Histone Acetyltransferase PCAF Is Required for Hedgehog– Gli-Dependent Transcription and Cancer Cell Proliferation

Published OnlineFirst August 13, 2013; DOI: 10.1158/0008-5472.CAN-12-4660

Cancer Tumor and Stem Cell Biology Research

Martina Malatesta1,2, Cornelia Steinhauer1,2, Faizaan Mohammad1,2, Deo P. Pandey1,2, Massimo Squatrito4, and Kristian Helin1,2,3

Abstract The Hedgehog (Hh) signaling pathway plays an important role in embryonic patterning and development of many tissues and organs as well as in maintaining and repairing mature tissues in adults. Uncontrolled activation of the Hh–Gli pathway has been implicated in developmental abnormalities as well as in several cancers, including brain tumors like medulloblastoma and glioblastoma. Inhibition of aberrant Hh–Gli signaling has, thus, emerged as an attractive approach for anticancer therapy; however, the mechanisms that mediate Hh–Gli signaling in vertebrates remain poorly understood. Here, we show that the histone acetyltransferase PCAF/KAT2B is an important factor of the Hh pathway. Speciﬁcally, we show that PCAF depletion impairs Hh activity and reduces expression of Hh target genes. Consequently, PCAF downregulation in medulloblastoma and glioblastoma cells leads to decreased proliferation and increased apoptosis. In addition, we found that PCAF interacts with GLI1, the downstream effector in the Hh–Gli pathway, and that PCAF or GLI1 loss reduces the levels of H3K9 acetylation on Hh target gene promoters. Finally, we observed that PCAF silencing reduces the tumor-forming potential of neural stem cells in vivo. In summary, our study identiﬁed the acetyltransferase PCAF as a positive cofactor of the Hh–Gli signaling pathway, leading us to propose PCAF as a candidate therapeutic target for the treatment of patients with medulloblastoma and glioblastoma. Cancer Res; 73(20); 1–11. 2013 AACR.

Introduction by Hh signaling, suggesting a regulation of the pathway The evolutionary conserved Hedgehog (Hh) signaling path- through both a positive and a negative feedback (8). way plays an important role in development, proliferation, and The importance of the Hh pathway in tumorigenesis was fi stem cell maintenance (1, 2). In agreement with such role, rst discovered in patients of Gorlin syndrome. This rare PTCH deregulation of the Hh pathway leads to several developmental pathology is caused by an inactivating mutation in , syndromes and tumors of different tissues (3, 4). In mammals, which leads to the development of tumors like basal cell Hh signaling is initiated when the secreted Hh proteins bind to carcinoma, medulloblastoma, and rhabdomyosarcoma (9). In and inhibit the transmembrane receptor PTCH. The interac- addition, inappropriate activation of Hh signaling has been tion between Hh and PTCH releases Smoothened (SMO), a shown to lead to the development of tumors in the lung, – second transmembrane protein, which in turn induces the gastrointestinal tract, and pancreas (10 12). GLI1 fi downstream components of the Hh signaling pathway and was originally isolated as a highly ampli ed gene in fi leads to the activation of the glioma-associated oncogene (GLI) human malignant glioma and subsequently found ampli ed transcription factor family (5). When activated, the GLI pro- in other tumor types, including liposarcoma, rhabdomyo- GLI1 teins induce the expression of genes that regulate multiple sarcoma, and osteosarcoma (9, 13, 14). is a key regu- cellular functions such as cell-cycle progression, proliferation, lator of glioma growth and of cancer stem cell self-renewal. and apoptosis (6, 7). In addition, GLI1 and PTCH are activated Moreover studies in transgenic mice have shown that ectop- ic expression of Gli1 is sufficienttoinducebasalcellcar- cinoma (15–17). Despite the importance of the Hh pathway in development and disease, the molecular mechanisms by which the Hh signal Authors' Affiliations: 1Biotech Research and Innovation Centre (BRIC) and 2Centre for Epigenetics, University of Copenhagen; 3The Danish Stem leads to the activation of GLI-regulated transcription is still not Cell Center (Danstem), Copenhagen, Denmark; and 4F-BBVA Cancer Cell completely understood. Here, we show that transcriptional Biology Programme, Centro Nacional de Investigaciones Oncologicas (CNIO), Madrid, Spain activation by the Hh pathway requires the histone acetyltransferase (HAT) PCAF. Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). Materials and Methods Corresponding Author: Kristian Helin, BRIC, University of Copenhagen, Ole Maaløes Vej 5, DK-2200 Copenhagen N. Phone: 45-35-32-56-66; Fax: Cell culture and reagents 45-35-32-56-69; E-mail: [email protected] Human cell lines (U87, U118, T98G, and Daoy) were doi: 10.1158/0008-5472.CAN-12-4660 purchased from the American Type Culture Collection 2013 American Association for Cancer Research. (ATCC) and used at low passage numbers. A short tandem

www.aacrjournals.org OF1

Malatesta et al.

repeat profile was done by the ATTC, and the cells tested entire statistical analysis was conducted using the statisti- negative for mycoplasm. 293-ShhN and NIH3T3 cells were cal software 'R'. kindlyprovidedbyDr.J.Taipale(KarolinskaInstitute,Stock- holm, Sweden) and grown as previously described (18). Cells RNA extraction and gene expression analysis were maintained at 37 Cand5%CO2 in Dulbecco's Modified Total RNA was isolated using the RNAeasy Mini Kit (Qiagen) Eagle's Medium (DMEM) supplemented with 100 mg/mL according to the manufacturer's instructions. cDNA was syn- penicillin, 100 mg/mL streptomycin, and 10% FBS (Gibco; thesized using the TaqMan Reverse Transcription Kit (Applied U87, U118, T98G, DAOY, and 293-ShhN) or 10% newborn calf Biosystems). Quantitative PCR (qPCR) was conducted using serum (NIH3T3). GBM543 neurospheres were isolated from SYBR Green PCR Master Mix (Applied Biosystems) on an ABI patients suffering from glioblastoma (GBM) and propagated Prism 7300 Real-Time PCR system (Applied Biosystems) or on as previously described (19). Ink4a-Arf-nullneuralstemcells a LightCycler 480 System (Roche Applied Science), using the (NSC) were isolated and grown as previously described (20). LightCycler 480 SYBR Green I Master Mix (Roche Applied Viral transductions were conducted using pLKO vectors. Science) according to the manufacturer's instructions. Error The different target sequences are available in Supplemen- bars represent SD of three PCR amplifications for each sample. tary Table S1. Cells were transduced with lentiviral particles Similar results were obtained in at least three independent for 16 hours, and selected with 2 mg/mL puromycin (Invi- experiments. The qPCR primers are available in Supplemen- trogen) 48 hours after transduction. Treatments using SAG tary Table S2. (ALX-270-426-M001 Alexis), cyclopamine (C4116 SIGMA), or anacardic acid (AA; G5173 SIGMA) were carried out 24 hours Immunoblotting and immunoprecipitation after cell plating or as indicated. To prepare whole-cell extracts for immunoblotting analysis, For cell proliferation assays, cells were seeded at two dif- cells were lysed in high salt buffer S300P (50 mmol/L Tris–HCl, ferent densities into 96-well plates (Nunc#161093) and three 300 mmol/L NaCl, 0.5% Igepal, 1 mmol/L EDTA, 1 mmol/L images per well were acquired every hour over a period of three DTT, 1 mmol/L phenylmethylsulfonylfluoride (PMSF), 1 mg/mL to five days using the Incucyte Zoom (Essen Biosciences). leupeptin, and 1 mg/mL aprotinin). For immunoprecipitations, Image analysis was conducted using the Incucyte Zoom soft- protein A/G-agarose beads (GE Healthcare) were pre-coupled ware package. O/N with the indicated antibodies. Equal amounts of protein lysates (S300P buffer) were used for each immunoprecipita- siRNA screening and luciferase assay tion. Immunoprecipitates were eluted from beads and ana- The screening was done on a customized siRNA library lyzed by immunoblotting with the indicated antibodies. The against 17 different mouse acetyltransferases, with each gene identity and the suppliers of the antibodies are provided in represented by three independent siRNA constructs (Sigma). Supplementary Table S3. The different target sequences are described in Supplementary Table S1. In addition, each plate contained four nontargeting Chromatin immunoprecipitation assay controls and four siRNAs against Smo as positive controls. The Chromatin immunoprecipitation (ChIP) was carried out as automated screening was done using a MicrolabSTAR liquid described previously (23). For each immunoprecipitation, 0.5 handling system (Hamilton Robotics). NIH3T3 reporter cells to 1 mg of chromatin was used except for histones and histone (21) were reverse transfected with the customized siRNA modifications where 100 mg were used. The qPCR primers are library including individual positive and negative controls available in Supplementary Table S2, and the antibodies in using Lipofectamine 2000 (Invitrogen) according to the man- Supplementary Table S3. ufacturer's instructions. After 24 hours of transfection, the medium was changed to conditioned medium containing Orthotopic transplantation / Sonic Hedgehog protein (Shh; derived from 293-ShhN cells) Transformed NSCs [tNSC, Ink4a-Arf NSCs expressing or respective culture medium and incubated for another 48 constitutively active EGF receptor (EGFR) as well as hours. Firefly and Renilla luciferase activities were determined luciferase] were cultured in serum-free NSC medium con- using the Dual-Luciferase Kit (Promega) according to the taining growth factors (human recombinant EGF and basic manufacturer's protocol. fibroblast growth factor, bFGF). A total of 100,000 control and Pcaf knockdown tNSCs were diluted in 5 mLofNSC Screening data analysis and statistics medium and orthotopically injected into the brain of immu- Individual Firefly-intensities were normalized over their nocompromised mice (2 mm lateral and 1 mm anterior to corresponding Renilla-readings and further normalized to bregma and 3 mm beneath the skull) using a stereotactic the median sample score of each plate. From this, a gene- device (24). The mice were followed daily for the develop- based hit list was generated using the statistical method ment of neurologic deficits. "redundant siRNA activity" (RSA) analysis (22). The RSA method first ranked individual siRNAs according to their Statistical analysis normalized scores andthencalculatedP values for each Survival curves were compared using the log-rank test. In gene on the basis of the likelihood for this distribution of other experiments, data were presented as mean SD. Two- siRNA ranks to occur by chance. This calculation of P values tailed Student t test was used for statistical analysis. P value less was based on the iterative hypergeometric distribution. The than 0.05 was considered as statistically significant.

OF2 Cancer Res; 73(20) October 15, 2013 Cancer Research

PCAF Is Required for GLI-Dependent Transcription

Results C, efficient depletion of Smo led to an impairment of the Shh- A siRNA screening to identify acetyltransferases involved dependent induction of the pathway. in the Hh pathway In the screening, we used three different siRNAs against each To identify novel transcriptional regulators of GLI1, we acetyltransferase, a scrambled siRNA sequence, and siRNA carried out a siRNA-based screening that targeted the 17 against the positive regulator Smo. The statistical analyses of different HATs that have been characterized in mammalian the results are described in detail in Material and Methods and cells (25). We used a siRNA library generated and validated in Supplementary Fig. S1. As expected, the downregulation of in our laboratory, where each HAT was targeted by three Smo led to a significant decrease in Gli-mediated transcrip- different siRNA oligonucleotides (26). As a cellular system, tional activation (Fig. 1D and Supplementary Fig. S1A). More- we used NIH3T3 cells containing a luciferase reporter fused over, downregulation of Ep300, Ncoa1, and Pcaf led to a with the responsive element for Gli repeated eight times (8 significant decrease in Gli1 activity. Thus, these three candi- 30Gli-BS Luc) and a Renilla reporter (RL) as internal control dates and the two hits closest to being significant, Gtf3c4 and (Fig.1A;ref.21).Inthissystem,luciferaseactivityreflects the Kat2a, were selected for further analysis (Fig. 1D). activity of the Hh pathway. To activate the Hh pathway, To validate the effect of these five candidates, we repeated we used conditioned medium containing the Hh protein the Hh luciferase assay (Supplementary Fig. S2A). In this assay, Shh, produced in 293T cells stably expressing this protein we observed a comparable negative effect on Hh pathway (Fig. 1B). activation by downregulating Pcaf, Ep300, or Smo. To further As a positive control for the screening, we knocked down corroborate this finding, we correlated the efficiency of mRNA Smo, which is one of the most upstream positive regulators of knockdown for each of the three siRNAs with each of the the Hh pathway, using a specific siRNA. As shown in Fig. 1B and candidate genes to the corresponding effect on the Hh

A D P = 5% siScr Luciferase 8X3´BSGli1 siMyst4 siCrebbp Shh siMyst3

siKat5 GLI1 Luciferase siHat1 8X3´BSGli1 siClock siTaf1 siMyst1

siElp3 BC siNcoa2 Smo siMyst2 1.00 1.00 siNcoa3 siKat2a 0.75 0.75 siGtf3c4

0.50 0.50 siPcaf *** *** siNcoa1 0.25 0.25 siSmo ****

Relative luciferase readings luciferase Relative siEp300

0.00 levels mRNA expression Relative 0.00 −2.00 −1.50 −1.00 −0.50 0.00 siScr siSmo siScr siSmo siScr siSmo siScr siSmo Log P value + Shh + Shh

Figure 1. Identiﬁcation of acetyltransferases involved in the Hh pathway. A, schematic representation of the cellular system used for the screening. B, Luciferase assay analysis in NIH3T3 cells transfected with a nontargeting scramble (Scr) control siRNA and with a siRNA against Smo. The Hh pathway is induced by treatment with Shh-conditioned medium. C, qRT-PCR expression analysis of Smo (normalized to Rplpo levels) in cells transfected with the indicated siRNAs. The expression is monitored both at steady state and on Shh treatment. For B and C, the mean and SD for three independent experiments are shown. , P < 0.0004; , P < 0.0001. D, RSA-based hit-list of the acetyltransferase siRNA screen. Each gene is assigned a single log P value on the basis of the ranking distribution of all three siRNAs.

www.aacrjournals.org Cancer Res; 73(20) October 15, 2013 OF3

Malatesta et al.

luciferase readings (Supplementary Fig. S2B). Pcaf showed efﬁcient Pcaf siRNAs from the screening (Fig. 2B and Sup- highly signiﬁcant correlation (R2 > 0.8) between the transcript plementary Fig. S2B) impair the activation of the Hh path- knockdown and the luciferase activity. way to the same extent as the Smo siRNA control. Moreover, two independent short hairpin RNAs (shRNA) targeting Pcaf Pcaf is required for the activation of a Hh pathway- in different regions than those targeted by the siRNAs, dependent promoter used in the screening, reduce Hh signaling with comparable We conducted additional tests to validate the role of Pcaf strength to a control shRNA directed against Gli1 (Fig. 2C). in the Hh pathway. As shown in Fig. 2A, the two most In addition, Pcaf and Gli1 silencing abolish the activation of

A B 1.00

siScr siSmo siPcaf#1 siPcaf#2 0.75 Smo

0.50 Pcaf **** **** **** Tubulin 0.25 Relative luciferase readings luciferase Relative 0.00 Figure 2. PCAF is required for Hh pathway activity. A, Luc-assay

siScr analysis in NIH3T3 cells siScr siSmo siPcaf#1siPcaf#2 transfected with siRNAs against + Shh Pcaf, Smo, and a scramble siRNA control (Scr). B, Western blot C Gli1 Pcaf analysis of protein extracts of cells transfected with the indicated 1.00 1.00 1.00 siRNAs using the indicated antibodies. Tubulin was used as a loading control. C, Luciferase 0.75 0.75 0.75 activity measured in NIH3T3 cells transduced with Pcaf, Gli1, and Scr **** * shRNAs. The Hh pathway was 0.50 0.50 0.50 * activated by Shh treatment. The **** **** ** corresponding expression of Pcaf Gli1 Rplpo 0.25 0.25 0.25 and (normalized to levels) was determined by qRT- PCR. D, Luciferase activity Relative luciferase readings luciferase Relative Relative mRNA expression levels mRNA expression Relative

0.00 0.00 levels mRNA expression Relative 0.00 analyzed in NIH3T3 cells upon stable knockdown of Pcaf, Gli1,

shScr shScr shScr shScr and Scr control. The Hh pathway shScr shScr shGli1 shGli1 shPcaf#1 was induced by treatment with shPcaf#1shPcaf#2 + Shh shPcaf#2 SAG, a compound homolog to + Shh + Shh Smo. The expression of Pcaf and Gli1 (normalized to Rplpo levels) D Gli1 Pcaf was measured by qRT-PCR. From A–D, the mean and the SD for 1.00 1.00 1.00 at least three independent experiments are represented. , P < 0.01; , P < 0.003; , P < 0.75 0.75 0.75 ** 0.0006; and , P < 0.0001. *** ** 0.50 0.50 0.50 * **** *

0.25 0.25 0.25 Relative luciferase readings luciferase Relative Relative mRNA expression levels mRNA expression Relative 0.00 levels mRNA expression Relative 0.00 0.00

shScr shScr shScr shScr shGli1 shScr shScr shGli1 shPcaf#1shPcaf#2 shPcaf#1shPcaf#2 + SAG + SAG + SAG

OF4 Cancer Res; 73(20) October 15, 2013 Cancer Research

PCAF Is Required for GLI-Dependent Transcription

the Hh pathway induced by treatment with the small-mol- PCAF affects cell proliferation and expression of Hh ecule agonist of Smo (SAG; Fig. 2D). Taken together, these target genes in glioblastoma and medulloblastoma cells results show that Pcaf, like Gli1, is required for the activation Hh signaling has been reported to play a role in glioma and of a Gli-dependent promoter. glioblastoma as well as in medulloblastoma, the most frequent childhood brain cancer (16). Indeed, deregulation of the Hh- Pcaf is required for the expression of Hh target genes pathway resulting from PTCH inactivation or from mutations The activation of the Hh pathway leads to increased expres- in transducers of the signaling pathway can induce medullo- sion of Gli-regulated target genes (16). To test the involvement blastoma in mice and in humans (27). of Pcaf on modulation of Hh target genes, we determined the To test if PCAF plays a role in modulating the Hh pathway expression of two of these, namely Gli1 and Ptch upon depletion also in glioblastoma and medulloblastoma cells, we stably of Pcaf in NIH3T3 cells. Shh stimulation of NIH3T3 cells leads to knocked down the acetyltransferase using two independent an increase in Gli1 and Ptch expression of seven- and fourfold, shRNAs in different glioblastoma and medulloblastoma cell respectively (Fig. 3A), and this increase is significantly reduced lines. In PCAF-depleted cells, the growth is slowed down in cells transfected with two different siRNAs against Pcaf. A compared with control transduced cells (Fig. 4A–C, right). similar effect was observed in Smo siRNA-treated cells (Fig. 3A). Importantly, a similar effect is observed when GLI1 expression The specificity of this observation was further supported by the is downregulated (Fig. 4A–C, right). Moreover, the downregu- fact that stable knockdown of Pcaf or Gli1 also impaired the lation of PCAF and GLI1 led to similar reductions in the activation of Gli1 and Ptch by Shh (Fig. 3B). Importantly, the expression of GLI1 and PTCH (Fig. 4A–C, left), indicating a impairment of the Shh-mediated upregulation of Gli1 and Ptch possible common mechanism by which the two proteins affect correlates with the efficiency of Pcaf knockdown (Fig. 3B). cell proliferation. In NIH3T3 cells, Pcaf depletion did not affect DNA synthesis It has been reported that inhibition of Hh signaling mediated and cell-cycle profile (Supplementary Fig. S3 and data not by treatment with the Smo antagonist cyclopamine, decreases shown), suggesting that increased expression of Hh target cell proliferation of different glioblastoma cells, including U87 genes is directly mediated by Pcaf and not caused by changes and U118 (28). In contrast, proliferation of the T98G glioblas- in proliferation. toma cells is not affected (28). To further investigate the

A Gli1 Ptch Smo Pcaf

1.00 1.00 1.00 1.00

0.75 Figure 3. Pcaf is required for 0.75 ** 0.75 0.75 *** **** *** expression of Hh target genes. A, **** **** **** qRT-PCR analysis for expression 0.50 0.50 0.50 0.50 **** of Hh target genes, Gli1 and Ptch (normalized to Rplpo levels), 0.25 **** Relative mRNA Relative 0.25 mRNA Relative 0.25 0.25 carried out in NIH3T3 cells mRNA Relative Relative mRNA Relative transfected with the reported siRNAs. The cells were treated with 0.00 0.00 0.00 0.00 Shh-conditioned medium. The siScr expression of Pcaf and Smo siScr siScrsiSmo siScr siSmo siScrsiSmo siScr siPcaf#1siPcaf#2 siPcaf#1siPcaf#2 siPcaf#2 (normalized to Rplpo levels) is also + Shh siPcaf#1 + Shh reported. B, qRT-PCR + Shh + Shh experiments conducted in cells with a stable knockdown shRNA- B Gli1 Ptch Pcaf mediated of Pcaf, Gli1, and Scr 1.00 1.00 1.00 control. The Gli–Hh pathway was activated using Shh treatment. **** Gli1, Ptch, and Pcaf expressions 0.75 0.75 0.75 Rplpo **** were normalized to levels. **** **** All data in this ﬁgure are average **** **** **** **** values and SD from at least three 0.50 0.50 0.50 experiments, each carried out in triplicate. , P < 0.001; , P <

Relative mRNA Relative 0.25 0.25 0.25 Relative mRNA Relative 0.0002; , P < 0.0001. mRNA Relative

0.00 0.00 0.00

shScr shScr shGli1 shScr shScr shGli1 shScr shPcaf#1shPcaf#2 shPcaf#1shPcaf#2 shPcaf#1shPcaf#2 + Shh + Shh + Shh

www.aacrjournals.org Cancer Res; 73(20) October 15, 2013 OF5

Malatesta et al.

A DAOY GLI1 PTCH PCAF 100.0 shScr 1.00 1.00 2.00 ** ** ** 0.75 0.75 1.50 **** **** *** 0.50 0.50 1.00 50.0

0.25 0.25 0.50 Confluence (%) Relative mRNA Relative ** Relative mRNA Relative Relative mRNA Relative shPcaf#1 *** shGli1 0.00 0.00 0.00 shPcaf#2 0.00 20.0 40.0 60.0 80.0 shScrshGli1 shScrshGli1 shScrshGli1 Time (h) B shPcaf#1shPcaf#2 shPcaf#1shPcaf#2 shPcaf#1shPcaf#2 80.0 U87 shScr GLI1 PTCH PCAF 1.00 1.00 1.50 60.0 *** 0.75 0.75 **** 1.00 40.0 shPcaf#1 0.50 **** 0.50 **** **** **** *** 0.50 Confluence (%) shGli1 Relative mRNA Relative 0.25 0.25 mRNA Relative **** 20.0 Relative mRNA Relative

0.00 0.00 0.00 shPcaf#2 0.00 Figure 4. PCAF affects proliferation 50.0 100.0 and expression of Hh target shScr shGli1 shScrshGli1 shScrshGli1 Time (h) shPcaf#1shPcaf#2 shPcaf#1shPcaf#2 shPcaf#1shPcaf#2 genes in glioblastoma and C medulloblastoma cells. A–E, left, qRT-PCR analysis for the U118 80.0 GLI1 PTCH PCAF shScr expression of the endogenous Hh 1.00 1.00 GLI1 PTCH 1.00 60.0 target genes, and * (normalized to Rplpo levels), 0.75 0.75 0.75 ** ** 40.0 measured on depletion of Pcaf and * 0.50 0.50 0.50 Gli1. All data in these panels (A–E) * are average values and SD from at 0.25 ** **** Confluence (%) 20.0 shGli1 Relative mRNA Relative Relative mRNA Relative 0.25 0.25 **** Relative mRNA Relative shPcaf#1 least three experiments, each 0.00 0.00 0.00 shPcaf#2 carried out in triplicate. , P < 0.01; 0.00 P < P < 100.0 , 0.003; , 0.0005; 50.0 shScrshGli1 shScrshGli1 shScrshGli1 Time (h) , P < 0.0001. A–E, right, growth shPcaf#1shPcaf#2 shPcaf#1shPcaf#2 shPcaf#1shPcaf#2 curve of cells stably depleted for D Gli1 and Pcaf by different shRNAs, 100.0 T98G shPcaf#1 compared with the nontargeting GLI1 PTCH PCAF shScr control shScr. 1.00 1.50 1.50 0.75 shGli1 shPcaf#2 1.00 1.00 50.0 0.50 **** 0.50 Confluence (%) 0.50 0.25 **** Relative mRNA Relative Relative mRNA Relative Relative mRNA Relative **** 0.00 0.00 0.00 0.00 20.0 40.0 60.0 80.0

shScrshGli1 shScrshGli1 shScrshGli1 Time (h) shPcaf#1shPcaf#2 shPcaf#1shPcaf#2 shPcaf#1shPcaf#2 E 2,500 GBM543 GLI1 PTCH PCAF shScr 1.00 1.00 1.00 2,000

* ) 2 0.75 0.75 0.75 * ** 1,500 0.50 0.50 0.50 **** Area ( µ m ** 1,000

0.25 ** mRNA Relative 0.25

Relative mRNA Relative 0.25 ****

Relative mRNA Relative shGli1 **** shPcaf#1 0.00 0.00 0.00 500 shPcaf#2 50.0 100.0 150.0 shScrshGli1 shScrshGli1 shScrshGli1 Time (h) shPcaf#1shPcaf#2 shPcaf#1shPcaf#2 shPcaf#1shPcaf#2

speciﬁcity of PCAF toward the Hh pathway, we stably depleted We then evaluated the impact of PCAF silencing on the the acetyltransferase or GLI1 in T98G cells. Interestingly, proliferation of human primary glioblastoma neurospheres, neither GLI1 nor PCAF knockdown affects GLI1 and PTCH propagated in the presence of EGF and bFGF. These cell expression (Fig. 4D). cultures more closely recapitulate the phenotype and genotype

OF6 Cancer Res; 73(20) October 15, 2013 Cancer Research

PCAF Is Required for GLI-Dependent Transcription

of the tumors than do serum-cultured cell lines (29). As shown in Fig. 4E and in Supplementary Fig. S4A, PCAF stable depletion A PTCH in GBM543 neurospheres leads to reduction of cell growth and 1.00 affects the expression of the Hh target genes. Moreover, the capability of GBM543 cells to grow as neurospheres is strongly 0.80 ** impaired upon PCAF or GLI1 depletion (Fig. 4E and Supple- mentary Fig. S4B). 0.60 In addition, the requirement of PCAF for cell proliferation was tested by colony formation assays in both glioblastoma 0.40 ** *** ** and medulloblastoma cell lines. Indeed, PCAF or GLI1 depletion strongly reduces the number of U87 glioblastoma mRNA Relative 0.20 and DAOY-MB cell colonies (Supplementary Fig. S5A and 5B). Taken together, these findings suggest a role for PCAF 0.00 in regulating proliferation of glioblastoma and medulloblas- tomacellsthroughitscontribution to Hh target gene AA 6h DMSO AA 24h Cycl 6h expression. Cycl 24h Previous results have shown that DAOY cells require a func- B tional Hh pathway for cell proliferation (30), and that cyclopa- Ctr AA Cycl mine treatment leads to a reduction in Hh target gene expression Casp3 cl. and increased apoptosis (31). In agreement with being required for the expression of Hh target genes (Fig. 4A), PCAF depletion Pcaf also induces apoptosis (Supplementary Fig. S5C). As further evidence for the requirement of PCAF for Hh Tubulin target gene expression in medulloblastoma cells, we used AA to inhibit the acetyltransferase activity (32, 33). AA treatment of PTCH C 50.0 DAOY-MB cells led to a strong reduction of expression *** and an increase in apoptosis, similar to the effect induced by * cyclopamine (Fig. 5A and C). Importantly, PCAF expression is 40.0 ** not affected by AA treatment (Supplementary Fig. S5D), sug- * gesting that the chemical compound specifically inhibits the 30.0 activity of PCAF. These results strongly corroborate the finding that PCAF is required for the expression of Hh target genes and, 20.0 consequently, for cell proliferation. 10.0 PCAF interacts with GLI1 and regulates H3K9 acetylation on Hh target gene promoters 0.0

To characterize the role and the mechanism of action of cellls V−positive Annexin Percentage PCAF in regulating Hh pathway, we tested if PCAF can interact DMSOAA 24hAA 48hAA 72h with the GLI1 transcription factor. We conducted ChIP assays Cycl 24hCycl 48hCycl 72h using cells in which we ectopically expressed PCAF and GLI1. As shown in Fig. 6A, PCAF can indeed bind to GLI1, and, most Figure 5. PCAF inhibitor and cyclopamine induce apoptosis in DAOY-MB importantly, this interaction was also observed between the cells. A, qRT-PCR analysis to measure the expression of the endogenous two endogenous proteins (Fig. 6B). PTCH (normalized to Rplpo levels) in DAOY cells treated with AA or PCAF is a coactivator of transcription that is recruited to the cyclopamine for the indicated time. B, Western blot analysis in cells treated for 24 hours with AA or cyclopamine. Levels of PCAF and of the region surrounding the transcription start site of target genes apoptotic marker cleaved caspase-3 were measured using specific (34). Furthermore, PCAF has been described to acetylate antibodies. Tubulin served as a loading control. C, detection of apoptotic histones on specific residues. In vitro, the recombinant protein DAOY cells. Cells were treated with AA, cyclopamine, or DMSO as control shows a preference toward lysine residues 9 and 14 on histone for the indicated time and stained using fluorochrome-conjugated fi H3 (H3K9 and H3K14) with a lower efficiency toward H4K8 and Annexin V and 7AAD. All data in this gure are average values and SD from three independent experiments. , P < 0.01; , P < 0.001; , P < 0.0009. H4K16 (35–37). Deletion of Pcaf together with its family member Gcn5 in vivo leads to a dramatic reduction of H3K9 acetylation (38). To gain further insight into the molecular the amount of H3K9Ac, but not H4Ac, associated with the two mechanisms underlying PCAF-dependent regulation of Hh promoters is significantly reduced in cells expressing Pcaf target genes, we examined if Pcaf can regulate the posttrans- shRNA. Furthermore, we showed that knockdown of PCAF lational modification of H3 on Gli1 and Ptch promoters. To do and GLI1 both led to a dramatic reduction of H3K9ac asso- this, we determined the amount of acetylated H3K9 (H3K9ac) ciated with the PTCH promoter (Fig. 6D). associated with the Gli1 and Ptch promoters by ChIP–qPCR To further characterize the mechanism by which PCAF analysis in NIH3T3 cells treated with SAG. As shown in Fig. 6C, regulates Hh target genes, we tested if it is associated with

www.aacrjournals.org Cancer Res; 73(20) October 15, 2013 OF7

Malatesta et al.

A B +++− +++− pcDNA − ++++− − − PCAF-Flag IP: IgG IP: Gli1 Figure 6. −− ++− − ++GLI1-His Input PCAF interacts with GLI1 and regulates H3K9 acetylation on Gli1 Gli1 promoters of Hh target genes. A, coimmunoprecipitation Pcaf Flag experiments in 293 cells ectopically expressing PCAF-flag IP Input IP Input and GLI1-His vectors using the IP: anti-Flag IP: anti-Gli1 indicated antibodies. IP, C immunoprecipitation. B, Gli1 Ptch endogenous binding between shScr + SAG PCAF and GLI1 in U87 cells. C, 20.0 6.0 shPcaf#2 + SAG ChIP–qPCR analysis of the histone modifications H3K9ac and H4ac in 15.0 4.0 NIH3T3 cells depleted for PCAF versus control cells on induction 10.0 with SAG treatment. H3K9ac and % Input % Input 2.0 H4ac signals were normalized to 5.0 histone density using an H3- specific antibody. D and E, ChIP– qPCR analysis of the PTCH 0.00 0.00 IgG H3K9Ac H4Ac IgG H3K9Ac H4Ac promoter in U87 cells stably depleted using shRNAs against Pcaf or Gli1 versus a Scr control. DEPTCH PTCH PTCH H3K9Ac- and Pcaf-specific shScr shScr shScr antibodies were used for the 10.0 0.60 10.0 shPcaf#2 shGli1 shPcaf#2 immunoprecipitation. H3K9Ac shGli1 8.0 8.0 signal was normalized to histone 0.40 density using an H3 antibody. All 6.0 6.0 data in this figure are average values and SD from three 4.0 4.0 % Input % Input

% Input 0.20 independent experiments, each 2.0 2.0 carried out in triplicate.

0.00 0.00 0.00 IgG H3K9Ac IgG H3K9Ac IgG Pcaf

the PTCH promoter. As shown in Fig. 6E, PCAF associates with formation through its requirement for the expression of the the PTCH promoter, and the binding is dependent on the Hh target genes in vivo. expression of GLI1. Discussion Pcaf depletion in tumorigenic neural stem cells affects In this study, we have identiﬁed PCAF as an important the expression of Hh target genes and impairs tumor cofactor of the Hh-GLI signaling pathway. We have shown that formation PCAF is required for the transcriptional activation of Hh–Gli In high-grade glioma, the inactivation of the locus Ink4a-Arf target genes. Moreover, we have shown that PCAF binds to is frequently found together with the activation of the EGFR. GLI1, and both proteins are required for the increased H3K9Ac Moreover, it has been shown that constitutively active EGFR levels on Hh target gene promoters in response to the Hh–Gli mutants (EGFR) are able to transform murine Ink4a-Arf-null activation. Because the association of PCAF with GLI-regulat- NSCs in the mouse brain (39). To investigate the role of Pcaf in ed promoters is dependent on GLI1, we propose a model regulating Hh pathway in this context, we stably knocked- (Fig. 7C) in which the activation of the Hh–Gli signaling / down the acetyltransferase in tumorigenic NSCs (Ink4a-Arf ; pathway leads to GLI1-dependent recruitment of PCAF. This, EGFR). As shown in Fig. 7A, PCAF depletion impairs the in turn, leads to H3K9acetylation of Hh target gene promoters expression of different Hh target genes and affects proliferation and their activation. of tNSCs (Supplementary Fig. S6). To test the role of PCAF for Deregulation of the Hh–Gli pathway contributes to the devel- the formation of tumors in vivo, we injected 100,000 control- opment and maintenance of several types of tumors, including depleted or PCAF-depleted tNSCs into the brain of immuno- glioblastomas and medulloblastomas. Interestingly, our results compromised mice. The control-depleted cells readily formed show that depletion of PCAF leads to impairment of cell prolif- tumors, and the mice died with a median latency of 7 weeks, eration and induction of apoptosis in glioblastoma and medul- whereas the mice injected with PCAF-depleted tNSCs survived loblastoma cell lines, whereas no effect was observed in ﬁbro- with no signs of tumor development (Fig. 7B). Taken together, blasts. These results are in agreement with the functional these results suggest a role for PCAF in regulating tumor requirement of the Hh–Gli signaling pathway in glioblastoma

OF8 Cancer Res; 73(20) October 15, 2013 Cancer Research

PCAF Is Required for GLI-Dependent Transcription

Figure 7. PCAF depletion in NSCs / A Gli1 Ptch Bmi1 Pcaf B Ink4a-Arf ; EGFR affects 100 expression of Hh target genes and 1.00 1.00 1.00 1.00 ** shPcaf#2 0.80 0.80 0.80 impairs tumor formation. A, qRT- 0.80 shScr ** PCR analysis for the expression of 0.60 0.60 0.60 ** 0.60 50 the endogenous Hh target genes Rplpo 0.40 0.40 0.40 0.40 (normalized to levels) ** Relative mRNA Relative Relative mRNA Relative Relative mRNA Relative Percentage survival Percentage Relative mRNA Relative measured on depletion of Pcaf. 0.20 0.20 0.20 0.20 0 Error bars represent mean and SD 0.00 0.00 0.00 0.00 0 50 100 150 from three independent Days shScr shScr shScr shScr experiments. , P < 0.004. B, shPcaf#2 shPcaf#2 shPcaf#2 shPcaf#2 survival curve of mice intracranially injected with NSCs Ink/Arf-null C EGFR engineered to express an Anti-Shh Ab shRNA against Pcaf or against a Smo inhibitors scramble control. C, model HH describing PCAF function. On Hh pathway activation, PCAF is recruited to Hh target gene Cytoplasm promoters through an interaction SMO PTCH Compounds acting at different with GLI1. The association between levels within the pathway PCAF and GLI1 leads to the H3K9Acetylation of the Hh target gene promoters, resulting in an PCAF inhibitor K9Ac K9Ac PCAF increase of the Hh target genes anti-Gli agents GLI1 Gli target genes expression. Known and potential novel Hh inhibitors are represented in the gray boxes. Nucleus

and medulloblastoma, and suggest that inhibition of PCAF interesting as drug targets. Recently, some natural HAT activity could be an attractive avenue to treat Hh/Gli-dependent inhibitors including AA, garcinol, and curcumin were shown tumors such as glioblastomas and medulloblastomas. to inhibit the activities of PCAF, p300, and CBP (33, 49). Different inhibitors targeting the Hh pathway have been Other compounds have been synthesized on the basis of proposed as anticancer agents (Fig. 7C). SMO inhibitors, like these natural products, and some of these have been shown cyclopamine, inhibit Hh-induced tumor growth in vivo and in to prevent growth of cancer cells without affecting prolif- vitro (40, 41). Recently, other small-molecule antagonists of eration of nonmalignant cells (50). SMO have been identified and, in particular, one compound As a proof-of-concept for the use of chemical inhibitors to (vismodegib) recently became the first U.S. Food and Drug PCAF to treat Hh–Gli-dependent tumors, we used AA and Administration (FDA)-approved Smo inhibitor for the treat- showed that AA treatment leads to reduced expression of PTCH, ment of basal cell carcinoma (BCC; ref. 42). Vismodegib and a Gli-target gene, and a clear increase in apoptosis in DAOY- other SMO antagonists are currently being tested in clinical medulloblastoma cells. Although we cannot exclude that AA trials as potential treatments for other tumors including treatments are inhibiting not only PCAF but also p300 and CBP, medulloblastoma, pancreatic, ovarian, and hematopoietic can- the observation that depletion of PCAF gives very similar effects cers. Use of Hh-inhibitors acting downstream of SMO have also suggests PCAF to be the main HAT in this setting. been reported as therapeutic alternatives, including inhibitors We hope that our results will inspire the development of to GLI1 (43–45). highly potent and specific inhibitors of PCAF for the treatment Histone deacetylase inhibitors (HDACi) have been describ- of patients who have cancer. In addition, such compounds will ed as multifunctional anticancer agents. Their role in brain be valuable for a better definition of PCAF as anticancer drug tumors has been studied, and, in particular, HDACis have been target, and for the understanding of the biologic function of shown to induce cell-cycle arrest associated with p21 increase PCAF in normal cells as well as in tumors. and to inhibit the growth of glioblastoma cell lines (46). fl Moreover, multiple oral HDACis have been studied in phase Disclosure of Potential Con icts of Interest No potential conflicts of interest were disclosed. I trial in pediatric patients (47). Recently, a role for HDACs in Hh pathway regulation has been reported. For instance, a Authors' Contributions recent study has shown that HDAC1 can deacetylate GLI1, Conception and design: M. Malatesta, M. Squartrito, K. Helin leading to the activation of GLI-target genes (48). Development of methodology: M. Malatesta, C. Steinhauer, M. Squartrito Acquisition of data (provided animals, acquired and managed patients, HATs play essential roles in normal cellular function; provided facilities, etc.): M. Malatesta, C. Steinhauer, F. Mohammad, D.P. however, they also contribute to the pathogenesis of differ- Pandey Analysis and interpretation of data (e.g., statistical analysis, biosta- ent diseases, including tumorigenesis and neurodegenera- tistics, computational analysis): M. Malatesta, C. Steinhauer, D.P. Pandey, tive disorders. HATs as a class of enzymes are, therefore, M. Squartrito

www.aacrjournals.org Cancer Res; 73(20) October 15, 2013 OF9

Malatesta et al.

Writing, review, and/or revision of the manuscript: M. Malatesta, Grant Support C. Steinhauer, F. Mohammad, M. Squartrito, K. Helin This work was ﬁnancially supported by the Danish National Advanced Administrative, technical, or material support (i.e., reporting or orga- Technology Foundation, The Danish National Research Foundation, the Danish nizing data, constructing databases): M. Malatesta Medical Research Council, the Lundbeck Foundation, and the Novo Nordisk Study supervision: M. Malatesta, K. Helin Foundation. The costs of publication of this article were defrayed in part by the payment of Acknowledgments page charges. This article must therefore be hereby marked advertisement in The authors thank Anna Fossum for assistance in FACS sorting and Michael accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Lees for helping with acquiring data for the proliferation experiments. The authors also thank Dr. Jussi Taipale for providing 239-ShhN and NIH3T3 cells and Received December 21, 2012; revised June 19, 2013; accepted July 24, 2013; all the members of the Helin laboratory for critical and fruitful discussions. published OnlineFirst August 13, 2013.

References 1. Ingham PW. Hedgehog signaling: a tale of two lipids. Science 20. Schmitz SU, Albert M, Malatesta M, Morey L, Johansen JV, Bak M, 2001;294:1879–81. et al. Jarid1b targets genes regulating development and is involved in 2. Ruiz i Altaba A, Sanchez P, Dahmane N. Gli and hedgehog in cancer: neural differentiation. EMBO J 2011;30:4586–600. tumours, embryos and stem cells. Nat Rev Cancer 2002;2:361–72. 21. Taipale J, Chen JK, Cooper MK, Wang B, Mann RK, Milenkovic L, et al. 3. McMahon AP, Ingham PW, Tabin CJ. Developmental roles and clinical Effects of oncogenic mutations in Smoothened and Patched can be significance of Hedgehog signaling. Curr Top Dev Biol 2003;53:1–114. reversed by cyclopamine. Nature 2000;406:1005–9. 4. Taipale J, Beachy PA. The Hedgehog and Wnt signalling pathways in 22. Konig R, Chiang CY, Tu BP, Yan SF, DeJesus PD, Romero A, et al. A cancer. Nature 2001;411:349–54. probability-based approach for the analysis of large-scale RNAi 5. Taipale J, Cooper MK, Maiti T, Beachy PA. Patched acts catalytically to screens. Nat Methods 2007;4:847–9. suppress the activity of Smoothened. Nature 2002;418:892–7. 23. Pasini D, Cloos PA, Walfridsson J, Olsson L, Bukowski JP, Johansen 6. Sasaki H, Hui C, Nakafuku M, Kondoh H. A binding site for Gli JV, et al. JARID2 regulates binding of the Polycomb repressive com- proteins is essential for HNF-3beta floor plate enhancer activity in plex 2 to target genes in ES cells. Nature 2010;464:306–10. transgenics and can respond to Shh in vitro. Development 1997;124: 24. Bruggeman SW, Hulsman D, Tanger E, Buckle T, Blom M, Zevenhoven 1313–22. J, et al. Bmi1 controls tumor development in an Ink4a/Arf-independent 7. Kinzler KW, Vogelstein B. The GLI gene encodes a nuclear protein manner in a mouse model for glioma. Cancer Cell 2007;12:328–41. which binds specific sequences in the human genome. Mol Cell Biol 25. Marmorstein R, Roth SY. Histone acetyltransferases: function, struc- 1990;10:634–42. ture, and catalysis. Curr Opin Genet Dev 2001;11:155–61. 8. Freeman M. Feedback control of intercellular signalling in develop- 26. Pasini D, Malatesta M, Jung HR, Walfridsson J, Willer A, Olsson L, ment. Nature 2000;408:313–9. et al. Characterization of an antagonistic switch between histone H3 9. Stein U, Eder C, Karsten U, Haensch W, Walther W, Schlag PM. GLI lysine 27 methylation and acetylation in the transcriptional regula- gene expression in bone and soft tissue sarcomas of adult patients tion of Polycomb group target genes. Nucleic Acids Res 2010;38: correlates with tumor grade. Cancer Res 1999;59:1890–5. 4958–69. 10. Thayer SP, di Magliano MP, Heiser PW, Nielsen CM, Roberts DJ, 27. Northcott PA, Korshunov A, Witt H, Hielscher T, Eberhart CG, Mack S, Lauwers GY, et al. Hedgehog is an early and late mediator of pancreatic et al. Medulloblastoma comprises four distinct molecular variants. cancer tumorigenesis. Nature 2003;425:851–6. J Clin Oncol 2011;29:1408–14. 11. Unden AB, Zaphiropoulos PG, Bruce K, Toftgard R, Stahle-Backdahl 28. Dahmane N, Sanchez P, Gitton Y, Palma V, Sun T, Beyna M, et al. The M. Human patched (PTCH) mRNA is overexpressed consistently in Sonic Hedgehog-Gli pathway regulates dorsal brain growth and tumor cells of both familial and sporadic basal cell carcinoma. Cancer tumorigenesis. Development 2001;128:5201–12. Res 1997;57:2336–40. 29. Lee J, Kotliarova S, Kotliarov Y, Li AG, Su Q, Donin NM, et al. Tumor 12. Watkins DN, Berman DM, Burkholder SG, Wang B, Beachy PA, Baylin stem cells derived from glioblastomas cultured in bFGF and EGF more SB. Hedgehog signalling within airway epithelial progenitors and in closely mirror the phenotype and genotype of primary tumors than do small-cell lung cancer. Nature 2003;422:313–7. serum-cultured cell lines. Cancer Cell 2006;9:391–403. 13. Roberts WM, Douglass EC, Peiper SC, Houghton PJ, Look AT. Ampli- 30. Stecca B, Ruiz i Altaba A. A GLI1-p53 inhibitory loop controls neural fication of the gli gene in childhood sarcomas. Cancer Res 1989;49: stem cell and tumour cell numbers. EMBO J 2009;28:663–76. 5407–13. 31. Bar EE, Chaudhry A, Farah MH, Eberhart CG. Hedgehog signaling 14. Kinzler KW, Bigner SH, Bigner DD, Trent JM, Law ML, O'Brien SJ, et al. promotes medulloblastoma survival via Bc/II. Am J Pathol 2007;170: Identification of an amplified, highly expressed gene in a human 347–55. glioma. Science 1987;236:70–3. 32. Ghizzoni M, Boltjes A, Graaf C, Haisma HJ, Dekker FJ. Improved 15. Nilsson M, Unden AB, Krause D, Malmqwist U, Raza K, Zaphiropoulos inhibition of the histone acetyltransferase PCAF by an anacardic acid PG, et al. Induction of basal cell carcinomas and trichoepitheliomas derivative. Bioorg Med Chem 2010;18:5826–34. in mice overexpressing GLI-1. Proc Natl Acad Sci U S A 2000;97: 33. Eliseeva ED, Valkov V, Jung M, Jung MO. Characterization of novel 3438–43. inhibitors of histone acetyltransferases. Mol Cancer Ther 2007;6: 16. Clement V, Sanchez P, de Tribolet N, Radovanovic I, Ruiz i Altaba A. 2391–8. HEDGEHOG-GLI1 signaling regulates human glioma growth, cancer 34. Wang Z, Zang C, Rosenfeld JA, Schones DE, Barski A, Cuddapah S, stem cell self-renewal, and tumorigenicity. Curr Biol 2007;17: et al. Combinatorial patterns of histone acetylations and methylations 165–72. in the human genome. Nat Genet 2008;40:897–903. 17. Grachtchouk M, Mo R, Yu S, Zhang X, Sasaki H, Hui CC, et al. Basal cell 35. Grant PA, Eberharter A, John S, Cook RG, Turner BM, Workman JL. carcinomas in mice overexpressing Gli2 in skin. Nat Genet 2000; Expanded lysine acetylation specificity of Gcn5 in native complexes. 24:216–7. J Biol Chem 1999;274:5895–900. 18. Varjosalo M, Bjorklund M, Cheng F, Syvanen H, Kivioja T, Kilpinen S, 36. Schiltz RL, Mizzen CA, Vassilev A, Cook RG, Allis CD, Nakatani Y. et al. Application of active and kinase-deficient kinome collection for Overlapping but distinct patterns of histone acetylation by the human identification of kinases regulating hedgehog signaling. Cell coactivators p300 and PCAF within nucleosomal substrates. J Biol 2008;133:537–48. Chem 1999;274:1189–92. 19. Bazzoli E, Pulvirenti T, Oberstadt MC, Perna F, Wee B, Schultz N, et al. 37. Kuo MH, Brownell JE, Sobel RE, Ranalli TA, Cook RG, Edmondson DG, MEF promotes stemness in the pathogenesis of gliomas. Cell Stem et al. Transcription-linked acetylation by Gcn5p of histones H3 and H4 Cell 2012;11:836–44. at specific lysines. Nature 1996;383:269–72.

OF10 Cancer Res; 73(20) October 15, 2013 Cancer Research

PCAF Is Required for GLI-Dependent Transcription

38. Jin Q, Yu LR, Wang L, Zhang Z, Kasper LH, Lee JE, et al. Distinct hog pathway blockade. Proc Natl Acad Sci U S A 2009;106: roles of GCN5/PCAF-mediated H3K9ac and CBP/p300-mediated 14132–7. H3K18/27ac in nuclear receptor transactivation. EMBO J 2011;30: 45. Hosoya T, Arai MA, Koyano T, Kowithayakorn T, Ishibashi M. Naturally 249–62. occurring small-molecule inhibitors of hedgehog/GLI-mediated tran- 39. Bachoo RM, Maher EA, Ligon KL, Sharpless NE, Chan SS, You MJ, scription. Chembiochem 2008;9:1082–92. et al. Epidermal growth factor receptor and Ink4a/Arf: convergent 46. Horing E, Podlech O, Silkenstedt B, Rota IA, Adamopoulou E, Nau- mechanisms governing terminal differentiation and transformation mann U. The histone deacetylase inhibitor trichostatin a promotes along the neural stem cell to astrocyte axis. Cancer Cell 2002;1: apoptosis and antitumor immunity in glioblastoma cells. Anticancer 269–77. Res 2013;33:1351–60. 40. Berman DM, Karhadkar SS, Hallahan AR, Pritchard JI, Eberhart CG, 47. New M, Olzscha H, La Thangue NB. HDAC inhibitor-based therapies: Watkins DN, et al. Medulloblastoma growth inhibition by Hedgehog can we interpret the code? Mol Oncol 2012;6:637–56. pathway blockade. Science 2002;297:1559–61. 48. Canettieri G, Di Marcotullio L, Greco A, Coni S, Antonucci L, Infante P, 41. Chen JK, Taipale J, Cooper MK, Beachy PA. Inhibition of Hedgehog et al. Histone deacetylase and Cullin3-REN(KCTD11) ubiquitin ligase signaling by direct binding of cyclopamine to Smoothened. Genes Dev interplay regulates Hedgehog signalling through Gli acetylation. Nat 2002;16:2743–8. Cell Biol 2010;12:132–42. 42. Sekulic A, Migden MR, Oro AE, Dirix L, Lewis KD, Hainsworth JD, et al. 49. Balasubramanyam K, Altaf M, Varier RA, Swaminathan V, Ravindran A, Efﬁcacy and safety of vismodegib in advanced basal-cell carcinoma. Sadhale PP, et al. Polyisoprenylated benzophenone, garcinol, a nat- N Engl J Med 2012;366:2171–9. ural histone acetyltransferase inhibitor, represses chromatin transcrip- 43. Lauth M, Bergstrom A, Shimokawa T, Toftgard R. Inhibition of GLI- tion and alters global gene expression. J Biol Chem 2004;279: mediated transcription and tumor cell growth by small-molecule 33716–26. antagonists. Proc Natl Acad Sci U S A 2007;104:8455–60. 50. Stimson L, Rowlands MG, Newbatt YM, Smith NF, Raynaud FI, Rogers 44. Hyman JM, Firestone AJ, Heine VM, Zhao Y, Ocasio CA, Han K, P, et al. Isothiazolones as inhibitors of PCAF and p300 histone et al. Small-molecule inhibitors reveal multiple strategies for Hedge- acetyltransferase activity. Mol Cancer Ther 2005;4:1521–32.

www.aacrjournals.org Cancer Res; 73(20) October 15, 2013 OF11

Histone Acetyltransferase PCAF Is Required for Hedgehog− Gli-Dependent Transcription and Cancer Cell Proliferation

Martina Malatesta, Cornelia Steinhauer, Faizaan Mohammad, et al.

Cancer Res Published OnlineFirst August 13, 2013.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-12-4660

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2013/08/13/0008-5472.CAN-12-4660.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/2013/10/04/0008-5472.CAN-12-4660. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.