PAK4 Phosphorylates Fumarase and Blocks Tgfb

Published OnlineFirst January 25, 2019; DOI: 10.1158/0008-5472.CAN-18-2575

Cancer Metabolism and Chemical Biology Research

PAK4 Phosphorylates Fumarase and Blocks TGFb- Induced Cell Growth Arrest in Lung Cancer Cells Tao Chen1, Ting Wang2,3, Wenhua Liang1, Qin Zhao2, Qiujing Yu2, Chun-Min Ma2, Lingang Zhuo2, Dong Guo2, Ke Zheng2, Chengzhi Zhou1, Shupei Wei1, Wenhua Huang1, Juhong Jiang1, Jing Liu1, Shiyue Li1, Jianxing He1,Yuhui Jiang2, and Nanshan Zhong1

Abstract

The metabolic activity of fumarase (FH) participates in gene formation. Physiologically, FH Ser 46 phosphorylation pro- transcription linking to tumor cell growth. However, whether motes tumorigenesis through its suppressive effect on FH Thr this effect is implicated in lung cancer remains unclear. Here, 90 phosphorylation–mediated cell growth arrest in NSCLC we show TGFb induces p38-mediated FH phosphorylation at cells and correlates with poor prognosis in patients with lung Thr 90, which leads to a FH/CSL (also known as RBP-Jk)/p53 cancer. Our ﬁndings uncover an uncharacterized mechanism complex formation and FH accumulation at p21 promoter underlying the local effect of FH on TGFb-induced gene under concomitant activation of Notch signaling; in turn, transcription, on which the inhibitory effect from PAK4 pro- FH inhibits histone H3 Lys 36 demethylation and thereby motes tumorigenesis in lung cancer. promotes p21 transcription and cell growth arrest. In addition, FH is massively phosphorylated at the Ser 46 by PAK4 in Signiﬁcance: Fumarase counteracts CSL via its meta- non–small cell lung cancer (NSCLC) cells, and PAK4- bolic activity to facilitate TGFb-induced cell growth phosphorylated FH binds to 14-3-3, resulting in cytosolic arrest, an effect largely blocked by PAK4-mediated phos- detention of FH and prohibition of FH/CSL/p53 complex phorylation of fumarase.

Introduction factor 6 (TRAF6)-TGFb–associated kinase 1 (TAK1) is critical for p38 activation in response to TGFb (6, 9). Functionally, previous TGFb is a key regulator of cell fate during embryogenesis ﬁndings demonstrated p38 activation is essential for either TGFb- and has also been implicated in tumor metastasis as a potent induced cellular apoptosis or epithelial-to-mesenchymal transi- driver (1–3). Canonically, TGFb stimulation leads to activation of tion (EMT; refs. 10, 11), although how its multifaceted effects are signal transduction through transmembrane type I and type II coordinated remains elusive. serine/threonine kinase receptors (TbRI and TbRII, respectively), Crosstalk between Notch and TGFb signaling pathway has which phosphorylates Smad2/Smad3 and promotes their com- been implicated in various physiologic events. TGFb stimula- plex formation with Smad4 and launches their transcriptional tion is able to induce expression of Jagged1, which promotes activity (4, 5). In addition, TGFb is also able to initiate a number of EMT (12), and Notch activation is found to be indispensable non-Smad signaling pathways including p38 pathway, PI3K for TGFb-mediated EMT during cardiac development and pathway, and the Ras–ERK–MAPK pathway (6–8). It has been tumorigenesis (13). HDAC6 has been reported to be required known that the axis of ubiquitin ligase TNF-receptor-associated for activation of Notch signaling by TGFb1 during EMT in lung cancer (14). Referring to downstream transcriptional factor, 1State Key Laboratory of Respiratory Diseases; National Clinical Research Center NICD and Smad3 can form physical complex and cooperate to of Respiratory Diseases; Guangzhou Institute of Respiratory Health; First Afﬁl- drive the activation of synthetic promoters containing multi- iated Hospital of Guangzhou Medical University, Guangzhou Medical University, merized CSL- or Smad3-binding sites (15). In terms of p21, Guangzhou, Guangdong, P.R. China. 2The Institute of Cell Metabolism and p53 plays a positive role in TGFb-induced p21 expression and Disease, Shanghai Key Laboratory of Pancreatic Disease, Shanghai General cell growth arrest at normal condition (16). A recent study Hospital, School of Medicine, Shanghai Jiaotong University, Shanghai, P.R. China. shows that CSL, which is the crucial transcriptional factor 3Department of Pharmacology, College of Basic Medical Sciences, Tianjin Medical University, Tianjin, China. under Notch signaling, blocks cell senescence through its inhibitory effects on p53-mediated p21 transcription at the Note: Supplementary data for this article are available at Cancer Research epigenetic level (17); this observation implies the regulatory Online (http://cancerres.aacrjournals.org/). effect of CSL on p53 is a potential linkage between TGFb and T. Chen, T. Wang, and W. Liang contributed equally to this article. Notch signaling. Corresponding Authors: Yuhui Jiang, The Institute of Cell Metabolism and Emerging evidence in recent years indicate that metabolic Disease, Shanghai Key Laboratory of Pancreatic Disease, Shanghai 200080, enzymes, in addition to the basic role in metabolic pathways, China. Phone: 8602-1377-97983; E-mail: [email protected]; and Jianxing also vitally participate in the epigenetic regulation through related He, [email protected] metabolic activity (18). Fumarase (FH), which separately distri- doi: 10.1158/0008-5472.CAN-18-2575 butes in mitochondria and cytosol, is responsible for the revers- 2019 American Association for Cancer Research. ible hydration and dehydration of fumarate to malate in the

www.aacrjournals.org 1383

Chen et al.

tricarboxylic acid cycle (19). Fumarate has been known to com- Materials pete with a-ketoglutarate to bind to a-ketoglutarate– Antibodies that recognize FH (ab110286), PAK4 (ab62509), dependent dioxygenases that are involved in histone and DNA CSL (ab25949), KDM2A (ab31739), H3 (ab1791), H3K36me2 demethylation, and thus exerts inhibitory effects (20). FH defi- (ab9049), Lamin B1 (ab65986), Smad3 (ab28379), 14-3-3 zeta ciency causes elevation of cellular fumarate level, which has been were purchased from Abcam. Antibodies that recognize p53 implicated in the development of renal cancer due to its inhibitory (sc-126) were obtained from Santa Cruz Biotechnology. Antibo- effect on TET-regulated DNA demethylation (21). FH is also dies against FH (#32975), Flag (#35535), His (#T505), and GST reported to be importantly involved in DNA repair (22, 23) or (#T509) were obtained from Signalway Antibody. Antibodies gene transcription in glucose-deficient conditions (24), in which against Flag (F3165), GST (G1160), and His (SAB4301134) were chromatin-associated FH precisely regulates H3K36me2 at the purchased from Sigma. Antibodies against Phospho-MAPKAPK-2 region of DNA breaks or gene promoter by antagonizing a-KG– (#3044), Phospho-JunB (D3C6, #8053), p38 (D13E1, #8690), dependent demethylase KDM2A/B. Given the tight relationship Phospho-p38 (Thr180/Tyr182; D3F9, #4511), Phospho-Akt between the functional status of FH and epigenetic regulation, it (Ser473; D9E, #4060), Phospho-SMAD2 (Ser465/Ser467; E8F3R, would be meaningful to extensively understand the concrete role #18338), Cleaved Notch1 (D3B8, #4147), SMRT (D8D2L, of FH in this regard under distinct genetic contexts. #62370), b-actin (#4970) were purchased from Cell Signaling In this study, we first found that TGFb induces the interaction Technology. FITC-labeled antibody against BrdU (11-5071) was between FH and CSL in a manner dependent on Notch signaling purchased from eBioscience. activation and p38-mediated phosphorylation of FH. This inter- Rabbit polyclonal FH pThr-90 and FH pSer-46 antibodies was action results in the formation of FH–CSL–p53 complex, and generated by Signalway Antibody. A peptide containing pThr-90 facilitates the recruitment of FH at the promoter of p53-targeted or FH pSer-46 was injected into rabbits. The rabbit serum was gene p21. Through inhibition of KDM2A activity, fumarate locally collected and purified using an affinity column with nonpho- produced from promoter-associated FH leads to the increase of sphorylated FH Thr-90 and Ser-46 peptide to exclude the anti- H3K36me2 and hence p21 transcription, which eventually relives bodies for nonphosphorylated FH, followed by an affinity col- the negative effects of CSL on p53-mediated cell growth arrest. umn with phosphorylated FH Thr-90 and Ser-46 peptide to bind Moreover, we found that FH is largely phosphorylated by PAK4 at to and purify the FH pThr-90 and FH pSer-46 antibody. The FH Ser 46 in lung cancer cells, and 14-3-3 binds to PAK4 phosphor- pThr-90 and FH pSer-46 antibodies were then eluted and ylated-FH, thus retaining FH in the cytosol and blocking its effect concentrated. on CSL–p53 axis. SP600125 and SB203580 were purchased from Cell Signal- ing Technology. AKT inhibitor and U0126 were purchased from EMD Biosciences. PAK4 inhibitor PF-3758309 was obtained Materials and Methods from Selleck Chemicals. Anti-FLAG M2 beads, alkaline phos- Cell culture phatase, and purified His-PAK4 were purchased from Abcam. Human normal bronchial epithelial cell line HBE and human TGFb was obtained from R&D Systems. Compound C, mono- NSCLC cell lines A549, NCI-H520, HCC-827, and NCI-H358, ethyl fumarate, diethyl-malate, BrdU, EDTA-free Protease were obtained from Shanghai Cell Bank (Shanghai Biological Inhibitor Cocktail, and PhosSTOP were purchased from Sigma. Sciences, Chinese Academy of Sciences, Shanghai, China) that Hygromycin, puromycin, DNase-free RNase A, and propidium were originally obtained from the ATCC, authenticated by short iodide were purchased from EMD Biosciences. Other materials tandem repeat profiling within 6 months. Cells were maintained were described previously (23). in RPMI1640 supplemented with 10% FBS. Immunoprecipitation and immunoblotting analysis DNA constructs and mutagenesis Proteins were extracted from cultured cells using a modifi- The sequence of Flag-tag of FH in the pcDNA6 vector is located ed buffer (50 mmol/L Tris-HCl (pH 7.5), 1% Triton X-100, at the C-terminus of FH. Flag-FH (N) with deletion of 21 150 mmol/L NaCl, 1 mmol/L DTT, 0.5 mmol/L EDTA, and N-terminal amino acids was constructed by PCR reactions and protease inhibitor cocktail or phosphatase inhibitor cocktail), the DNA constructs and mutagenesis were described previous- followed by immunoprecipitation and immunoblotting with ly (23, 25). pGIPZ human FH shRNA was generated with the corresponding antibodies, as described previously (26). For 0 0 0 the oligonucleotides 5 -GGAATTTAGTGGTTATGTT-3 and 5 - immunoprecipitation of endogenous FH, FH antibody from 0 AAATTGATATAAGCATCCA -3 . pGIPZ human CSL shRNA was Abcam (ab110286) was used for immunoprecipitation (IP) and 0 generated with the oligonucleotide 5 -ATAATTTCGCATAG- antibody from Signalway Antibody (#32975) was used for immu- 0 0 0 CTTCC-3 and 5 -TCTGATTCATCATCATCCA-3 . pGIPZ human noblotting. Nuclear extract was collected using NE-PER Nuclear 0 p53 shRNAs were generated with the oligonucleotides 5 -GAAAC- and Cytoplasmic Extraction Reagents (Thermo Fisher Scientific) 0 0 0 CACTGGATGGAGAA-3 and 5 -TTATCAGCAGTATGATAGC-3 . according to the manufacturer's instructions. The protein concen- pGIPZ human 14-3-3 zeta shRNAs were generated with tration was determined using Bradford assay. Proteins from cell 0 0 0 the oligonucleotides 5 -ACGGTTCACATTCCATTAT-3 and 5 - lysates or nuclear extracts were separated by SDS-PAGE, trans- 0 AGAACACAGAGAAGTTAAG-3 . Human p21 siRNAs were ferred onto polyvinylidene difluoride membrane (Millipore Cor- 0 generated with the oligonucleotides sense 5 -GAUGGAACUUC- poration) and probed with the indicated antibodies. GACUUUGUUU-30; antisense: 50-ACAAAGUCGAAGUUCCA- 0 UCUU-3 . The pGIPZ controls were generated with control Recombinant protein purification 0 0 0 oligonucleotide 5 -GCTTCTAACACCGGAGGTCTT-3 or 5 - WT and mutant GST-FH were expressed in bacteria and puri- 0 GCCCGAAAGGGTTCCAGCTTA-3 . fied, as described previously (27).

1384 Cancer Res; 79(7) April 1, 2019 Cancer Research

PAK4 Blocks TGFb-Induced Cell Growth Arrest

Gene expression analysis Chromatin treatment in vitro We isolated total RNA using RNAzol RT (Molecular Research Isolation of chromatin extract from cultured cells with recon- Center) following the manufacturer's instructions. cDNA was stituted expression of WT or mutant rFH was performed as synthesized from 1 mg total RNA using iScript cDNA Synthesis described previously (23). Briefly, cells were washed with PBS Kit (Bio-Rad) and quantified mRNA levels by real-time qRT-PCR and resuspended in solution A [10 mmol/L HEPES (pH 7.9), using SYBR Green (Bio-Rad). We ran samples in technical tripli- 10 mmol/L KCl, 1.5 mmol/L MgCl2, 0.34 M sucrose, 10% glyc- cates and calculated relative mRNA levels normalized to actin erol, 1 mmol/L dithiothreitol, 10 mmol/L NaF, 1 mmol/L mRNA levels in the same samples. The qPCR primer sequences Na2VO3, and protease inhibitors]. Triton X-100 was added to a were listed as follows: p21-50-CGACTGTGATGCGCTAATGG (for- final concentration of 0.1%, and the cells were incubated for 5 ward), 50-GGCGTTTGGAGTGGTAGAAATC (reverse); GAPDH- minutes on ice, followed by low-speed centrifugation (4 minutes 50-AGGTGAAGGTCGGAGTCAAC (forward), 50-GACAAGCTT- at 1,300 g at 4C) to separate the cytoplasmic proteins from CCCGTTCTCAG (reverse). the nuclei. The isolated nuclei were then lysed in solution B (3 mmol/L EDTA, 0.2 mmol/L EGTA, 1 mmol/L DTT, and p38 treatment protease inhibitors). Insoluble chromatin was collected by cen- Purified WT and mutant GST-FH were incubated with p38 in trifugation. The resuspended chromatin extracts were incubated kinase assay buffer supplemented with 0.2 mmol/L AMP and cold with the various concentration of malate for 30 minutes, ATP in the presence or absence of 0.2 mCi/mL hot ATP (ICN followed by incubation with purified KDM2A in reaction buffer Biochemicals) for 20 minutes at 30C. After the reaction, p38 was (20 mmol/L Tris-HCl at pH 7.5, 150 mmol/L NaCl, 50 mmol/L removed by extensive washing with RIPA buffer and kinase assay [NH4]2Fe[SO4]2, 100 mmol/L a-ketoglutarate, 2 mmol/L Vc, and buffer. The GST-FH bound beads were recovered by centrifuga- 10 mmol/L phenylmethylsulfonylfluoride) for 3 hours. After tion. For kinase phosphorylation analyses, the GST-FH–bound reaction, the chromatin extracts were fixed and sonicated. The beads were subjected to SDS-PAGE and then autoradiography CHIP analysis with H3K36me2 antibody was performed using an after incubation with EN3HANCE (PerkinElmer). The GST-FH– Upstate Biotechnology kit. bound beads after p38 treatment in absence of hot ATP were further incubated with his-CSL for protein interaction assay as Transfection indicated. HBE, A549, and NCI-H520 cells were transfected with various plasmids and siRNA using Lipofectamine 3000 (Invitrogen) Chromatin Immunoprecipitation assay according to the vendor's instructions. A chromatin immunoprecipitation (ChIP) assay was performed using an Upstate Biotechnology kit as described pre- Cell proliferation assay viously (23). Quantitative real-time PCR was used to measure After treatment, cells were incubated with 50 mmol/L BrdU for 1 the amount of bound DNA, and the value of enrichment hour. Cells were collected by tryptase and centrifugation and fixed was calculated according to the relative amount of input and in 70% ethanol at 4C for 1 hour and subsequently incubated the ratio to IgG. The primers covering p53 DNA-binding with 2 N HCl/0.5% Triton X-100 for 30 minutes, 0.1 mol/L borate sites (p53RE) of human p21 gene promoter region [50- sodium for 2 minutes. After anti-BrdU-FITC (eBioscience) incu- GTGGCTCTGATTGGCTTTCTG-30 (forward), 50-CTGAAAA- bation and washing, BrdU incorporation rate was analyzed by CAGGCAGCCCAAG-30 (reverse)] were used for the real-time flow cytometry. Cell-cycle profile was measured by propidium PCR (28). iodide staining and analyzed by flow cytometry. The primers covering proximal promoter region of human p21 gene [50-GAAGTGCCCTCCTGCAGCACGCGAG-30 (forward) and Enzyme activity assay 50-CACCTCCTCTGAGTGCCTCGGTG-30 (reverse)] were used for Measurement of FH activity was performed as described pre- the real-time PCR. The primers covering promoter region of viously (23). Briefly, purified WT or mutant FH proteins human Smad7 gene [50-CCTCTGCTCGGCTGGTTCCACTGC-30 were added to an enzyme assay buffer (50 mmol/L malate and (forward) and 50-TAGAAACCCGATCTGTTGTTTGCG-30 (reverse)] 10 mmol/L potassium phosphate at pH 7.3) and the absorbance were used for the real-time PCR. at OD 240 nm was recorded.

Mass spectrometry analysis Fumarate measurement FH phosphorylation was analyzed by LC/MS-MS as Fumarate concentrations in cells were measured using the described previously (23). Briefly, Flag-FH–associated proteins Fumarate Detection Kit purchased from Abcam. from the IP assay were acetone-precipitated in vitro at 20C overnight and resuspended in 50 mmol/L ammonium bicar- IHC bonate buffer containing Rapigest (Waters Corp). The sample IHC was performed on paraffin-embedded sections of human washeatedto95C for 10 minutes and allowed to cool down lung cancer (611 lung cancer tissues). Formalin-fixed, paraffin- before 100 ng of sequencing-grade modified trypsin (Promega) embedded consecutive human lung cancer tissue sections was added. The digestion proceeded overnight at 37Candwas (3–5 mm) were deparaffinized and rehydrated. Antigen retrieval analyzed by LC/MS-MS using an Orbitrap-Elite mass spectrom- was performed by boiling tissue sections in 10 mmol/L citrate eter (Thermo Fisher Scientific). Proteins were identified by buffer (pH 6.0) in a microwave oven for 5 minutes. The activity of comparing the fragment spectra against those in the SwissProt endogenous peroxidase was blocked with 3% hydrogen peroxide protein database (EBI) using Mascot v.2.3 (Matrix Science) and in methanol for 10 minutes at room temperature. After washing, Sequest v.1.20 via Proteome Discoverer v.1.3 (Thermo Fisher nonspecific binding sites were blocked by incubating the slides Scientific) software. with 10% FBS/PBS for 30 minutes at room temperature. Sections

www.aacrjournals.org Cancer Res; 79(7) April 1, 2019 1385

Chen et al.

were subsequently incubated with antibodies against PAK4 basally binds to CSL, and their interaction was decreased by NICD (ab62509, 1:50) and FH pSer-46 (Signalway Antibody, 1:50) at expression (Fig. 1A, right). These data suggest that both TGFb and 4C overnight. After incubation with the primary antibodies, the Notch signaling are required for FH–CSL complex formation, sections were washed and incubated with secondary antibodies and the replacement of corepressor such as SMRT by NICD in and DAB staining reagent with GTVision Detection System/ CSL-related complex would facilitate FH–CSL interaction. Flow Mo&Rb Kit according to the manufacturer's instructions. After cytometry analyses showed that TGFb notably induced G0–G1 counterstain with hematoxylin and dehydration, the sections arrest and suppressed proliferation both in parent HBE (Supple- were mounted and imaged using the Leica microscope. Clinical mentary Fig. S1A) and NICD-overexpressed HBE (NI-HBE) cells information of 681 consecutive resected NSCLC cases between (Fig. 1B; Supplementary Fig. S1B). However, we found that TGFb- 2009 and 2010 in the first affiliated hospital of Guangzhou induced cell growth arrest was inhibited by FH depletion only in Medical University (Guangzhou, China) were collected. To illus- NI-HBE (Fig. 1B; Supplementary Fig. S1B), and this effect could trate the natural prognosis of patients without confounding be further reversed by CSL depletion (Fig. 1B; Supplementary effects from postrecurrent treatments, only resected NSCLCs Fig. S1B). These results reveal that FH is required for TGFb- (stage I–III) were eligible and disease-free survival (DFS) was induced growth arrest upon Notch activation and at this condi- selected as endpoint. The corresponding tissue array (with 681 tion exerts opposite effect to CSL. cases and a total of 23 holes as normal control) was subjected to TGFb-induced phosphorylation of Smad2, p38, and AKT IHC staining for FH pS46 expression. Immunoreactivity was in HBE (Supplementary Fig. S1C, left) or NI-HBE cells qualitatively evaluated according to the area of staining despite (Supplementary Fig. 1C, right) was not affected by FH depletion. the intensity to avoid artificial cut-off effect: the area of stained Coimmunoprecipitation analyses showed CIP treatment in pre- cancer cells was recorded as 0% (negative staining), > 0% (positive cipitates from Flag-FH–disrupted FH-CSL complex formation staining) of all cancer cells. Statistical analyses of the survival time (Fig. 1C), revealing FH interacts with CSL in a phosphoryla- as well as the illustration of survival curves were performed using tion-dependent manner. In addition, NI-HBE cells were pre- the Cox multivariate regression model in the SPSS 19 Software. treated with AMPK inhibitor Compound C, JNK inhibitor Statistical significance was set at P < 0.05. SP600125, and p38 inhibitor SB203580, respectively (Supple- mentary Fig. S1D), and TGFb-induced FH–CSL interaction was Mouse found to be exclusively blocked by SB203580 (Fig. 1D), suggest- Six-week-old male nu/nu mice were injected with 2 106 A549 ing TGFb-induced FH–CSL interaction is dependent on p38 cells in a volume of 150 mL of PBS/Matrigel [1/1 (v/v)]. Injections activity and raising the possibility that FH could be the substrate were made subcutaneously in the left and right flanks, respectively of p38. This assumption was verified by an in vitro protein at day 0. Tumor volume was measured using Vernier calipers phosphorylation assay, which showed purified activated p38 weekly. All animal experiments were performed in accordance could phosphorylate FH (Fig. 1E). Analysis of human FH amino with the regulations drafted by the Association for Assessment and acid sequence indicated Thr 90 is the potential p38-phosphory- Accreditation of Laboratory Animal Care in Shanghai and were lated residue as S/TP site, and further in vitro protein phosphor- approved by the East China Normal University Center for Animal ylation assay indicated FH T90A mutant was resistant to p38- Research. mediated FH phosphorylation (Fig. 1F). Subsequently, coimmunoprecipitation analyses showed FH Ethics approval and consent to participate T90A mutant lost CSL-binding ability in NI-HBE cells under All animal experiments were approved by the East China TGFb treatment (Fig. 1G). Consistently, a GST pull-down assay Normal University Center for Animal Research. The usage of showed that purified CSL can interact with WT FH, but not FH human tissues was approved by the Institutional Ethics Board T90A, in the presence of p38 (Fig. 1H) and a phosphor-mimic of the First Affiliated Hospital of Guangzhou Medical University form FH T90D (Supplementary Fig. S1E). These results suggest and conforms to the Helsinki Declaration and to local legislation. p38-mediated FH T90 phosphorylation is required for FH Written informed consent was obtained from the patients. binding to CSL. To investigate whether cellular location of FH affects its interaction with CSL, Flag-FH was immunoprecipi- Statistical analysis tated separately from nuclear and cytosolic fractions. As a Statistical analysis was conducted with the two-tailed unpaired result, CSL was largely enriched in precipitates in the nucleus Student t test. All data represent the mean SEM of three of NI-HBE cells stimulated with TGFb (Fig. 1I, left), whereas independent experiments. FH-90 phosphorylation was detected in either the nuclear or cytosolic lysates (Fig. 1I). In addition, the limited effect of CSL depletion on the amount of nuclear FH-90 phosphorylation Results (Supplementary Fig. S1F) suggests that the nuclear transloca- p38 mediates FH–CSL interaction induced by TGFb tion of p38-phosphorylated FH is independent of CSL. Fur- To determine whether FH is involved in the regulation of TGFb thermore, we depleted endogenous FH in NI-HBE cells and signaling under Notch activation, the potential relation between reconstituted the expression of RNAi-resistant rFH(N), which FH and CSL was examined in human bronchial epithelial cell line lost the mitochondrial localization, as 21 amino-terminal HBE cells. The reciprocal immunoprecipitation analysis indicated amino acid residues were deleted (23). As a result, FH(N) TGFb stimulation dramatically increased the interaction between showed comparable FH-T90 phosphorylation to FH of full- FH and CSL in HBE cells expressed with Notch intracellular length under TGFb stimulation (Supplementary Fig. S1G), and domain (NICD), while TGFb treatment or NICD expression alone FH-T90 phosphorylation was not detected in the mitochondria could not (Fig. 1A). FH was not associated with SMRT (Fig. 1A, extracts (Supplementary Fig. S1H), suggesting mitochondria left), a known transcriptional corepressor of CSL (29), which localization is dispensable for FH-T90 phosphorylation.

1386 Cancer Res; 79(7) April 1, 2019 Cancer Research

PAK4 Blocks TGFb-Induced Cell Growth Arrest

Figure 1. p38 mediates FH–CSL interaction induced by TGFb. A, HBE cells overexpressing Flag-FH and with or without NICD were treated or untreated with TGFb (10 ng/mL) for 2 hours. Whole cellular extracts were subjected to immunoprecipitation with an anti-FH antibody (left) or anti-CSL antibody (right). B, HBE cells overexpressing NICD (NI-HBE cells) were transfected with or without FH shRNA or/and CSL shRNA. Cells were treated or untreated with TGFb (10 ng/mL) for 48 hours. The proportion of cells in cell cycle was examined by FACS analyses. The values are presented as mean SEM (n ¼ 3 independent experiments). , P < 0.01 (Student t test) between indicated groups. C, NI-HBE cells and Flag–FH were treated or untreated with TGFb (10 ng/mL) for 1 hour. The immunoprecipitates were treated with CIP (10 units) and analyzed by immunoblotting. D, NI-HBE cells overexpressing Flag-FH were pretreated with Compound C (10 mmol/L), SP600125 (20 mmol/L), or SB203580 (10 mmol/L) for 1 hour (left), before being treated or untreated with TGFb (10 ng/mL) for 2 hours. E, In vitro phosphorylation analyses were performed by mixing the purified p38 with GST-FH proteins in the presence of [g-32P]ATP. F, The indicated purified GST-FH protein was mixed with or without purified p38. G, NI-HBE cells expressed with indicated Flag–FH were treated or untreated with TGFb (10 ng/mL) for 2 hours. H, The indicated purified GST-FH protein was mixed with purified His-CSL protein with or without purified p38. GST pull-down analyses were performed. I, NI-HBE cells overexpressing Flag-FH were treated or untreated with TGFb (10 ng/mL) for 2 hours; nuclear (left) or cytosolic (right) extracts were subjected to immunoprecipitation. In A–I, immunoblotting analyses were performed using the indicated antibodies; data represent one of three experiments.

www.aacrjournals.org Cancer Res; 79(7) April 1, 2019 1387

Chen et al.

FH forms a complex with CSL-p53 and promotes p21 promotes TGFb-induced growth arrest. p53 has been known to transcription through local fumarate production be critical for TGFb-induced growth arrest (16). Coimmuno- Next, FH was depleted in NI-HBE cells and reconstituted with precipitation analyses showed NICD overexpression induced expression of shRNA-resistant WT rFH(N) and rFH(N) T90A CSL–p53 complex formation (Supplementary Fig. S2C, left), (Supplementary Fig. S2A). Consequently, rFH(N) T90A expres- which was not affected by TGFb treatment or FH depletion sion signiﬁcantly inhibited TGFb-induced G0–G1 arrest (Fig. 2A) (Supplementary Fig. S2C, right). In contrast, TGFb largely induced and suppressed proliferation (Supplementary Fig. S2B) compared FH–p53 interaction in NICD-expressed HBE cells (Supplemen- with its WT counterpart, indicating FH-T90 phosphorylation tary Fig. S2D) in a CSL (Fig. 2B, left) and FH-T90 phosphorylation

Figure 2. FH forms complex with CSL-p53 and promotes p21 transcription through local fumarate production. A, NI-HBE cells overexpressing indicated Flag-FH were treated or untreated with TGFb (10 ng/mL) for 48 hours. Cell-cycle profile was examined by FACS analyses. B, NI-HBE cells with or without depleted CSL (left) or overexpressing indicated Flag–FH (right) were treated or untreated with TGFb (10 ng/mL) for 2 hours. Whole cellular extracts were subjected to immunoprecipitation with indicated antibodies. C, NI-HBE cells with or without p53 depletion, and with or without CSL and/or p53 depletion were treated or untreated with TGFb (10 ng/mL) for 48 hours. Immunoblotting analyses were performed (left). Cell-cycle profile was examined by FACS analyses (middle and right). D, HBE or NI-HBE cells with or without FH depletion were treated or untreated with TGFb (10 ng/mL) for 12 hours. The mRNA level of p21 was analyzed by real-time PCR. E, NI-HBE cells with or without FH depletion and/or CSL depletion were treated or untreated with TGFb (10 ng/mL) for 12 hours. The mRNA level of p21 was analyzed by real-time PCR. F, NI-HBE cells overexpressing indicated rFH(N) or with or without p53 depletion were treated or untreated with TGFb (10 ng/mL) for 12 hours. The mRNA level of p21 was analyzed by real-time PCR. G, NI-HBE cells with or without p53 depletion, and with or without CSL and/or p53 depletion were overexpressed with indicated Flag-FH; cells were treated or untreated with TGFb (10 ng/mL) for 6 hours. ChIP analyses were performed by using antibodies of anti-Flag. The y-axis shows the value normalized to the input. H, NI-HBE cells with FH depletion and reconstituted expression of indicated Flag-rFH (N) were treated or untreated with TGFb (10 ng/mL). ChIP analyses were performed with an anti-Flag antibody. The y-axis shows the value normalized to the input. I, NI-HBE cells with FH depletion and reconstituted expression of indicated Flag–rFH(N) were treated or untreated with TGFb (10 ng/mL). Cells were treated with indicated concentration of monoethyl fumarate. The mRNA level of p21 was analyzed by real-time PCR. J, NI-HBE cells with FH depletion and reconstituted expression of indicated Flag-rFH(N) were treated or untreated with TGFb (10 ng/mL). Cells were treated with indicated concentration of monoethyl fumarate. Cell-cycle profile was examined by FACS analyses. In B and C, immunoblotting analyses were performed using the indicated antibodies. In A–J, data represent one of three experiments. In A and C–J, the values are presented as mean SEM (n ¼ 3 independent experiments). , P < 0.01 (Student t test) between the indicated groups.

1388 Cancer Res; 79(7) April 1, 2019 Cancer Research

PAK4 Blocks TGFb-Induced Cell Growth Arrest

(Fig. 2B, right) dependent manner. Functional analyses indicated and NCI-H520 cells (Supplementary Fig. S3A). Accordingly, A549 that p53 depletion largely inhibited TGFb-induced growth arrest (Supplementary Fig. S3B) and NCI-H520 (Supplementary Fig. in NI-HBE cells, and this effect was not changed by CSL depletion S3C) cells did not show significant enhancement of growth arrest (Fig. 2C; Supplementary Fig. S2E), demonstrating the dominative after TGFb treatment, unless with CSL depletion. Given this role of p53 here. Along with aforementioned results, these data difference, we set forth to look for the upstream component that delineate a proposed model in terms of the functional relation- would be specific for the regulation of FH–CSL interaction in ship among FH, CSL, and p53 under TGFb and Notch signaling cancer cells. A panel of inhibitors was collected and used to block (Supplementary Fig. S2F). Depletion of p21, a primary p53 target activities of relevant kinases that are involved in tumorigenesis. gene relevant to growth arrest, apparently inhibited TGFb- Intriguingly, we found that the treatment of the p21 protein induced growth arrest in either HBE cells (Supplementary (Cdc42/Rac)-activated kinase 4, PAK4 inhibitor PF-3758309 Fig. S2G) or NI-HBE cells (Supplementary Fig. S2H). In line with largely enhanced FH–CSL interaction in A549 (Fig. 3A) and results shown in Fig. 1B, FH depletion notably decreased TGFb- NCI-H520 (Supplementary Fig. S3D) cells under TGFb stimulus. induced p21 transcription in NI-HBE cells, but not in parental PAK4 belongs to serine/threonine protein kinase and its dysre- HBE cells (Fig. 2D), and CSL depletion reversed these effects in gulation has been implicated in lung cancer (33). Further coim- NI-HBE cells (Fig. 2E). Meanwhile, TGFb dramatically promoted munoprecipitation analysis showed that PAK4 interacted with FH p21 transcription in WT FH-, but not FH T90A-expressed NI-HBE instead of CSL (Supplementary Fig. S3E), implying that FH would cells (Fig. 2F). Moreover, ChIP analysis showed WT FH-, but not also be a substrate of PAK4. As revealed in the protein phosphor- FH T90A, was efficiently accumulated at p21 promoter region in ylation assay, purified activated PAK4 phosphorylated FH and the TGFb-treated NI-HBE cells, which was blocked by either CSL or phosphorylation was recognized by an antibody against Ser p53 depletion (Fig. 2G). These results reveal FH is recruited to phosphorylation (Fig. 3B). Mass spectrometric analysis indicated promoter by CSL-p53 and thereby promotes p21 transcription that FH Ser46 underwent phosphorylation in A549 cells (Fig. 3C). and growth arrest by counteracting the suppressive effects of CSL Mutagenesis analyses showed that FH S46A mutant, which did on p53. not affect p38-mediated FH T90 phosphorylation (Supplemen- FH is able to regulate nonmetabolic events through its meta- tary Fig. S3F), robustly abolished PAK4-mediated FH phosphor- bolic function (22, 23). Expression of the catalytically inactive ylation, as demonstrated by autography and immunoblotting mutant Flag-tagged FH(N) R233H showed TGFb induced com- analysis by using a specific FH-S46 phosphorylation antibody parable FH-90 phosphorylation and FH-CSL complex to that in (Fig. 3D). WT FH-expressed NI-HBE cells (Supplementary Fig. S2I). In line Subsequently, immunoblotting analysis showed that PAK4- with this, TGFb readily induced accumulation of FH(N) R233H as mediated FH-S46 phosphorylation is insensitive to TGFb well as WT FH(N) at p21 promoter regions (Fig. 2H). However, stimulation in A549 cells (Supplementary Fig. S3G, left), and either reconstituted expression of rFH(N) R233H or rFH(N) T90A hardly detected in NI-HBE cells (Supplementary Fig. S3G, notably inhibited TGFb-induced p21 expression (Fig. 2I) and right). Meanwhile, PF-3758309 treatment, which blocked growth arrest (Fig. 2J) in NI-HBE cells. Of note, FH(N) T90A, FH-S46 phosphorylation, did not affect TGFb-induced FH which lost the enrichment at p21 promoter (Fig. 2H), showed a T90 phosphorylation in A549 cells (Supplementary Fig. comparable catalytic activity to WT FH(N) and FH(N) T90D S3H). Immunoprecipitation analyses indicated that TGFb stim- (Supplementary Fig. S2J). These data suggest both FH activity ulation failed to significantly promote the binding of phos- and promoter location are required for FH-CSL-p53–mediated phor-mimic mutant FH S46D to CSL in NI-HBE cells (Supple- downstream events. To further determine the effect of FH activity mentary Fig. S3I). In contrast, FH S46A displayed an increased on TGFb-induced growth arrest, exogenous monoethyl fumarate bindingtoCSLandp53inTGFb-treated A549 and NCI-520 was added into the cell-cultured medium with various amounts, cells (Fig. 3E; Supplementary Fig. S3J). Functionally, TGFb which accordingly resulted in elevation of intracellular fumarate treatment largely promoted cell growth arrest in A549 cells concentration (Supplementary Fig. S2K). Consequently, FH (N) with PAK4 inhibitor treatment (Supplementary Fig. S3K), or R233H or FH(N) T90A-repressed p21 transcription (Fig. 2I) and with reconstituted expression of rFH(N) S46A, but not rFH(N) growth arrest (Fig. 2J; Supplementary Fig. S2L) in TGFb-treated S46A/T90A (Fig. 3F; Supplementary Figs. S3L and S3M). These NI-HBE cells were partially reversed by exogenous fumarate in a data strongly suggest FH-S46 phosphorylation by PAK4 dose-dependent manner, while these rescue effects were blocked impedes FH–CSL interaction, and thus releases the repressive by p53 depletion (Supplementary Fig. S2M and S2N). On the effects of CSL on TGFb-induced growth arrest. In addition, other hand, addition of exogenous malate that led to a moderate TGFb-induced growth arrest was inhibited by p53 depletion increase of fumarate levels (Supplementary Fig. S2O) did not in A549 cells with FH S46A expression (Fig. 3G; Supplementary soundly affect TGFb-induced p21 transcription (Supplementary Fig. S3N), and TGFb treatment displayed limited effects on cell Fig. S2P) and growth arrest (Supplementary Fig. S2Q) in rFH growth in p53-null H1299 cells, which was not changed by CSL R233H- or rFH T90A-expressed NI-HBE cells. Together, these data depletion (Supplementary Fig. S3O), supporting the pivotal indirectly demonstrate promoter-associated fumarate production role of p53 in TGFb-induced cell growth arrest. Consistently, activity of FH is indispensable for p53-p21–mediated down- TGFb readily stimulated p21 transcription (Fig. 3H, left) as well stream event under TGFb signaling. as FH promoter accumulation (Fig. 3I) in FH S46A-expressed, but not WT FH- or FH S46A/T90A-expressed A549 cells, which PAK4 phosphorylates FH and blocks FH–CSL interaction were repressed by p53 depletion [Fig. 3H (right) and I]. In terms The negative effect of TGFb signaling on cell growth is always of the effect of metabolic activity of FH, we found exogenous impeded during tumor development (30–32). In contrast with fumarate (Supplementary Fig. S3P) partially promoted p21 the observation in HBE cells (Fig. 1A), TGFb stimulation only transcription (Fig. 3J) and growth arrest (Fig. 3K) in A549 cells weakly promotes FH–CSL interaction in NICD-expressed A549 treated with TGFb.

www.aacrjournals.org Cancer Res; 79(7) April 1, 2019 1389

Chen et al.

Figure 3. PAK4 phosphorylates FH and blocks FH–CSL interaction. A, A549 cells overexpressing NICD (NI-A549 cells) were transfected with Flag-FH and were pretreated with PF-3758309 (10 mmol/L), U0126 (20 mmol/L), or LY294002 (10 mmol/L) for 1 hour (left), before being treated or untreated with TGFb (10 ng/mL) for 2 hours. Immunoprecipitation analysis was performed using anti-Flag antibody. B, In vitro phosphorylation analyses were performed by mixing the purified PAK4 with GST-FH proteins in the presence of [g-32P]ATP. C, NI-A549 cells were overexpressed with Flag-FH. Immunoprecipitation analysis was performed using anti-Flag antibody, and the extracts were analyzed by mass spectrometry. The results of a mass spectrometric analysis of a tryptic fragment at m/z 510.70 (mass error, 1.13 ppm) matched those of the doubly charged peptide 111-2-120, suggesting that S46 was phosphorylated. The Sequest score for this match was Xcorr ¼ 2.38. The Mascot score was 288.46. D, The indicated purified GST-FH proteins were mixed with or without purified PAK4. E, NI-A549 cells expressed with indicated Flag-FH were treated or untreated with TGFb (10 ng/mL) for 2 hours. Immunoprecipitation analysis was performed using anti-Flag antibody. F, NI-A549 cells overexpressing indicated Flag-FH were treated or untreated with TGFb (10 ng/mL) for 48 hours. Cell-cycle profile was examined by FACS analyses. G, NI-A549 cells with or without p53 depletion were overexpressed with indicated Flag-FH. Cells were treated or untreated with TGFb (10 ng/mL) for 48 hours. Cell-cycle profile was examined by FACS analyses. H, NI-A549 cells with or without CSL depletion (left) or p53 depletion (right) were expressed with indicated Flag-FH. Cells were treated or untreated with TGFb (10 ng/mL) for 48 hours. The mRNA level of p21 was analyzed by real-time PCR. I, NI-A549 cells with or without CSL depletion (left) or p53 depletion (right) were overexpressed with indicated Flag-FH. Cells were treated or untreated with TGFb (10 ng/mL) for 6 hours. ChIP analyses were performed with an anti-Flag antibody. The y-axis shows the value normalized to the input. J, A549 cells incubated with the indicated concentrations of monoethyl-fumarate were treated or untreated with TGFb (10 ng/mL) for 12 hours. The mRNA level of p21 was analyzed by real-time PCR. K, A549 cells incubated with the indicated concentrations of monoethyl-fumarate were treated or untreated with TGFb (10 ng/mL) for 48 hours. Cell-cycle profile was examined by FACS analyses. In A, B, D,andE, immunoblotting analyses were performed using the indicated antibodies. In A, B and D–K, data represent one of three experiments. In F–K, the values are presented as mean SEM (n ¼ 3 independent experiments). , P < 0.05; , P < 0.01 (Student t test) between the indicated groups.

PAK4-phosphorylated FH binds to 14-3-3 tance of FH-T90 phosphorylation for FH binding to CSL, these Sequential in vitro kinase assays indicated p38- and PAK4- data are in contrast with the result that FH S46A–CSL interaction mediated FH phosphorylation were not mutually affected (Sup- was largely enhanced in TGFb-treated NSCLC cells (Fig. 3E), and plementary Fig. S4A). In A549 cells, TGFb induced FH-T90 implies that the inhibitory effect of PAK4 on FH–CSL interaction phosphorylation in a p38 activity–dependent manner (Supple- is attributed to an additional mechanism. Cellular fraction anal- mentary Fig. S4B), which was not signiﬁcantly changed under ysis indicated that FH-S46 phosphorylation showed an exclusive FH-S46A expression (Supplementary Fig. S4C). Given the impor- localization in the cytosol of A549 cells (Fig. 4A) and expression of

1390 Cancer Res; 79(7) April 1, 2019 Cancer Research

PAK4 Blocks TGFb-Induced Cell Growth Arrest

Figure 4. PAK4-phosphorylated FH binds to 14-3-3. A, NI-A549 cells overexpressing indicated Flag-FH were treated or untreated with TGFb (10 ng/mL) for 2 hours; nuclear (left) or cytosolic (right) extracts were subjected to immunoprecipitation with an anti-Flag antibody. B, NI-A549 cells overexpressing indicated Flag-FH were treated or untreated with TGFb (10 ng/mL) for 2 hours; nuclear (left) or cytosolic (right) were subjected to immunoprecipitation with an anti-Flag antibody. C, NI-A549 cells overexpressing indicated Flag-FH. Cellular extracts were collected and incubated with or without GST–14-3-3. D, NI-A549 cells were transfected with 14-3-3 zeta shRNA and were treated or untreated with TGFb (10 ng/mL) for 2 hours. Nuclear extracts were collected. E, A cartoon showing 14-3-3 binds to Ser46-phosphorylated FH and leads to its sequestration in the cytosol. In A–D, immunoblotting analyses were performed using the indicated antibodies. Data represent one of three experiments.

FH S46A led to a dramatic increase of FH–CSL interaction in the to protein phospho-Ser/Thr–containing sequence motifs (34). nucleus, but not cytosol, under TGFb stimulation. In contrast, the Scansite analysis showed that FH Ser46 is located in the sequence nuclear level of FH-T90 phosphorylation was significantly as a potential binding site motif for 14-3-3 zeta. We wonder increased in FH S46A–expressed A549 cells with TGFb stimulus, whether 14-3-3 zeta is responsible for the cytosolic arrest of FH. with a concomitant drop in the cytosol (Fig. 4A). Further cellular Ectopic and endogenous coimmunoprecipitation analysis show- fraction analysis in rFH (N) S46A-expressed A549 cells indicated ed that 14-3-3 zeta was associated with FH in the cytosol (Fig. 4B, that FH with FH-T90 phosphorylation accounts for around 20% left) but not nucleus (Fig. 4B, right) of A549 cells. In contrast, FH of total nuclear FH (Supplementary Fig. S4D, left), a large appor- S46A failed to interact with 14-3-3 zeta (Fig. 4B, left). Consistent tion of which, interacted with CSL (Supplementary Fig. S4D, with the limited FH-S46 phosphorylation levels, NI-HBE cells right). In addition, rFH(N) R233H and rFH T90A showed a showed a limited complex formation of FH-14-3-3 zeta (Supple- comparable nuclear accumulation to WT FH (Supplementary Fig. mentary Fig. S4F). The pull-down assay showed purified GST-14- S4E, left), while rFH(N) S46A and rFH S46A/T90A with nuclear 3-3 zeta associated with WT FH, but not FH S46A, in A549 cells location increased by around 23% compared with that of WT FH (Fig. 4C). Consistent with results shown in Supplementary (Supplementary Fig. S4E, right) in A549 cells treated with TGFb. Fig. S4E, immunoblotting of nuclear extracts indicated expression These data suggest that FH T90 or S46 phosphorylation accounts of 14-3-3 zeta shRNA (Supplementary Fig. S4G) only partially for a small proportion of total FH, and FH pS46 exerts negative enhanced the nuclear translocation of total FH, while leading to a effects on the nuclear translocation of T90-phosphorylated FH significant increase of nuclear FH-T90 phosphorylation in A549 and the subsequent FH–CSL interaction. Of note, the limited cells treated with TGFb (Fig. 4D). These data indicate that FH pS46 effect of FH pS46 on nuclear translocation of total FH implies that exerts critical effects on nuclear translocation of FH-T90 phos- there remain unidentified mechanisms involved in FH nuclear phorylation and the subsequent FH–CSL interaction (Fig. 4E). In translocation. addition, 14-3-3 zeta has been reported to destabilize p53 (30); 14-3-3 proteins are critically implicated in signal transduction this implies that TGFb-induced p53-FH S46A interaction might be during various biological processes through the specific binding attributed to the potential change of p53 protein levels resulting

www.aacrjournals.org Cancer Res; 79(7) April 1, 2019 1391

Chen et al.

from FH S46A expression. However, immunoblotting analysis S46A/R233H-suppressed H3K36me2 at p21 promoter (Fig. 5G) indicated that reconstituted expression of rFH S46A in A549 cells induced by TGFb, in which FH S46 phosphorylation did not did not result in significant change of p53 protein levels (Sup- affect FH metabolic activity (Supplementary Fig. S5M). At the plementary Fig. S4H). same condition, addition of high dosage of exogenous malate did not exert significant effect on H3K36me2 accumulation at p21 Local fumarate maintains H3K36Me2 through inhibition of promoter (Supplementary Fig. S5N), p21 transcription (Supple- KDM2 activity mentary Fig. S5O), or cell growth (Supplementary Fig. S5P). Fumarate can regulate histone methylation via its antagonized Measurement of fumarate indicated high dosage of exogenous effect on a-ketoglutarate (aKG)-dependent demethylases (22). malate only led to moderate increase of intracellular fumarate TGFb stimulation in NI-HBE cells increased H3K36me2 and levels (Supplementary Fig. S5Q). To determine the direct effects H3K4me3, but not H3K27me2 or H3K9me2 levels at the p21 of fumarate on KDM2A activity, chromatin was collected from promoter region covering p53RE (Supplementary Fig. S5A), while A549 cells expressing rFH(N) S46A or rFH(N) S46A/R233H. only H3K36me2 accumulation, which is associated with tran- In vitro assay showed that KDM2A abolished TGFb-induced scriptional activation (35), was apparently blocked by rFH T90A- H3K36me2 on the p21 promoter region, which was partially or rFH R233H expression in TGFb-treated NI-HBE cells (Supple- inhibited by the malate incubation (Fig. 5H, 1st–4th lanes). In mentary Fig. S5B). We further expressed H3K36R in NI-HBE cells contrast, malate failed to affect H3K36me2 from FH R233H- and found H3K36R was readily accumulated at p21 promoter expressed A549 cells (Fig. 5H, 5th–7th lanes). These data indicate region (Supplementary Fig. S5C). p53, CSL, and p38 protein FH catalytic activity for fumarate production is critical for TGFb- levels remained at steady levels under H3K36R expression (Sup- induced growth arrest through inhibition of KDM2A-mediated plementary Fig. S5D) and H3K36R expression largely repressed demethylation at p21 promoter. TGFb-induced p21 transcription in NI-HBE cells (Supplementary Fig. S5E) or rFH(N) S46A-expressed A549 cells (Supplementary FH S46 phosphorylation is required for tumorigenesis Fig. S5F). In addition, rFH(N) T90A expression did not show a Seahorse analysis indicated only A549 cells expressed with rFH significant effect on TGFb-induced transcription of smad7, of R233H, but not rFH T90A, rFH S46A, or rFH S46A/T90A, dis- which the gene promoter region contains binding site of SBE played a lower mitochondrial oxygen consumption rate (OCR), (Smad-binding element), but not p53RE, compared with WT compared with WT counterpart (Supplementary Fig. S6A), counterpart (Supplementary Fig. S5G). Meanwhile, ChIP analysis excluding that FH-S46 or FH-T90 phosphorylation might affect indicated FH accumulation was hardly to be detected at the cell growth through its potential impact on mitochondrial activ- promoter region of smad7, and H3K36me2 deposition on smad7 ity. To further analyze the physiologic relevance of FH Ser46 promoter was not changed by TGFb treatment (Supplementary phosphorylation, four lung cancer cell lines NCI-H520, A549, Fig. S5H), suggesting FH-regulated H3K36me2 accumulation at HCC-827, and NCI-H358 were collected. As a result, each cancer promoter would be prone to promoter of genes that cover p53RE. cell line exhibited distinct FH Ser46 phosphorylation levels, In A549 cells, rFH(N) S46A expression largely promoted which was in line with the PAK4 protein levels and inversely H3K36me2 accumulation at p21 promoter (Fig. 5A) and p21 related to the amount of FH–CSL complex (Fig. 6A). Functional transcription (Supplementary Fig. S5I) induced by TGFb com- analysis indicated that FH-S46 phosphorylation and PAK4 levels pared with WT rFH(N), while this failed to be recapitulated in cells are positively correlated with cell growth in these cell lines under expressing rFH(N) S46A/R233H or S46A/T90A (Fig. 5B). In TGFb stimulation (Fig. 6B). As shown in Supplementary Fig. S6B, addition, expression of rFH(N) S46A/R233H or S46A/T90A had expression of rFH S46A, but not rFH S46A/T90A, largely blocked no additional effects on H3K36R-repressed p21 transcription cell growth in NCI-H520 cells under TGFb treatment. To further (Fig. 5C). These data indicate promoter-associated FH activity is determine the functional impact of PAK4-mediated FH-S46 phos- required for H3K36me2 regulation at promoter region. phorylation on tumorigenesis, A549 cells with reconstituted KDM2 is the primary histone demethylase responsible for expression of WT rFH, rFH T90A, rFH S46A, or rFH S46A/T90A H3K36me2 demethylation (22). ChIP analysis showed that were subcutaneously injected into athymic nude mice. Compared KDM2A was constantly detected at the p21 promoter region in with WT counterpart, A549 cells expressing rFH T90A did not NI-HBE cells or rFH(N) S46A-expressed A549 cells (Supplemen- show remarkable change of tumor growth (Supplementary Fig. tary Fig. S5J). KDM2A depletion notably inhibited rFH(N) S6C), while expression rFH S46A, but not rFH S46A/T90A, S46A/T90A- and rFH(N) S46A/R233H-repressed H3K36me2 notably inhibited tumor growth (Fig. 6C), revealing the critical accumulation at p21 promoter (Fig. 5B), p21 transcription effect of FH S46 phosphorylation. Consistently, 14-3-3 depletion (Fig. 5D), and growth arrest (Fig. 5E; Supplementary Fig. S5K) or PF-3758309 treatment in A549 cells significantly repressed in TGFb-treated A549 cells. These data indicate that FH promo- tumor growth (Supplementary Fig. S6D). High activity of PAK4 tes TGFb-induced growth arrest through inhibition of KDM2A- correlates with poor prognosis for patients with lung cancer (33). mediated demethylation of H3K36me2. Furthermore, exogenous To further determine the clinical relevance of our finding that monoethyl fumarate was added into A549 cells with KDM2A FH-S46 phosphorylation promotes tumor cell growth, we per- overexpression that led to a decrease in overall H3K36me2 levels formed IHC analyses to examine the levels of PAK4 and FH-S46 (Supplementary Fig. S5L). Ectopic expression of KDM2A robustly phosphorylation in serial sections of human lung tumor speci- suppressed H3K36me2 at p21 promoter, which was blocked by mens (Fig. 6D, left and middle). Staining quantification showed exogenous fumarate in a dosage-dependent manner (Fig. 5F). that FH-S46 phosphorylation levels were positively correlated However, if exogenous a-KG was concomitantly added into with PAK4 expression levels (Fig. 6D, right). No case in the normal cells with fumarate, the inhibitory effect from KDM2A over- controls showed significant FH pS46 signals (Supplementary Fig. expression was significantly reversed (Fig. 5F). Moreover, exoge- S6E). As illustrated in Table 1, the expression rate of FH pS46 was nous fumarate partially reversed FH(N) S46A/T90A or FH(N) higher in female than in male, and was increased as the N stage

1392 Cancer Res; 79(7) April 1, 2019 Cancer Research

PAK4 Blocks TGFb-Induced Cell Growth Arrest

Figure 5. Local fumarate maintains H3K36me2 through inhibition of KDM2 activity. A, NI-A549 cells with FH depletion and reconstituted expression of indicated Flag-rFH (N) were treated or untreated with TGFb (10 ng/mL) for 6 hours. B, NI-A549 cells with FH depletion and reconstituted expression of indicated Flag–rFH (N) and with or without KDM2A depletion were treated or untreated with TGFb (10 ng/mL) for 6 hours. C, NI-A549 cells with FH depletion and reconstituted expression of indicated Flag-rFH(N) and indicated H3. Cells were treated or untreated with TGFb (10 ng/mL) for 12 hours. Immunoblotting analyses were performed using the indicated antibodies (left). The mRNA level of p21 was analyzed by real-time PCR (right). D, NI-A549 cells with FH depletion and reconstituted expression of indicated Flag-rFH(N) were transfected with or without KDM2A shRNA. Cells were treated or untreated with TGFb (10 ng/mL) for 12 hours. The mRNA level of p21 was analyzed by real-time PCR. E, NI-A549 cells with FH depletion and reconstituted expression of indicated Flag-rFH(N) were transfected with or without KDM2A shRNA. Cells were treated or untreated with TGFb (10 ng/mL) for 48 hours. Cell-cycle proﬁle was examined by FACS analyses. F, NI-A549 cells with FH depletion and reconstituted rFH(N) S46A were overexpressed with or without KDM2A. Cells added with indicated concentration of monoethyl fumarate and octyl-aKG were treated or untreated with TGFb (10 ng/mL) for 12 hours. The mRNA level of p21 was analyzed by real-time PCR. G, NI-A549 cells with FH depletion and reconstituted expression of indicated Flag-rFH(N) were treated with indicated concentration of monoethyl fumarate. Cells were treated or untreated with TGFb (10 ng/mL) for 6 hours. H, NI-A549 cells with FH depletion and reconstituted expression of indicated Flag-rFH(N) were treated with TGFb (10 ng/mL) for 6 hours. Chromatin extracts were collected and mixed with or without the indicated concentration of malate for 30 minutes and followed by incubation with KDM2A, and then chromatin extracts were ﬁxed and used for ChIP analyses (left). Chromatin extracts subjected to immunoblotting analysis with an indicated antibody (right). In A, B, G,andH, ChIP analyses with indicated antibodies were performed. The y-axis shows the value normalized to the input. The values represent mean SEM (n ¼ 3 independent experiments). , P < 0.01 (Student t test) between indicated groups. In A–H, data represent one of three experiments.

and TNM stage increased. However, the proportions were similar 1.040; 95% CI, 0.772–1.400, P ¼ 0.797). However, patients with between different age, smoking history, histology, and tumor size. FH pS46 expression were potentially associated with inferior DFS In survival analyses, the censored rate was 37.6% with a median in patients with squamous cell carcinoma (HR 1.668; 95% CI, follow-up time of 5.06 (range 0.05–6.99) years. After adjusting for 0.863–3.222; P ¼ 0.128; Fig. 6E, left). and in those with lymph sex, age, histology, and stage, we found that the expression status node metastases (HR 1.435; 95% CI, 0.956–2.154; P ¼ of FH pS46 did not stratify prognosis in overall population (HR 0.082; Fig. 6E, right).

www.aacrjournals.org Cancer Res; 79(7) April 1, 2019 1393

Chen et al.

Figure 6. FH S46 phosphorylation is required for tumorigenesis. A, NCI-H520 A549, NCI-H520, HCC-827, and NCI-H358 cells overexpressing NICD were treated with TGFb (10 ng/mL) for 48 hours. Whole cellular extracts were subjected to immunoprecipitation with an anti-FH antibody. B, NCI-H520 A549, NCI-H520, HCC-827, and NCI-H358 cells overexpressing NICD were treated or untreated with TGFb (10 ng/mL) for 48 hours. Cell-cycle proﬁle was examined by FACS analyses (left). Cellular proliferation rate was examined by BrdU incorporation assay and FACS analyses (right). The values are presented as mean SEM (n ¼ 3 independent experiments). , P < 0.05; , P < 0.01 (Student t test) between indicated groups. C, A total of 2 106 A549 cells with overexpression of NICD and FH depletion and reconstituted expression of indicated rFH(N) were subcutaneously injected into the athymic nude mice. Representative tumor xenografts are shown (left). Tumor volumes were measured using length (a) and width (b) and calculated using the following equation: V ¼ ab2/2. Data represent the means SEM (n ¼ 7, right). D, IHC staining with anti-PAK4 and anti-FH pSer46 antibodies was performed on 611 human lung tumor specimens. Representative photos of staining in squamous cell carcinoma are shown. Scale bar, 20 mm (left and right). Relation between categorized variables was examined by x2 test (right). E, The survival times for patients with negative (black curve) versus positive (red curve) FH pS46 were compared. The Kaplan–Meier method and log-rank tests indicate the signiﬁcance level of the association of FH pS46 (P ¼ 0.01182) with inferior DFS in patients with squamous cell carcinoma (HR, 1.668; 95% CI, 0.863–3.222; P ¼ 0.128) and in patients with lymph node metastases (HR, 1.435; 95% CI, 0.956–2.154; P ¼ 0.082). F, A schematic model representing integration of TGFb-p38-FH and NICD-CSL signaling and its negative regulation in tumor cells. TGFb induces p38 activation, which can phosphorylate FH at T90. Upon Notch activation, nuclear NICD promotes the interaction between CSL and p38-phosphorylated FH; this in turn promotes FH/CSL/p53/Smad complex formation and the enrichment of FH on p21 promoter region. Eventually, the local production of fumarate by FH inhibits KDM2A-mediated histone H3 demethylation and promotes p21 expression responsible for growth arrest. On the other hand, PAK4 is able to phosphorylate FH and impedes FH–CSL interaction. In normal cells, cells are prone to growth arrest under TGFb treatment, while upregulated PAK4 activity prevents growth arrest induced by TGFb treatment in cancer cells.

1394 Cancer Res; 79(7) April 1, 2019 Cancer Research

PAK4 Blocks TGFb-Induced Cell Growth Arrest

Table 1. Summary of patient characteristics and their correlation with FH pS46 expression FH pS46 expression Characteristics N (%) Negative (%) Positive (%) P Total 681 (100) 33.9 66.1 Sex 0.025 Male 397 (58.4) 37.6 62.4 Female 283 (41.6) 28.8 71.2 Age 0.144 <70 551 (80.9) 32.5 67.5 70 130 (19.1) 40.2 59.8 Smoking history 0.231 No 405 (63.4) 31.8 68.2 Yes 234 (36.6) 36.9 63.1 Histology 0.651 Squamous 150 (22.1) 37.5 62.5 Adenocarcinoma 479 (70.5) 33 67 Others 50 (7.4) 34 66 Tumor size (cm) 0.243 <3 307 (45.7) 36.3 63.7 3–5 212 (31.5) 33 67 5–7 101 (15.0) 31.3 68.7 7 52 (7.7) 29.8 70.2 N stage 0.073 0 410 (63.1) 36.2 63.8 1 90 (13.8) 30.5 69.5 2 150 (23.1) 28.1 71.9 TNM stage 0.020 Ia 143 (22.0) 39.2 60.8 Ib 202 (31.1) 36.8 63.2 IIa 105 (16.2) 31.9 68.1 IIb 36 (5.5) 18.8 81.3 IIIa 161 (24.8) 29.5 70.5 IIIb 3 (0.5) 0 100

Discussion study and it has become critical to develop the novel experimental method in future to precisely detect the amount of local meta- The physiologic consequence of TGFb signaling is compli- bolites and to better clarify the local metabolic effects of FH under cated by the concrete context and its crosstalk with other the specific context. Our results indicate that CSL upon Notch signaling pathways. A recent study showed CSL, the central activation is shown to block the transcriptional activity p53 under effector under Notch signaling pathway, can interact with p53 TGFb signaling, which is in line with the previous finding indi- and thereby block p53-mediated cell senescence (17). Given cating the inhibitory effects of CSL on p53-mediated transcrip- the critical role of p53 in the TGFb-induced cell growth arrest, tion (17). The interaction between CSL and T90-phoshorylated this finding provides a potential regulatory linkage between FH promotes the formation of the protein complex containing TGFb and Notch signaling, and prompts us to further investi- FH/CSL/p53, which is the prerequisite for FH accumulation and gate the underlying mechanism.Inthisstudy,wefoundthat local effect at the promoter of targeted gene; Of note, it has been TGFb induces FH Thr 90 phosphorylation by p38. Upon Notch known that Notch activation is accompanied with the alteration activation, nuclear NICD promotes the interaction between of CSL-associated components (36), thus, this might be critical for CSL and p38-phosphorylated FH and thus FH/CSL/p53/Smad the accessibility of CSL binding for T90-phoshorylated FH. Con- complex formation; this facilitates FH recruitment to p53- sistently, the essential role of Notch–CSL activation for FH targed p21 promoter, where FH inhibits KDM2A-mediated involvement in TGFb-p53–p21 signaling is supported by results demethylation of H3K36me2 through local production of showing that exclusive FH–CSL interaction in the nucleus appears fumarate; accumulation of H3K36me2 relieves the inhibitory when both signaling pathways are triggered. Hence, it seems that effect of CSL on p53 and promotes TGFb-induced cell growth the additively negative control of CSL by FH is able to make gene arrest (Fig. 6F). These results elucidate a novel mechanism of transcription regulated in a more precise manner on the signaling H3K36me2 regulation by p38/FH signaling axis, and demon- crosstalk. However, it remains to be further investigated whether strate that the metabolic effect of FH on transcriptional regu- this effect is also dependent on Notch activation under other lation is importantly implicated in cellular response to TGFb signaling context. signaling. Besides the inhibitory effects on cell growth, TGFb is known to Previous studies indicated that FH localized in the nucleus paradoxically drive tumor metastasis during cancer develop- displayed a local effect on histone methylation (23). Consistently ment (2). It has been a hot issue to decipher how the functional here, we found that promoter-associated FH facilitates p53-medi- role of TGFb is converted. In contrast with observations in HBE ated transcription through inhibition of H3K36me2 demethyla- cells, we found that TGFb-induced FH–CSL interaction and cell tion (Fig. 6F), in which FH-catalyzed metabolic reaction in the growth arrest are notably suppressed in lung cancer cells, which direction from fumarate to malate is shown to be crucial. How- was reversed by the PAK4 inhibition. Overexpression or hyper- ever, the supporting evidence in this regard remains indirect in this activation of PAK4 is recorded in human lung cancer (33). In

www.aacrjournals.org Cancer Res; 79(7) April 1, 2019 1395

Chen et al.

accordance, our study indicates that PAK4 is highly expressed in Acquisition of data (provided animals, acquired and managed patients, lung cancer cells and clinical samples. Further analyses show that provided facilities, etc.): T. Chen, T. Wang, W. Liang, Q. Zhao, K. Zheng, PAK4 can phosphorylate FH at Ser 46; and this phosphorylation C. Zhou, S. Wei, J. Jiang, S. Li, J. He Analysis and interpretation of data (e.g., statistical analysis, biostatistics, sequesters FH at cytosol through its binding to 14-3-3; in turn, FH computational analysis): T. Chen, W. Liang, Q. Zhao, D. Guo, C. Zhou, J. He Ser 46 phosphorylation blocks nuclear function of FH relevant to Writing, review, and/or revision of the manuscript: Y. Jiang, T. Chen, T. Wang, TGFb-induced cell growth arrest (Fig. 6F). Collectively, these W. Liang, Q. Yu results suggest that the aberrant levels of PAK4 would be an Administrative, technical, or material support (i.e., reporting or organizing important strategy undertook by cancer cells to avoid negative data, constructing databases): Q. Zhao, Q. Yu, C.-M. Ma, L. Zhuo, K. Zheng, effects of TGFb on cell growth, and hence to indirectly potentiate C. Zhou, S. Wei, J. Liu, S. Li, J. He Study supervision: Y. Jiang, J. He, N. Zhong tumor metastasis under TGFb signaling. Given the important physiologic effects of FH phosphorylation by PAK4, it could be Acknowledgments predicted that FH expression levels and activity would be tightly We thank D.-L. Li at the East China Normal University for mice study related to tumorigenesis in lung cancer cells with high expression assistance, and L.-J. Liao at the East China Normal University for assistance levels of PAK. of mass spectrometry analysis. This work was supported by National Nature The regulation of metabolic enzymes is fundamentally Science Foundation of China 81773006 (to Y. Jiang), Shanghai Committee linked to various physiologic events in addition to the direct of Science and Technology 16QA1403200 (to Y. Jiang), Shanghai Municipal Education Commission-Gaofeng Clinical Medicine Grant 20161319 effect on metabolism (37, 38). The local effect of FH on TGFb- (to Y. Jiang), Program for Eastern Young Scholar at Shanghai Institutions fi induced gene transcription exempli es the subtle interaction of Higher Learning (to Y. Jiang), The First Affiliated Hospital of Guangzhou between metabolic pathway and cellular growth signaling. Medical University, the Open Project of State Key Lab of Respiratory Disease Meanwhile, the inhibitory effect of PAK4 on p38/FH axis Grant SKLRD2016OP006 (to Y. Jiang), SKLRDQN201702 (to T. Chen), uncovers a novel mechanism of spatial regulation of FH activ- National Natural Science Foundation of China 81490534 (to N. Zhong), ity. With respect to tumorigenesis, the dysregulation of PAK4 81871893 (to W. Liang), 81501996 (to W. Liang), Training of Excellent fi fi Youth in Shanghai Health System 2017YQ056 (to T. Wang), Guangdong activity is identi ed here as a tumor-speci c way to escape from Provincial Science and Technology Planning Project 2018A030313525 (to T. thenegativeeffectsofTGFb, which potentially provides a Chen), the First Affiliated Hospital of Guangzhou Medical University, the molecular basis for improving the clinical treatment against "Young Elite Talents Program," and the "Science and Technology Support tumors with upregulated PAK4 activity. Projects 201520-gyfyy" (to T. Chen).

Disclosure of Potential Conflicts of Interest The costs of publication of this article were defrayed in part by the No potential conflicts of interest were disclosed. payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate Authors' Contributions this fact. Conception and design: Y. Jiang, T. Chen, W. Liang, J. He, N. Zhong Development of methodology: T. Chen, T. Wang, W. Liang, Q. Yu, C. Zhou, Received August 17, 2018; revised December 6, 2018; accepted January 23, W. Huang, J. He, N. Zhong 2019; published first January 25, 2019.

References 1. Pickup MW, Owens P, Moses HL. TGF-b, bone morphogenetic protein, 11. Horowitz JC, Lee DY, Waghray M, Keshamouni VG, Thomas PE, Zhang and activin signaling and the tumor microenvironment. Cold Spring H, et al. Activation of the pro-survival phosphatidylinositol 3-kinase/AKT Harb Perspect Biol 2017;9. doi: 10.1101/cshperspect.a022285. pathway by transforming growth factor-beta1 in mesenchymal cells is 2. Zhang Y, Alexander PB, Wang XF. TGF-beta family signaling in the control mediated by p38 MAPK-dependent induction of an autocrine growth of cell proliferation and survival. Cold Spring Harb Perspect Biol 2017;9. factor. J Biol Chem 2004;279:1359–67. doi: 10.1101/cshperspect.a022145. 12. Zavadil J, Cermak L, Soto-Nieves N, Bottinger EP. Integration of TGF-beta/ 3. Colice GL, Stukel TA, Dain B. Laryngeal complications of prolonged Smad and Jagged1/Notch signalling in epithelial-to-mesenchymal transi- intubation. Chest 1989;96:877–84. tion. EMBO J 2004;23:1155–65. 4. Xu P, Lin X, Feng XH. Posttranslational regulation of smads. Cold Spring 13. Timmerman LA, Grego-Bessa J, Raya A, Bertran E, Perez-Pomares JM, Harb Perspect Biol 2016;8. doi: 10.1101/cshperspect.a022087. Diez J, et al. Notch promotes epithelial-mesenchymal transition during 5. Heldin CH, Moustakas A. Role of Smads in TGFbeta signaling. Cell Tissue cardiac development and oncogenic transformation. Genes Dev 2004; Res 2012;347:21–36. 18:99–115. 6. Sorrentino A, Thakur N, Grimsby S, Marcusson A, von Bulow V, Schuster 14. Deskin B, Lasky J, Zhuang Y, Shan B. Requirement of HDAC6 for N, et al. The type I TGF-beta receptor engages TRAF6 to activate TAK1 in a activation of Notch1 by TGF-beta1. Sci Rep 2016;6:31086. receptor kinase-independent manner. Nat Cell Biol 2008;10:1199–207. 15. Blokzijl A, Dahlqvist C, Reissmann E, Falk A, Moliner A, Lendahl U, et al. 7. Bakin AV, Tomlinson AK, Bhowmick NA, Moses HL, Arteaga CL. Cross-talk between the Notch and TGF-beta signaling pathways mediated Phosphatidylinositol 3-kinase function is required for transforming by interaction of the Notch intracellular domain with Smad3. J Cell Biol growth factor beta-mediated epithelial to mesenchymal transition and 2003;163:723–8. cell migration. J Biol Chem 2000;275:36803–10. 16. Cordenonsi M, Dupont S, Maretto S, Insinga A, Imbriano C, 8. Holm TM, Habashi JP, Doyle JJ, Bedja D, Chen Y, van Erp C, et al. Piccolo S. Links between tumor suppressors: p53 is required for Noncanonical TGFbeta signaling contributes to aortic aneurysm progres- TGF-beta gene responses by cooperating with Smads. Cell 2003;113: sion in Marfan syndrome mice. Science 2011;332:358–61. 301–14. 9. Yamashita M, Fatyol K, Jin C, Wang X, Liu Z, Zhang YE. TRAF6 mediates 17. Procopio MG, Laszlo C, Al Labban D, Kim DE, Bordignon P, Jo SH, et al. Smad-independent activation of JNK and p38 by TGF-beta. Mol Cell 2008; Combined CSL and p53 downregulation promotes cancer-associated 31:918–24. ﬁbroblast activation. Nat Cell Biol 2015;17:1193–204. 10. Heldin CH, Landstrom M, Moustakas A. Mechanism of TGF-beta 18. Boukouris AE, Zervopoulos SD, Michelakis ED. Metabolic enzymes signaling to growth arrest, apoptosis, and epithelial-mesenchymal transi- moonlighting in the nucleus: metabolic regulation of gene transcription. tion. Curr Opin Cell Biol 2009;21:166–76. Trends Biochem Sci 2016;41:712–30.

1396 Cancer Res; 79(7) April 1, 2019 Cancer Research

PAK4 Blocks TGFb-Induced Cell Growth Arrest

19. Kornberg HL, Krebs HA. Synthesis of cell constituents from C2-units by a 29. Kao HY, Ordentlich P, Koyano-Nakagawa N, Tang Z, Downes M, Kintner modified tricarboxylic acid cycle. Nature 1957;179:988–91. CR, et al. A histone deacetylase corepressor complex regulates the Notch 20. Xiao M, Yang H, Xu W, Ma S, Lin H, Zhu H, et al. Inhibition of alpha-KG- signal transduction pathway. Genes Dev 1998;12:2269–77. dependent histone and DNA demethylases by fumarate and succinate that 30. Xu J, Acharya S, Sahin O, Zhang Q, Saito Y, Yao J, et al. 14–3–3zeta turns are accumulated in mutations of FH and SDH tumor suppressors. Genes TGF-beta's function from tumor suppressor to metastasis promoter in Dev 2012;26:1326–38. breast cancer by contextual changes of Smad partners from p53 to Gli2. 21. Sciacovelli M, Goncalves E, Johnson TI, Zecchini VR, da Costa AS, Gaude Cancer Cell 2015;27:177–92. E, et al. Fumarate is an epigenetic modifier that elicits epithelial-to- 31. Levy L, Hill CS. Smad4 dependency defines two classes of transforming mesenchymal transition. Nature 2016;537:544–7. growth factor {beta} (TGF-{beta}) target genes and distinguishes TGF- 22. Tsukada Y, Fang J, Erdjument-Bromage H, Warren ME, Borchers CH, {beta}-induced epithelial-mesenchymal transition from its antiprolifera- Tempst P, et al. Histone demethylation by a family of JmjC domain- tive and migratory responses. Mol Cell Biol 2005;25:8108–25. containing proteins. Nature 2006;439:811–6. 32. David CJ, Huang YH, Chen M, Su J, Zou Y, Bardeesy N, et al. TGF-beta 23. Jiang Y, Qian X, Shen J, Wang Y, Li X, Liu R, et al. Local generation of tumor suppression through a lethal EMT. Cell 2016;164:1015–30. fumarate promotes DNA repair through inhibition of histone H3 demeth- 33. Cai S, Ye Z, Wang X, Pan Y, Weng Y, Lao S, et al. Overexpression of P21- ylation. Nat Cell Biol 2015;17:1158–68. activated kinase 4 is associated with poor prognosis in non-small cell lung 24. Wang T, Yu Q, Li J, Hu B, Zhao Q, Ma C, et al. O-GlcNAcylation of cancer and promotes migration and invasion. J Exp Clin Cancer Res 2015; fumarase maintains tumour growth under glucose deficiency. Nat Cell Biol 34:48. 2017;19:833–43. 34. Obsilova V, Kopecka M, Kosek D, Kacirova M, Kylarova S, Rezabkova 25. Zheng Y, Li X, Qian X, Wang Y, Lee JH, Xia Y, et al. Secreted and O- L, et al. Mechanisms of the 14–3–3 protein function: regulation of GlcNAcylated MIF binds to the human EGF receptor and inhibits its protein function through conformational modulation. Physiol Res activation. Nat Cell Biol 2015;17:1348–55. 2014;63 Suppl 1;S155–64. 26. Lu Z, Liu D, Hornia A, Devonish W, Pagano M, Foster DA. Activation of 35. Balbach ST, Orkin SH. An Achilles' Heel for MLL-rearranged leukemias: protein kinase C triggers its ubiquitination and degradation. Mol Cell Biol writers and readers of H3 lysine 36 dimethylation. Cancer Discov 2016;6: 1998;18:839–45. 700–2. 27. Xia Y, Wang J, Liu TJ, Yung WK, Hunter T, Lu Z. c-Jun downregulation by 36. Collu GM, Hidalgo-Sastre A, Brennan K. Wnt-Notch signalling crosstalk in HDAC3-dependent transcriptional repression promotes osmotic stress- development and disease. Cell Mol Life Sci 2014;71:3553–67. induced cell apoptosis. Mol Cell 2007;25:219–32. 37. Krall AS, Christofk HR. Cancer: a metabolic metamorphosis. Nature 2013; 28. Hayakawa J, Mittal S, Wang Y, Korkmaz KS, Adamson E, English C, et al. 496:38–40. Identification of promoters bound by c-Jun/ATF2 during rapid large-scale 38. Gut P, Verdin E. The nexus of chromatin regulation and intermediary gene activation following genotoxic stress. Mol Cell 2004;16:521–35. metabolism. Nature 2013;502:489–98.

www.aacrjournals.org Cancer Res; 79(7) April 1, 2019 1397

PAK4 Phosphorylates Fumarase and Blocks TGFβ-Induced Cell Growth Arrest in Lung Cancer Cells

Tao Chen, Ting Wang, Wenhua Liang, et al.

Cancer Res 2019;79:1383-1397. Published OnlineFirst January 25, 2019.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-18-2575

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2019/01/25/0008-5472.CAN-18-2575.DC1

Cited articles This article cites 35 articles, 11 of which you can access for free at: http://cancerres.aacrjournals.org/content/79/7/1383.full#ref-list-1

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 Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/79/7/1383. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.