphorylation and increased tumor On the other hand, the binding deﬁciencies of Aa mutants growth (28). However, our results underscore that this model, with B56/B0 and PR72/B00 subunits were much more diverse, based on the characterization of a few, sporadically occurring Aa highlighting the challenges associated with the existence of so mutants, should not be generalized. many different subunits and warning for over-generalization. Our binding assays revealed that 10 out of 11 Aa mutants Only one subunit, B56d,boundallAa mutants in comparable tested showed an unexpectedly complicated pattern of inter- amounts as WT. For the striatin/B000 subunits, our data revealed actions. Only Aa-R249H behaved similar to WT-Aa,andmay unexpected gain-of-binding to Aa mutant proteins. Increased therefore be a nonpathologic passenger mutant. Crystallo- binding of striatin/B000 to the Aa mutants may in part be an graphic data support this view, as R249 is a buried residue, indirect consequence of their generally decreased binding to while all other mutated residues face the interaction surface subunits of other subclasses (B55/B, B56/B0,PR72/B00). The with B and C subunits (17). All other Aa mutants showed loss general decrease in PP2Ac binding to 10/11 Aa mutants is of binding to B55/B subunits, indicating that a single missense consistent with reported decreased PP2Ac binding to Aa HR5 mutation in HR5 or HR7 severely affects binding to isoforms of and HR7 deletion mutants (14), highlighting a complex rela- this B subunit class. Due to B-C cooperativity in A subunit tionship between B and C subunits within the holoenzyme: not binding (14, 15, 19, 20), loss of B55/B binding is likely also one only B subunits may require stabilizing contacts with PP2Ac to of the indirect causes of the observed overall decrease in C bind A and form stable trimers (15, 20, 22, 39), stable PP2Ac binding to the A mutants. binding to A may also depend on the presence of a B subunit.

OF10 Cancer Res; 76(19) October 1, 2016 Cancer Research

Mechanism of Action of PP2A Aa Subunit Cancer Mutants

Thus, the overall reduced C binding to the Aa mutants is oncogenic, via dephosphorylation of Mst1/2 and Mst3/4 kinases probably an indirect consequence of reduced or loss of binding (51–53), impacting, e.g., on regulation of NF-kB (54), Hippo of specific B-type subunits. This view is further confirmed by (23, 24), and MEK/ERK (51, 53). The reported role of PP2A retained PP2Ac binding to mutant Aa-B56d (Fig. 3) and mutant striatins in regulating cancer cell migration and invasion (55) Aa-TIPRL1 complexes (Supplementary Fig. S5B). remains worthwhile to follow up, particularly because serous An unbiased MS-based approach to identify differences in uterine tumors are highly metastatic. Reduced ERK phosphory- mutant Aa binding to constituents of the broader Aa inter- lation, found in all mutant Aa-expressing HEC-1-A cells, is in actome in HEC-1-A cells, unexpectedly, revealed TIPRL1 as the any case consistent with a disturbed phosphorylation balance only protein showing significant gain-of-binding to all Aa within STRIPAK due to enhanced STRN–PP2A function. mutants tested. TIPRL1 binding occurred through HR1-10 of In summary, our data provide biochemical and functional Aa, independently of C binding. Addition of recombinant evidence that cancer-associated Aa mutants trigger EC cell TIPRL1 in vitro resulted in decreased PP2A activity, specifically growth through a novel dominant-negative mechanism depen- in the PP2A A-C dimer and in isolated mutant Aa-B56d/g1-C dent on retained binding to tumor suppressive B56/B0 subunits, complexes. Thus, increased TIPRL1 binding results in catalyt- such as B56g and B56d, and gain-of-binding to TIPRL1. ically impaired, mutant Aa-containing PP2A trimers, which act Although TIPRL1 biochemistry remains poorly understood, as dominant-negatives toward WT Aa-containing trimers. The cBioportal data indicate TIPRL overexpression/mutation in sev- dominant-negative nature of the mutants was significantly eral human cancers (38). This all supports a novel pathological further underscored by increased anchorage-independent cell role of TIPRL1 in the biology of PPP2R1A-mutated endometrial and tumor growth, and corresponding increased oncogenic cancers and identifies TIPRL1 as a novel therapeutic target in signaling upon their ectopic expression in HEC-1-A cells, with- these tumors. Our data also highlight the therapeutic potential out prior downregulation of endogenous Aa.Thisdefinitely of treating PPP2R1A mutant tumors with kinase inhibitors argues against a mechanism of haploinsuffciency and virtually directed against components of the Akt and mTOR/p70S6K eliminates the involvement of loss-of-binding to specificB-type signaling pathways, either as monotherapies, or in combination. subunits (e.g., B55 and others) as a major cause of the observed This opens perspectives for the use of PPP2R1A as therapeutic phenotype. stratification marker and improved clinical management of The functional differences between the mutants were these tumors. reflected well at the signaling level as hyperphosphorylation of p70S6K, S6, Akt, and GSK3b were clearly triggered by Aa Disclosure of Potential Conflicts of Interest mutants R183G, R183Q, and S256F, while Aa mutant R182W No potential conflicts of interest were disclosed. triggered hyperphosphorylation of GSK3b and S6, and Aa mutant P179R only of GSK3b, even if increasingly expressed Authors' Contributions (Supplementary Fig. S9). It is well established that activation Conception and design: D. Haesen, L.A. Asbagh, A. Sablina, V. Janssens of these pathways stimulates protein translation, cell prolif- Development of methodology: D. Haesen, L.A. Asbagh, E. Waelkens eration, and survival. Although all mutants showed increased Acquisition of data (provided animals, acquired and managed patients, binding to STRN3, a gain-of-function of PP2A–STRN3 com- provided facilities, etc.): D. Haesen, R. Derua, A. Hubert, S. Schrauwen, plexes cannot explain the observed hyperphosphorylation of E. Waelkens Analysis and interpretation of data (e.g., statistical analysis, biostatistics, the above kinases. Instead, Akt, GSK3b, and p70S6K are computational analysis): D. Haesen, R. Derua, A. Hubert, E. Waelkens, – – reported substrates of PP2A B56g and B56d complexes in V. Janssens several cellular contexts (27, 28, 48–50). Accordingly, in HEC- Writing, review, and/or revision of the manuscript: D. Haesen, E. Waelkens, 1-A cells, the R183Q mutant, causing the most severe tumor A. Sablina, V. Janssens phenotype, retains binding to B56g2/g3 and B56d.This Administrative, technical, or material support (i.e., reporting or organizing mutant also shows the strongest interaction with TIPRL1. For data, constructing databases): Y. Hoorne, F. Amant, A. Sablina Study supervision: A. Sablina, V. Janssens the other mutants, a similar reasoning could be applied, whereby the severity of the observed growth phenotype may 0 Acknowledgments eventually be the combinatory result of (i) the specificB56/B 0 We thank Profs. S. Dilworth and B. Hemmings for kind gift of antibodies, subunit that retains binding, (ii) the number of B56/B sub- V. Feytons for phosphopeptide synthesis, and lab members for stimulating units that retain binding, (iii) the absolute strength of the discussions. retained binding (or affinity) to a specificB56/B0 subunit, and finally (iv) the binding efficiency to TIPRL1. Notably, the more 0 Grant Support B56/B subunits retain Aa mutant binding to form different Funding was provided by the KU Leuven Research Fund (OT/13/094 substrate-trapping complexes, the better putative functional to V. Janssens and A. Sablina), the IAP program of the Belgian federal redundancies between different trimers for dephosphorylation government (P7/13 to V. Janssens and E. Waelkens), Research Founda- of a given substrate will be avoided. Nevertheless, our results tion – Flanders (G.0582.11N to V. Janssens and E. Waelkens; G.0B01.16N clearly demonstrated the TIPRL1 dependency of the observed to V. Janssens), and Flemish Agency for Innovation by Science and phenotypes. Technology (D. Haesen). The costs of publication of this article were defrayed in part by the Although STRN3 gain-of-binding to mutant Aa could not payment of page charges. This article must therefore be hereby marked explain hyperphosphorylation of Akt, p70S6K, or GSK3b in advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate mutant Aa expressing HEC-1-A cells, it remains possible that this this fact. gain-of-function mechanism still contributes to the stronger oncogenic phenotype of these cells. The role of PP2A striatins Received December 18, 2015; revised July 14, 2016; accepted July 14, 2016; within several STRIPAK complexes is indeed considered to be published OnlineFirst August 2, 2016.

www.aacrjournals.org Cancer Res; 76(19) October 1, 2016 OF11

Haesen et al.

References 1. Shih IeM, Wang TL. Mutation of PPP2R1A: a new clue in unveiling the reveals mechanisms of kinase-phosphatase interactions. Sci Signal pathogenesis of uterine serous carcinoma. J Pathol 2011;224:1–4. 2013;6:rs15. 2. The Cancer Genome Atlas Research Network. Integrated genomic charac- 25. Zwaenepoel K, Goris J, Erneux C, Parker PJ, Janssens V. Protein phospha- terization of endometrial carcinoma. Nature 2013;497:67–73. tase 2A PR130/B00alpha1 subunit binds to the SH2 domain-containing 3. Kuhn E, Wu RC, Guan B, Wu G, Zhang J, Wang Y, et al. Identification of inositol polyphosphate 5-phosphatase 2 and prevents epidermal growth molecular pathway aberrations in uterine serous carcinoma by genome- factor (EGF)-induced EGF receptor degradation sustaining EGF-mediated wide analyses. J Natl Cancer Inst 2012;104:1503–13. signaling. FASEB J 2010;24:538–47. 4. Le Gallo M, O'Hara AJ, Rudd ML, Urick ME, Hansen NF, O'Neil NJ, et al. 26. Sablina AA, Chen W, Arroyo JD, Corral L, Hector M, Bulmer SE, et al. The Exome sequencing of serous endometrial tumors identifies recurrent tumor suppressor PP2A Abeta regulates the RalA GTPase. Cell somatic mutations in chromatin-remodeling and ubiquitin ligase complex 2007;129:969–82. genes. Nat Genet 2012;44:1310–5. 27. Sablina AA, Hector M, Colpaert N, Hahn WC. Identification of PP2A 5. Zhao S, Choi M, Overton JD, Bellone S, Roque DM, Cocco E, et al. complexes and pathways involved in cell transformation. Cancer Res Landscape of somatic single-nucleotide and copy-number mutations in 2010;70:10474–84. uterine serous carcinoma. Proc Natl Acad Sci U S A 2013;110:2916–21. 28. Chen W, Arroyo JD, Timmons JC, Possemato R, Hahn WC. Cancer- 6. McConechy MK, Ding J, Cheang MC, Wiegand KC, Senz J, Tone AA, et al. associated PP2A Aalpha subunits induce functional haploinsufficiency Use of mutation profiles to refine the classification of endometrial carci- and tumorigenicity. Cancer Res 2005;65:8183–92. nomas. J Pathol 2012;228:20–30. 29. Chen W, Possemato R, Campbell KT, Plattner CA, Pallas DC, Hahn WC. 7. McConechy MK, Anglesio MS, Kalloger SE, Yang W, Senz J, Chow C, et al. Identification of specific PP2A complexes involved in human cell trans- Subtype-specific mutation of PPP2R1A in endometrial and ovarian carci- formation. Cancer Cell 2004;5:127–36. nomas. J Pathol 2011;223:567–73. 30. Khanna A, Pimanda JE, Westermarck J. Cancerous inhibitor of protein 8. Shih IeM, Panuganti PK, Kuo KT, Mao TL, Kuhn E, Jones S, et al. Somatic phosphatase 2A, an emerging human oncoprotein and a potential cancer mutations of PPP2R1A in ovarian and uterine carcinomas. Am J Pathol therapy target. Cancer Res 2013;73:6548–53. 2011;178:1442–7. 31. Perrotti D, Neviani P. Protein phosphatase 2A: a target for anticancer 9. Nagendra DC, Burke J3rd, Maxwell GL, Risinger JI. PPP2R1A mutations are therapy. Lancet Oncol 2013;14:e229–8. common in the serous type of endometrial cancer. Mol Carcinog 2012; 32. Haesen D, Sents W, Lemaire K, Hoorne Y, Janssens V. The basic biology of 51:826–31. PP2A in hematologic cells and malignancies. Front Oncol 2014;4:347. 10. Janssens V, Goris J, Van Hoof C. PP2A: the expected tumor suppressor. Curr 33. Wang SS, Esplin ED, Li JL, Huang L, Gazdar A, Minna J, et al. Alterations of Opin Genet Dev 2005;15:34–41. the PPP2R1B gene in human lung and colon cancer. Science 1998;282: 11. Lambrecht C, Haesen D, Sents W, Ivanova E, Janssens V. Structure, 284–7. regulation, and pharmacological modulation of PP2A phosphatases. 34. Calin GA, di Iasio MG, Caprini E, Vorechovsky I, Natali PG, Sozzi G, et al. Methods Mol Biol 2013;1053:283–305. Low frequency of alterations of the alpha (PPP2R1A) and beta (PPP2R1B) 12. Janssens V, Rebollo A. The role and therapeutic potential of Ser/Thr isoforms of the subunit A of the serine-threonine phosphatase 2A in phosphatase PP2A in apoptotic signalling networks in human cancer cells. human neoplasms. Oncogene 2000;19:1191–5. Curr Mol Med 2012;12:268–87. 35. Ruediger R, Pham HT, Walter G. Disruption of protein phosphatase 2A 13. Groves MR, Hanlon N, Turowski P, Hemmings BA, Barford D. The structure subunit interaction in human cancers with mutations in the A alpha of the protein phosphatase 2A PR65/A subunit reveals the conformation of subunit gene. Oncogene 2001;20:10–5. its 15 tandemly repeated HEAT motifs. Cell 1999;96:99–110. 36. Ruediger R, Pham HT, Walter G. Alterations in protein phosphatase 2A 14. Ruediger R, Roeckel D, Fait J, Bergqvist A, Magnusson G, Walter G. subunit interaction in human carcinomas of the lung and colon with Identification of binding sites on the regulatory A subunit of protein mutations in the A beta subunit gene. Oncogene 2001;20:1892–9. phosphatase 2A for the catalytic C subunit and for tumor antigens of 37. Ruediger R, Ruiz J, Walter G. Human cancer-associated mutations in the simian virus 40 and polyomavirus. Mol Cell Biol 1992;12:4872–82. Aa subunit of protein phosphatase 2A increase lung cancer incidence in 15. Ruediger R, Hentz M, Fait J, Mumby M, Walter G. Molecular model of the A Aa knock-in and knockout mice. Mol Cell Biol 2011;31:3832–44. subunit of protein phosphatase 2A: interaction with other subunits and 38. Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, et al. The cBio tumor antigens. J Virol 1994;68:123–9. cancer genomics portal: an open platform for exploring multidimensional 16. Ruediger R, Fields K, Walter G. Binding specificity of protein phosphatase cancer genomics data. Cancer Discov 2012;2:401–4. 2A core enzyme for regulatory B subunits and T antigens. J Virol 39. Longin S, Zwaenepoel K, Louis JV, Dilworth S, Goris J, Janssens V. Selection 1999;73:839–42. of protein phosphatase 2A regulatory subunits is mediated by the 17. Shi Y. Serine/threonine phosphatases: mechanism through structure. Cell C terminus of the catalytic Subunit. J Biol Chem 2007;282:26971–80. 2009;139:468–84. 40. Houge G, Haesen D, Vissers LE, Mehta S, Parker MJ, Wright M, et al. B56d- 18. Wlodarchak N, Guo F, Satyshur KA, Jiang L, Jeffrey PD, Sun T, et al. related protein phosphatase 2A dysfunction identified in patients with þ Structure of the Ca2 -dependent PP2A heterotrimer and insights into Cdc6 intellectual disability. J Clin Invest 2015;125:3051–62. dephosphorylation. Cell Res 2013;23:931–46. 41. Seiler CY, Park JG, Sharma A, Hunter P, Surapaneni P, Sedillo C, et al. 19. Kamibayashi C, Lickteig RL, Estes R, Walter G, Mumby MC. Expression of DNASU plasmid and PSI:Biology-Materials repositories: resources to accel- the A subunit of protein phosphatase 2A and characterization of its erate biological research. Nucleic Acids Res 2014;42:D1253–60. interactions with the catalytic and regulatory subunits. J Biol Chem 42. Glatter T, Wepf A, Aebersold R, Gstaiger M. An integrated workflow for 1992;267:21864–72. charting the human interaction proteome: insights into the PP2A system. 20. McCright B, Virshup DM. Identification of a new family of protein Mol Syst Biol 2009;5:237. phosphatase 2A regulatory subunits. J Biol Chem 1995;270:26123–8. 43. Goudreault M, D'Ambrosio LM, Kean MJ, Mullin MJ, Larsen BG, 21. Dovega R, Tsutakawa S, Quistgaard EM, Anandapadamanaban M, Low€ C, Sanchez A, et al. A PP2A phosphatase high density interaction network Nordlund P. Structural and biochemical characterization of human PR70 identifies a novel striatin-interacting phosphatase and kinase complex in isolation and in complex with the scaffolding subunit of protein linkedtothecerebralcavernousmalformation3(CCM3)protein. phosphatase 2A. PLoS One 2014;9:e101846. Mol Cell Proteomics 2009;8:157–71. 22. Janssens V, Longin S, Goris J. PP2A holoenzyme assembly: in cauda 44. Herzog F, Kahraman A, Boehringer D, Mak R, Bracher A, Walzthoeni T, et al. venenum (the sting is in the tail). Trends Biochem Sci 2008;33:113–21. Structural probing of a protein phosphatase 2A network by chemical cross- 23. Ribeiro PS, Josue F, Wepf A, Wehr MC, Rinner O, Kelly G, et al. Combined linking and mass spectrometry. Science 2012;337:1348–52. functional genomic and proteomic approaches identify a PP2A complex as 45. Sacco F, Boldt K, Calderone A, Panni S, Paoluzi S, Castagnoli L, et al. a negative regulator of Hippo signaling. Mol Cell 2010;39:521–34. Combining affinity proteomics and network context to identify new 24. Couzens AL, Knight JD, Kean MJ, Teo G, Weiss A, Dunham WH, et al. phosphatase substrates and adapters in growth pathways. Front Genet Protein interaction network of the mammalian Hippo pathway 2014;5:115.

OF12 Cancer Res; 76(19) October 1, 2016 Cancer Research

Mechanism of Action of PP2A Aa Subunit Cancer Mutants

46. McConnell JL, Gomez RJ, McCorvey LR, Law BK, Wadzinski BE. Identiﬁ- phosphatase-2A catalytic subunit levels and consequent suppres- cation of a PP2A-interacting protein that functions as a negative regulator sion of inhibitory Raf-1 phosphorylation. J Biol Chem 2010;285: of phosphatase activity in the ATM/ATR signaling pathway. Oncogene 15076–87. 2007;26:6021–30. 52. Gordon J, Hwang J, Carrier KJ, Jones CA, Kern QL, Moreno CS, et al. Protein 47. Arroyo JD, Lee GM, Hahn WC. Liprin alpha1 interacts with PP2A B56gam- phosphatase 2a (PP2A) binds within the oligomerization domain of ma. Cell Cycle 2008;7:525–32. striatin and regulates the phosphorylation and activation of the mamma- 48. Liu L, Eisenman RN. Regulation of c-Myc protein abundance by a protein lian Ste20-Like kinase Mst3. BMC Biochem 2011;12:54. phosphatase 2A-glycogen synthase kinase 3b-negative feedback pathway. 53. Lin JL, Chen HC, Fang HI, Robinson D, Kung HJ, Shih HM. MST4, a new Genes Cancer 2012;3:23–36. Ste20-related kinase that mediates cell growth and transformation via 49. Hahn K, Miranda M, Francis VA, Vendrell J, Zorzano A, Teleman AA. PP2A modulating ERK pathway. Oncogene 2001;20:6559–69. regulatory subunit PP2A-B0 counteracts S6K phosphorylation. Cell Metab 54. Chen C, Shi Z, Zhang W, Chen M, He F, Zhang Z, et al. Striatins contain 2010;11:438–44. a noncanonical coiled coil that binds protein phosphatase 2A A subunit 50. Rocher G, Letourneux C, Lenormand P, Porteu F. Inhibition of B56- to form a 2:2 heterotetrameric core of striatin-interacting phosphatase containing protein phosphatase 2As by the early response gene IEX-1 leads and kinase (STRIPAK) complex. J Biol Chem 2014;289:9651–61. to control of Akt activity. J Biol Chem 2007;282:5468–77 55. Madsen CD, Hooper S, Tozluoglu M, Bruckbauer A, Fletcher G, Erler JT, 51. Kilili GK, Kyriakis JM. Mammalian Ste20-like kinase (Mst2) indirectly et al. STRIPAK components determine mode of cancer cell migration and supports Raf-1/ERK pathway activity via maintenance of protein metastasis. Nat Cell Biol 2015;17:68–80.

www.aacrjournals.org Cancer Res; 76(19) October 1, 2016 OF13

Recurrent PPP2R1A Mutations in Uterine Cancer Act through a Dominant-Negative Mechanism to Promote Malignant Cell Growth

Dorien Haesen, Layka Abbasi Asbagh, Rita Derua, et al.

Cancer Res Published OnlineFirst August 2, 2016.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-15-3342

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2016/08/02/0008-5472.CAN-15-3342.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2016/09/14/0008-5472.CAN-15-3342. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.