HDAC5 and HDAC9 in Medulloblastoma: Novel Markers for Risk Stratification and Role in Tumor Cell Growth

Published OnlineFirst April 22, 2010; DOI: 10.1158/1078-0432.CCR-10-0395 Published OnlineFirst on June 8, 2010 as 10.1158/1078-0432.CCR-10-0395

Clinical Imaging, Diagnosis, Prognosis Cancer Research HDAC5 and HDAC9 in Medulloblastoma: Novel Markers for Risk Stratification and Role in Tumor Cell Growth

Till Milde1, Ina Oehme1, Andrey Korshunov2,5, Annette Kopp-Schneider3, Marc Remke4,6, Paul Northcott7, Hedwig E. Deubzer1,6, Marco Lodrini1,6, Michael D. Taylor7, Andreas von Deimling2,5, Stefan Pfister4,6, and Olaf Witt1,6

Abstract Purpose: Medulloblastomas are the most common malignant brain tumors in childhood. Survivors suffer from high morbidity because of therapy-related side effects. Thus, therapies targeting tumors in a specific manner with small molecules such as histone deacetylase (HDAC) inhibitors are urgently war- ranted. This study investigated the expression levels of individual human HDAC family members in primary medulloblastoma samples, their potential as risk stratification markers, and their roles in tumor cell growth. Experimental Design: Gene expression arrays were used to screen for HDAC1 through HDAC11. Using quantitative real time reverse transcriptase-PCR and immunohistochemistry, we studied the expression of HDAC5 and HDAC9 in primary medulloblastoma samples. In addition, we conducted functional studies using siRNA-mediated knockdown of HDAC5 and HDAC9 in medulloblastoma cells. Results: HDAC5 and HDAC9 showed the highest expression in prognostically poor subgroups. This finding was validated in an independent set of medulloblastoma samples. High HDAC5 and HDAC9 expression was significantly associated with poor overall survival, with high HDAC5 and HDAC9 expression posing an independent risk factor. Immunohistochemistry revealed a strong expression of HDAC5 and HDAC9 proteins in most of all primary medulloblastomas investigated. siRNA-mediated knockdown of HDAC5 or HDAC9 in medulloblastoma cells resulted in decreased cell growth and cell viability. Conclusion: HDAC5 and HDAC9 are significantly upregulated in high-risk medulloblastoma in comparison with low-risk medulloblastoma, and their expression is associated with poor survival. Thus, HDAC5 and HDAC9 may be valuable markers for risk stratification. Because our functional studies point toward a role in medulloblastoma cell growth, HDAC5 and HDAC9 may potentially be novel drug targets. Clin Cancer Res; 16(12); 3240–52. ©2010 AACR.

Medulloblastoma is the most common malignant intra- meters, such as age, dissemination at diagnosis, and extent cranial tumor in childhood (1) and represents a very het- of surgical resection, are used for risk stratification. Recent- erogeneous group as far as outcome is concerned. 5-Year ly, novel molecular markers, most notably DNA copy- event-free survival can be as low as 34.7% in patients with number variations of chromosomal regions 6q and 17q, metastasized disease (2) compared with 81% in patients and MYC/MYCN have been proposed for risk stratifica- with localized disease (3). Traditionally, clinical para- tion, which clearly separate prognostic subgroups (4). Many of the surviving patients suffer from therapy-related Authors' Affiliations: 1Clinical Cooperation Unit Pediatric Oncology, side effects, especially if radiotherapy is included in the 2Clinical Cooperation Unit Neuropathology, 3 Department of treatment regimen during infancy (5, 6). It is therefore Biostatistics, 4Division Molecular Genetics, German Cancer Research Center; Departments of 5Neuropathology and 6Pediatric Oncology, most important to develop novel treatment strategies that Hematology and Immunology, University Hospital Heidelberg, help to increase the survival of high-risk patients and at Heidelberg, Germany; and 7Division of Neurosurgery, Program in Developmental and Stem Cell Biology, Arthur and Sonia Labatt Brain the same time induce less side effects. Tumour Research Centre, Hospital for Sick Children, University of One of the strategies that have been followed to achieve Toronto, Toronto, Ontario, Canada this goal is the use of small molecules inhibiting histone Note: Supplementary data for this article are available at Clinical Cancer deacetylases (HDAC; ref. 7). A wide body of literature pro- Research Online (http://clincancerres.aacrjournals.org/). vides evidence for effective treatment of different tumor Corresponding Author: Till Milde, Clinical Cooperation Unit Pediatric cells using HDAC inhibitors (HDACi) in vitro and in vivo, Oncology (G340), German Cancer Research Center, Im Neuenheimer Feld 280, 69120 Heidelberg, Germany. Phone: 49-6221-423388; Fax: such as leukemia (8), lymphoma (9), lung cancer (10, 11), 49-6221-423277; E-mail: [email protected]. retinoblastoma (12), and neuroblastoma (13, 14). Brain doi: 10.1158/1078-0432.CCR-10-0395 tumor cells seem to be susceptible to treatment with ©2010 American Association for Cancer Research. HDACi as has been shown for glioblastoma (15, 16),

3240 Clin Cancer Res; 16(12) June 15, 2010

HDAC5 and HDAC9 in Medulloblastoma

seem to have nonredundant and specific functions con- Translational Relevance trolling hallmarks of cancer biology, such as proliferation, apoptosis, differentiation, migration, resistance to chemo- Patients with medulloblastoma experience poor therapy, and angiogenesis (25, 26). outcomes, especially in metastasized disease, and treat- The specific functions of single HDAC family members ment of all stages is associated with strong side effects, furthermore seem to be tumor specific, and accordingly, resulting in impaired quality of life. Specific therapies our group has started to dissect the function of individual for such high-risk patients are therefore urgently need- HDAC family members in distinct tumor entities. For ed to resolve this unsatisfactory situation. Histone dea- example, HDAC8 expression was found to correlate with cetylase (HDAC) inhibitors are a promising novel class poor outcome of neuroblastoma tumors, a highly ma- of targeted therapeutics. Our group and others have pre- lignant childhood cancer derived from neural crest pro- viously shown that differential expression of single genitor cells. Specific inhibition of HDAC8 induces HDAC family members is associated with poor outcome differentiation of neuroblastoma cells (14, 27). Of note, in solid tumors and that selective targeting of the identi- HDAC8 disruption in mice impairs neural crest cell fate fied HDACs could be a successful strategy. Here, we de- (28).ClassIHDAC1,HDAC2,andHDAC3havebeen scribe HDAC5 and HDAC9 as independent prognostic shown to be highly expressed in colorectal and gastric can- markers for overall survival in primary medulloblasto- cers (29, 30). The expression of some of the isoenzymes ma, and we show a functional role of HDAC5 and can be associated with prognosis (29, 30), and in the case HDAC9 in tumor cell growth in medulloblastoma cell of colorectal carcinomas, targeting of the specific isoen- lines. This finding is of general interest in oncology be- zymes was a successful strategy in cell culture models (30). cause class I HDACs, especially HDAC1 and HDAC2, Here, we examined the expression patterns and functions have thus far been considered to be the most relevant of HDAC isoenzymes in medulloblastoma and correlated HDAC family members to be targeted in cancer. isoenzyme expression with clinical course. For the first time, we provide evidence for a role of the class IIa HDACs, HDAC5 and HDAC9 in medulloblastoma cell growth and propose HDAC5 and HDAC9 as novel prognostic markers. atypical teratoid/rhabdoid tumor (17), and medulloblastoma (17–19). HDACis have only recently been introduced Materials and Methods in the clinical setting of cancer treatment, with Vorinostat being the first HDACi approved for the treatment of cutane- Patients ous T-cell lymphoma by the Food and Drug Administration Material from patients from the first set (n =37snap (20). However, most of the HDACis target either all or at frozen samples in liquid nitrogen at -196°C), as well as least a wide range of HDACs (21). This creates the problem the paraffin embedded medulloblastoma samples, were of unspecific inhibition of several HDACs, whereas the randomly collected at the Department of Neuropathology, targeted blockade of specific single HDACs might be more Burdenko Neurosurgical Institute (Moscow, Russia) be- desirable instead. Class-specific side effects of pan-HDACis tween 1993 and 2003. Approval to link laboratory data have been reported (22), supporting the requirement of to clinical data was obtained by the Institutional Review selective inhibitor development. Board. Two neuropathologists confirmed the diagnoses ac- The zinc-dependent HDAC1 through HDAC11 com- cording to the 2000 WHO classification. None of the pa- prise 11 members grouped into four classes (I, IIa, IIb, tients had received irradiation or chemotherapy before and IV; ref. 21). Knockout studies in mice suggest nonre- collection of specimens. Metastatic state (M stage) was dundant and specific functions of single HDACs in the determined by magnetic resonance imaging and cerebro- physiologic setting. For example, HDAC1 and HDAC3 spinal fluid cytopathology at diagnosis. Clinical and knockouts are early embryonic lethal in accordance with histopathologic data are summarized in Supplementary their essential und ubiquitous function as components Table S1. of repressor complexes and cell cycle progression; HDAC2, Material from patients from the second set (n =103 HDAC5, and HDAC9 control myocardial development samples) was obtained in accordance with the Research and function; HDAC4 is involved in bone and chondro- Ethics Board at the Hospital for Sick Children (Toronto, cyte development; HDAC7 plays a central role in endothe- Ontario, Canada), from the Co-operative Human Tissue lial cell adhesion and HDAC6 in tubulin acetylation; and Network (Columbus, OH), and the Brain Tumor Tissue HDAC8 knockouts show a distinct neural crest cell pheno- Bank (London, Ontario, Canada) as described (31). type (23). Thus, these data already point to distinct and specific functions of individual HDAC family members. DNA extraction and array-based comparative It can be expected that tissue- and time-specific disruption genomic hybridization of single HDAC will uncover even more physiologic func- Extraction of high–molecular weight DNA and RNA from tions of particular HDACs since liver-specific disruption of frozen tumor samples was carried as previously described HDAC3 causes deregulation of carbohydrate and lipid (32). Selection of genomic clones, isolation of bacterial metabolism (24). In cancer, deregulated HDACs also artificial chromosome DNA, performance of degenerate

www.aacrjournals.org Clin Cancer Res; 16(12) June 15, 2010 3241

Milde et al.

oligonucleotide primer-PCR, and preparation of microar- immunohistochemical stainings were done on 5-μm-thick rays were done as described (33). Labeling, hybridization, sections of formalin-fixed, paraffin-embedded microdis- and washing procedure were done as reported (34). sected specimens. The antibodies used were obtained from Abcam: HDAC5, ab55403 (1:500); and HDAC9, ab59718 RNA extraction, cDNA synthesis, and quantitative (1:20). Immunohistochemistry for HDAC5 and HDAC9 real-time reverse transcriptase-PCR (RT-PCR) was done with an automated stainer (Benchmark XT, Ven- RNA was extracted from biopsy samples after milling of tana) following the protocols of the manufacturer. All the frozen primary medulloblastoma sample in a Mikro- analyses of immunohistochemical stainings were carried Dismembrator S (B. Braun). RNA from primary medullo- out by two investigators (A. Korshunov and M. Remke) blastoma samples was extracted using Trizol (Invitrogen), who were blinded to clinical and molecular variables us- and RNA from cell culture experiments was extracted ing a scoring system. Fluorescence in situ hybridization was using the RNeasy Mini Kit (Qiagen), both according to carried out as described (38). manufacturer's instructions. cDNA was synthesized using the First Strand cDNA Synthesis Kit (Fermentas) according Cell culture and siRNA-mediated knockdown to manufacturer's instructions. Quantitative real-time Daoy cells were obtained through American Type Cul- PCR was done using an ABI Prism 7700 thermal cycler ture Collection, UW228-2 cells were a friendly gift from (Applied Biosystems) in standard mode with Platinum John Silber (Seattle, WA), UW228-3 cells were a friendly SYBR Green qPCR SuperMix-UDG (Invitrogen). The quan- gift from Steven Clifford (Newcastle, United Kingdom), titative real-time PCR conditions were 50°C (2 min), 95°C ONS76 was obtained from the Institute for Fermentation (10 min), 40 cycles of 95°C (15 s), and 60°C (1 min). (Japan), and Med8A were a friendly gift from R. Gilbertson Primers were obtained through Thermo Electron (se- (Memphis, TN). Cells were tested for mycoplasma, quences in Supplementary Table S2). The software used viral, and cell contamination using the in-house Multi- to analyze the data was SDS v. 1.3.1 (Applied Biosystems). plex cell Contamination Testing Service (39).For siRNA- The ΔΔCt method was used to obtain relative quantifica- mediated knockdown, cells were- seeded in 6-well tion. ACTB was used as a control gene whereas normal cer- plates on the day before transfection. On day 0, cells ebellum RNA as a control sample. Normal cerebellum were transfected with siRNA at 25 nmol/L concen- RNA was purchased from Clontech. trations using HiPerfect transfection reagent (Qiagen) according to manufacturer's instructions. siRNAs were Gene expression microarray obtained through Qiagen and Dharmacon/Thermo Fisher Hybridized microarrays were scanned at 5-μm resolution Scientific (catalog numbers in Supplementary Table S3). in a two-color Agilent Scanner G25505B (Agilent) with au- RNA was extracted at indicated time points as described tomatically adjusted photomultiplier tube (PMT) voltages above. according to manufacturer's specification. Array raw data were generated from scanned images using Axon GenePix- Western blot analysis and image processing Pro Software (version 6.1.0.2). The data was preprocessed, Protein concentrations of cell lysates were determined quality controlled, and analyzed with our in-house-devel- using the Bradford assay (Bio-Rad) according to manufac- oped ChipYard framework for microarray data analysis turer's instructions. The following antibodies were used: (http://www.dkfz.de/genetics/ChipYard/) using R (35) polyclonal rabbit anti-human HDAC5 (1:500; catalog and Bioconductor (36) software packages. Feature signals no. 2082; Cell Signaling), polyclonal rabbit anti-human had to fulfill the following criteria to be considered for anal- HDAC9 (1:500; catalog no. ab53102; Abcam), and mouse ysis: minimal signal to background ratio ≥1.2 in at least one monoclonal anti–β-actin (clone AC-15; Sigma-Aldrich). channel; mean to median spot intensity ≤75% quartile + 3 Detection was done using Amersham ECL Western Blot- times the interquartile range of all features on the array; and ting Detection System (GE Healthcare) and Amersham feature replicate SD ≤ 0.25 per array. Normalization of raw Hyperfilm ECL (GE Healthcare). Developed films were signals was done using variance stabilization normalization scanned using an Epson Perfection V700 Photo (Seiko Ep- (37). Probes with >40% missing values across all samples son Corp.). Uncropped images were contrast enhanced were removed. Based on BLASTing the probes sequence in- and subsequently cropped using Photoshop CS2 Version formation against the genome, biological annotations were 9.0 (Adobe Systems). retrieved from EnsEMBL (version 54; NCBI Build 36 of the human genome reference sequence). Sample preparation, Cell number, cell growth kinetic, and viability hybridization, and data analysis of the second separate me- Cells were seeded in 6-well plates 24 h before transfec- dulloblastoma patient set was done as described (31). tion. Transfection was done as described above. Cells were collected at indicated time points, and cell numbers were Preparation of medulloblastoma tissue microarray, measured using a Z2 Series Coulter Counter (Beckman immunohistochemical staining, and fluorescence Coulter). Growth kinetic curves were plotted and doubling in situ hybridization times calculated using GraphPad Prism version 3.03 for The medulloblastoma tissue microarray was prepared Windows (GraphPad Software). Viability was determined from blocks of patient material as described (4). All using trypan blue exclusion staining.

3242 Clin Cancer Res; 16(12) June 15, 2010 Clinical Cancer Research

HDAC5 and HDAC9 in Medulloblastoma

Measurement of the sub-G0 fraction and Results caspase-3–like activity The sub-G0 fraction of cultured cells was measured High HDAC5 and HDAC9 mRNA expression is as described (13) using Nicoletti stain and the flow associated with poor prognosis in medulloblastoma cytometer FACS Canto II (Beckman Coulter). For data To determine the expression of HDAC family members analysis, FACSDiva (version 6.1.2; Beckman Coulter) in medulloblastoma with unfavorable versus favorable was used. Caspase-3–like activity of cultured cells was clinical outcome in a screening approach, we did an measured using the Caspase-3 Fluorometric Assay Kit mRNA expression profiling of a small set of pooled patient (Biovision, Inc.) according to manufacturer's instruc- samples with either chromosome 6q loss (a marker for fa- tions. The positive control for caspase-3–like activity vorable prognosis; n = 5 samples) or gain of chromosome measurements consisted of untransfected cells treated 6q (a marker for poor prognosis; n = 4 samples; Fig. 1A) or with UV light (35 mJ/cm2) 16 h before caspase-3–like pools of patients with balanced chromosome 17 status activity measurement. (correlating with favorable prognosis; n = 11 samples) or 17q gain (defined as having either an isochromosome Statistical analysis i17q or a gain of chromosome arm 17q, correlating with Statistical analysis was done using GraphPad Prism poor prognosis; n = 10 samples; Fig. 1B). These molecular version 3.03 for Windows (GraphPad Software) and R markers on chromosome 6 and 17 have recently been (R version 2.4.1, 2006; The R Foundation for Statistical identified as powerful outcome predictors of prognosis Computing) with the package maxstat (40) as follows: in medulloblastoma patients (4). Our screening revealed GraphPad Prism, nonparametric Mann-Whitney U test of HDAC5 and HDAC9 to be highly expressed in the prog- quantitative real-time RT-PCR measurements of HDAC5 nostic unfavorable groups (Fig. 1A and B). HDACs that and HDAC9 in patient samples and of gene expression were found downregulated in the prognostic unfavorable measurements in the validation cohort; and R, Kaplan- groups are HDAC4 and HDAC1 to a lesser extent. Meier survival analysis and log-rank statistics, cut-point To confirm the mRNA expression of HDAC5 and analysis of quantitative real-time RT-PCR measurements HDAC9 in a set of 37 individual samples of patients with of HDAC5 and HDAC9 in patient samples using maximal- medulloblastoma, we used quantitative real time RT-PCR. ly selected rank statistics to determine the value separating Tumors harboring a gain of chromosome 6q again showed a group into two groups with the most significant differ- a significantly higher mRNA expression of HDAC5 and of ence when used as a cut-point, and ANOVA analysis of cell HDAC9 when compared with tumors with 6q deletion P P numbers, sub-G0 fraction, and caspase-3 activity after ( < 0.05 and < 0.05; Fig. 1C and D). Tumors with bal- knockdown of HDAC5 and HDAC9 using a linear mixed anced 6q also showed a significantly higher HDAC5 and model with fixed factors (“siRNA against” and “no. of HDAC9 mRNA expression level when compared with tu- siRNA,” and random intercept for the “number of mea- mors with 6q deletion (P < 0.05 and P < 0.005; Fig. 1C surement”). Grouping of patients according to median and D). The same was true when patients were grouped of quantitative real-time RT-PCR measurements was done using the DNA copy-number status of chromosome 17q; as follows: HDAC5 ≤ 0.33, HDAC5 low; HDAC5 > 0.33, tumors exhibiting gain of 17q revealed a significantly HDAC5 high; HDAC9 ≤ 0.87, HDAC9 low; and HDAC9 > higher expression of HDAC5 and of HDAC9 mRNA when 0.87, HDAC9 high. Grouping according to calculated compared with tumors displaying a balanced chromo- optimal cut-points was done as follows: HDAC5 ≤ 0.5, some 17 status (P < 0.05 and P < 0.005; Fig. 1E and F). HDAC5 low; HDAC5 > 0.5, HDAC5 high; HDAC9 ≤ 1.3, We therefore conclude that medulloblastoma with either HDAC9 low; and HDAC9 > 1.3, HDAC9 high. Combined balanced chromosome 6 or gain of chromosome 6q or HDAC5 and HDAC9 grouping according to calculated op- 17q have a higher HDAC5 and higher HDAC9 mRNA ex- timal cut-points was done as follows: HDAC5 ≤ 0.5 and pression level than tumors with chromosome 6q deletion HDAC9 ≤ 1.3 (group A), HDAC5 ≤ 0.5 and HDAC9 > or balanced status of 17q. 1.3 or HDAC5 >0.5andHDAC9 ≤ 1.3 (group B), and To investigate the potential of HDAC5 and HDAC9 HDAC5 > 0.5 and HDAC9 > 1.3 (group C). The stratified mRNA expression to predict the survival of medulloblasto- Cox-regression model was calculated in R (R version ma patients, we did a log-rank analysis. When analyzed for 2.7.1, 2008-06-23; The R Foundation for Statistical overall survival using the median as a cut-point, the groups Computing). A stratified Cox-model analysis was used displayed statistically significant differences in overall sur- to determine prognostic factors in a multivariate analysis vival probability for HDAC5 and HDAC9 (P < 0.05 and P < with HDAC5 and HDAC9 dichotomized at the previous- 0.005; log-rank test; Supplementary Fig. S1). Therefore, ei- ly determined cut-points. Because no deaths were ob- ther HDAC5 or HDAC9 mRNA expression separates the served in the group with chromosome 6q loss or with patients into two groups with distinct overall survival. desmoplastic histology, the Cox-regression model strati- To analyze the potential of combined HDAC5 and fied for chromosome 6 and histology was fitted. This HDAC9 expression data to separate groups of patients approach allows for different baseline hazards functions with distinct overall survival probabilities, we combined for the combinations of chromosome 6 and histology HDAC5 and HDAC9 expression data, separating the categories. patients into three groups. First, we used optimal cut-point

www.aacrjournals.org Clin Cancer Res; 16(12) June 15, 2010 3243

Milde et al.

Fig. 1. Increased HDAC5 and HDAC9 mRNA expression in primary medulloblastoma samples with unfavorable prognosis. A, a pool of n = 4 patient samples with gain of chromosome 6q (chr 6q gain) was compared with a pool of n = 5 patient samples with loss of chromosome 6q (chr 6q loss). The log 2 ratios for each HDAC represented on the array chip were calculated by expression value of chr 6q gain pool divided by the expression value of chr 6q loss pool. HDACs shown to the left are upregulated in the pool with poorer prognosis; HDACs shown to the right are downregulated in the pool with poorer prognosis. B, a pool of n = 10 patient samples with 17q gain (chr 17q gain) was compared with a pool of n = 11 patient samples with balanced chromosome 17 status (chr 17 bal); log 2 ratios for each HDAC represented on the array chip were calculated by expression value of 17q gain pool divided by expression value of chromosome 17 balanced pool. HDACs shown to the left are upregulated in the pool with poor prognosis; HDACs shown to the right are downregulated in the pool with poor prognosis. C-F, HDAC5 and HDAC9 mRNA expression was measured by quantitative real time RT-PCR in n = 37 individual samples of medulloblastoma patients, which were grouped according to chromosome 6q (C and D) and chromosome 17 status (E and F). Relative expression is normalized to normal cerebellum RNA. C and D, patients grouped according to chromosome 6q showed a significantly higher expression of HDAC5 (C) and of HDAC9 (D) in the prognostically unfavorable group with gain of chromosome 6q when compared with patients with loss of chromosome 6q (P = 0.0152 for HDAC5 and P = 0.0260 for HDAC9;Mann-WhitneyU test). HDAC5 and HDAC9 expression was also significantly higher in the group with balanced chromosome 6 status when compared with the prognostically favorable group with loss of chromosome 6q (P = 0.0108 for HDAC5 and P = 0.0025 for HDAC9;Mann-WhitneyU test). E and F, when patients were grouped according to the status of chromosome 17q, the prognostically unfavorable group with 17q gain (either gain on 17q or i17q) exhibited a significantly higher expression of HDAC5 (E) and HDAC9 (F; P = 0.0051 for HDAC5 and P = 0.0024 for HDAC9;Mann-WhitneyU test). Chr, chromosome; bal, balanced. *, P < 0.05; **, P < 0.001; ***, P < 0.0001.

3244 Clin Cancer Res; 16(12) June 15, 2010 Clinical Cancer Research

HDAC5 and HDAC9 in Medulloblastoma

analysis to identify the optimal cut-point separating the patients in two groups with high or low expression levels of either HDAC5 or HDAC9. For overall survival, the optimal cut-point for HDAC5 was 0.5, however, without reaching statistical significance in separating the two groups after correction for overfitting (Fig. 2A). For HDAC9, the optimal cut-point for overall survival was 1.3, dividing the two groups with high statistical significance after correction for overfitting (P < 0.001; log-rank test; Fig. 2B). In a second step, patients were grouped according to optimal cut-points as follows (Fig. 2C): a, HDAC5 and HDAC9 low; b, HDAC5 low and HDAC9 high or HDAC5 high and HDAC9 low; and c, HDAC5 high and HDAC9 high. Kaplan-Meier analysis revealed significant differences about survival probabilities upon subgrouping (P < 0.0001; log-rank test). In summary, patients expressing low HDAC5 and- low HDAC9 had a significantly higher overall survival probability than patients with either HDAC5 or HDAC9 high expression, especially than patients with HDAC5 and HDAC9 high expression. We therefore conclude that the level of mRNA expression of HDAC5 and HDAC9 may be a candidate for risk stratification of patients with medulloblastoma. Using a stratified Cox-regression analysis, we determined independent prognostic factors in a multivariate analysis in our patient cohort, including the parameters age, metastatic stage at diagnosis, extent of surgical resection, histology, DNA copy-number status of chromosome 6 and 17, and HDAC5 and HDAC9 mRNA expression (Table 1). Combined analysis of HDAC5 and HDAC9 was carried out in the same manner as done in the survival analysis. High expression of HDAC5 and HDAC9 (group c) proved to be a significant independent risk factor when compared with the group with low expression for HDAC5 and HDAC9 (group a; P < 0.005), as well as when compared with the group with high expression of only HDAC5 or HDAC9 (group b; P <0.05; Table 1). Thus, we conclude that the combination of high expression of HDAC5 and HDAC9 may be of prognostic significance in medulloblastoma.

HDAC5 and HDAC9 expression correlates with prognostic markers in an independent large cohort of medulloblastoma patients Fig. 2. HDAC5 and HDAC9 mRNA expression is associated with survival To test if HDAC5 and HDAC9 expression correlates with probability in medulloblastoma patients. Kaplan-Meier analysis of prognostic markers in a separate validation set of medul- survival of medulloblastoma patients grouped according to HDAC5 and HDAC5 HDAC9 HDAC9 expression as measured by quantitative real-time RT-PCR. loblastoma samples, we measured and A and B, after doing optimal cut-point analysis for HDAC5 and for HDAC9 expression in an independent set of n = 103 medulloblas- using maximally selected rank statistics, patients were grouped into two toma samples using gene expression arrays (31). Tumors groups for each HDAC5 (cut-point = 0.5) and HDAC9 (cut-point = 1.3). The with balanced chromosome 6 status showed a signifi- difference in overall survival probability between HDAC5 low versus HDAC5 HDAC9 HDAC5 cantly higher mRNA expression for and high expressing patients (A) showed a trend toward significance P after correction for overestimation (P = 0.08; log rank). The difference in when compared with patients with 6q deletion ( < 0.05 overall survival probability between HDAC9 low and HDAC9 high and P < 0.0005; Supplementary Fig. S2A and B). Tumors expressing patients (B) was statistically significant after correction for with a gain of chromosome 6q showed a significantly overfitting (P = 0.0008; log rank). C, when three groups were formed using higher mRNA expression of HDAC9 when compared the optimal cut-points for HDAC5 and HDAC9 as determined by P cut-point analysis, the group with patients expressing HDAC5 and with tumors with 6q deletion ( < 0.05; Supplementary HDAC9 at low levels (group a) showed a significantly better overall survival Fig. S2B). For HDAC5, expression was also higher in probability compared with the two other groups (P = 0.00004; log rank). tumors with gain of chromosome 6q when compared with

www.aacrjournals.org Clin Cancer Res; 16(12) June 15, 2010 3245

Milde et al.

loss of chromosome 6q, however, without reaching statis- HDAC5 and HDAC9 protein expression and cellular tical significance (Supplementary Fig. S2A). Tumors localization in primary medulloblastoma samples exhibiting a gain of 17q showed a significantly higher To investigate if HDAC5 and HDAC9 are also expressed expression of HDAC5 mRNA when compared with tumors at the protein level and could thus represent potential drug displaying a balanced chromosome 17 status (P < 0.0001; targets, we stained sections of a medulloblastoma tissue Supplementary Fig. S2C). However, no difference in microarray for HDAC5 and HDAC9. In addition, we stud- HDAC9 mRNA expression was found between these two ied the cellular localization because both proteins belong groups (Supplementary Fig. S2D). These data confirm the to class IIa HDACs known to shuttle between nucleus and association of high HDAC5 expression with prognosti- cytoplasm. HDAC5 was predominantly located in the cally unfavorable chromosome 17q gain and high nucleus (Fig. 3A; Supplementary Table S4); HDAC9 was HDAC9 expression with prognostically unfavorable gain primarily located in the cytoplasm (Fig. 3B; Supplemen- on chromosome 6. tary Table S4). Overall, >95% of cells stained positive Previous publications have shown that, in medullo- for either HDAC5 or HDAC9 protein (Supplementary blastoma, monosomy 6 is concurrent with activating Table S4). For n = 125 samples, both HDAC5 and HDAC9 mutations of CTNNB1 (41) and all of the tumors with staining was available. Samples were evaluated for nuclear chromosome 6 loss investigated have an activation of HDAC5 staining and cytosolic HDAC9 staining. More than the WNT pathway (42). We therefore went on to group 74% of the tumor samples showed a strong immunoreac- patients according to their gene expression signatures, that tivity for HDAC5 and HDAC9 (Supplementary Table S4). is, WNT and sonic hedgehog (SHH) pathway signatures Of note, staining for HDAC5 and HDAC9 was stronger in and signatures C and D as published previously (31, 41). desmoplastic nodules than in the surrounding tissue in HDAC5 expression correlated significantly with molecular most patients with desmoplastic medulloblastoma (4 of 7 subgroups, that is, it was lowest in the groups with WNT tumors and 6 of 8 tumors, respectively; Fig. 3A and B). In and SHH signatures and highest in groups C and D (Supple- summary,wewereabletoshowstrongHDAC5and mentary Fig. S2E). HDAC9 expression showed a similar HDAC9 protein levels in most primary medulloblastoma trend, with the expression being significantly lowest in samples and show HDAC5 to be localized predominantly the WNT group when compared with group C or D but in the nucleus, whereas HDAC9 is mostly localized in the showed no statistically significant difference between the cytoplasm. Considering HDAC5 and HDAC9 as potentially SHH group and group C or D (Supplementary Fig. S2F). “druggable” proteins, >95% of the medulloblastoma In conclusion,these data show a statistically significant tumors are positive for these targets. association of high HDAC5 expression with chromosome 17q gain and of high HDAC9 expression with chromo- siRNA-mediated knockdown of HDAC5 and HDAC9 some 6 gain, both correlating with poor prognosis, in a in medulloblastoma cell lines reduces cell growth second independent sample set. Furthermore, HDAC5 and viability and HDAC9 expression significantly correlates with mo- To investigate whether HDAC5 and HDAC9 are of lecular subgroups characterized by distinct gene expres- functional relevance in medulloblastoma cells, we chose sion signatures. an in vitro cell culture model using siRNA-mediated

Table 1. Stratified Cox regression model and hazard ratio (n = 37)

Variable (multivariate analysis) Effect Hazard ratio P

Age <4 vs ≥4 y 6.26 NS

Metastasis M2 or M3 vs M0 or M1 1.40 NS Level of resection STR vs GTR 28.12 <0.05 Histology Large cell anaplastic vs classic or desmoplastic 1.99 NS Chromosome 6 6 Bal vs monosomy 6 n/a n/a 6 Gain vs monosomy 6 n/a n/a Chromosome 17 17q gain vs 17q bal 10.79 NS HDAC5 and HDAC9 expression Group c (high/high) vs group a (low/low) 0.02 <0.005 Group c (high/high) vs group b (high/low) 0.02 <0.05

NOTE: Hazard ratio prediction using a stratified Cox regression model revealed that high expression of HDAC5 and HDAC9 is a significant risk factor. The only other significant risk factor found in this model using the combined analysis proved to be level of resection. Because no deaths were observed in the groups with 6q loss and the group with M stage 0, no HR can be estimated for these groups. Abbreviations: STR, subtotal resection; GTR, gross total resection; bal, balanced; NS, not significant; n/a, not available.

3246 Clin Cancer Res; 16(12) June 15, 2010 Clinical Cancer Research

HDAC5 and HDAC9 in Medulloblastoma

Fig. 3. HDAC5 and HDAC9 protein expression in primary medulloblastoma samples. A and B, two representative medulloblastoma samples are shown for each of the three standard histologic categories: classical, large cell anaplastic (LCA) and desmoplastic medulloblastoma. Original magnification, ×100 (left) and ×400× (right). Pink staining, HDAC5 (A) or HDAC9 (B); blue staining, nuclei. The cellular staining pattern for HDAC5 was predominantly nuclear, and for HDAC9, it was predominantly cytoplasmic.

knockdown of HDAC5 or HDAC9 expressioninestab- a reduction in cell growth was seen, with the strongest ef- lished medulloblastoma cell lines Daoy, UW228-2, fect seen in all cell populations with knockdown of UW228-3, ONS76, and Med8A. After transient transfec- HDAC5 but also in two out of three populations with tion with three different siRNAs against each HDAC5 or knockdown of HDAC9 (Fig. 4A). Accordingly, the dou- HDAC9, medulloblastoma cell lines showed a knockdown bling time was increased up to 2.04-fold (Fig. 4B). Because WAF1/CIP1 of up to 80% of HDAC5 and HDAC9 mRNA expression we did not observe p21 mRNA induction (data after 72 hours (Supplementary Fig. S3A). Western blot not shown), a marker typically associated with cell cycle confirmed the knockdown of HDAC5 and HDAC9 protein inhibition upon HDAC inhibition, we went on to investi- expression at 72-hour exemplary in Daoy cells (Supple- gate viability and cell death after knockdown of HDAC5 or mentary Fig. S3B). When cell counts were measured for HDAC9. Following knockdown of HDAC5 or HDAC9,we Daoy cells over the course of 0 to 5 days after transfection, observed a significant increase of up to 5-fold in trypan

www.aacrjournals.org Clin Cancer Res; 16(12) June 15, 2010 3247

Milde et al.

blue–positive cells (P < 0.005 to P < 0.0001; Fig. 5A). Of induced the weakest reduction in HDAC9 protein level, note, knockdown efficacy paralleled the extent of viability also showed the least effects on subsequent cell counts, decrease, that is, Med8A cell exhibiting the highest remain- sub-G0, and caspase-3–like activity analyses, indicating a ing HDAC5 and HDAC9 mRNA levels after knockdown al- dose-response relationship. In summary, knockdown of so showed the least increase in trypan blue-positive cells. HDAC5 or HDAC9 reduces cell growth and viability of Flow-cytometric analysis of propidium iodide–stained medulloblastoma cells in vitro, associated with induction Daoy cells showed a significant increase in sub-G0 fraction of apoptosis. following knockdown of HDAC5 or HDAC9 (up to 34% and 31%; P <0.05andP < 0.05; Fig. 5B), suggestive of Discussion apoptosis. We therefore determined caspase-3–like activity in Daoy cells as an indicator for apoptosis. Knockdown of The treatment of medulloblastoma patients is still HDAC5 and HDAC9 resulted in increased caspase-3–like challenging in terms of long-term survival, as well as neu- activity up to 2.65-fold, suggestive of apoptosis induction rologic, cognitive, and endocrinological sequelae of che- (P < 0.05; Fig. 5C). Of note, HDAC9 siRNA 2, which motherapy- and radiation-based treatment protocols. To

Fig. 4. Knockdown of HDAC5 and HDAC9 reduces cell population growth and increases doubling time. A, cell counts of Daoy cells were measured at 0 hour and every 24 hours for 5 days after transfection with three different siRNAs against HDAC5 (HDAC5 siRNA 1-3) or HDAC9 (HDAC9 siRNA 1-3), two different siRNA controls (negative ctrl 1 and 2) and transfection reagent only (transfection ctrl; representative plot). Cell population growth was decreased in cells with knockdown of either HDAC5 or HDAC9. B, doubling times (days; means and SD from four independent measurements) were calculated for Daoy cells after siRNA-mediated knockdown of HDAC5 and HDAC9. The increase in doubling time was significant for cells with knockdown of HDAC5 (P = 0.0361) but not for cells with knockdown of HDAC9 (P = 0.0592; ANOVA). However, when only the two HDAC9 siRNAs with sufficient reduction of HDAC9 protein (HDAC9 siRNAs 1 and 3; Supplementary Fig. S3B) were compared with the negative controls; the difference in doubling time was statistically significant (P = 0.0272; ANOVA). ctrl, control; NS, not significant; n/a, not available.

3248 Clin Cancer Res; 16(12) June 15, 2010 Clinical Cancer Research

HDAC5 and HDAC9 in Medulloblastoma

Fig. 5. Knockdown of HDAC5 and HDAC9 decreases viability in medulloblastoma cell lines. A, the percentage of dead cells as determined by trypan blue exclusion staining was increased in a statistically significant manner 72 hours after transfection with three different siRNAs against HDAC5 (HDAC5 siRNA 1-3) or HDAC9 (HDAC9 siRNA 1-3) compared with two different siRNA controls (negative ctrl 1 and 2) in five different cell lines (P = 0.0014 to

P < 0.0001; ANOVA). B, measurement of sub-G0 fraction 72 hours after knockdown of HDAC5 or HDAC9 (Nicoletti method) in Daoy cells. The sub-G0 fraction is significantly increased (P = 0.0092 and P = 0.0266; ANOVA). C, caspase-3–like activity is significantly increased in Daoy cells 72 hours after knockdown of HDAC5 (P = 0.0115; ANOVA), Daoy cells with knockdown of HDAC9 showed an increase in caspase-3 activity 72 hours after knockdown, however, without reaching statistical significance, because of the lack of caspase-3–like activity increase in one of three siRNAs against HDAC9. When only the two HDAC9 siRNAs with sufficient reduction of HDAC9 protein (HDAC9 siRNAs 1 and No.3; Supplementary Fig. S3B) were compared with the negative controls, the difference in caspase-3–like activity was statistically significant (P < 0.005; ANOVA). Bars in A-C, averages from at least three independent measurements; error bars, SD. *, P < 0.05; **, P < 0.001; ***, P < 0.0001.

www.aacrjournals.org Clin Cancer Res; 16(12) June 15, 2010 3249

Milde et al.

explore the possibility of treatment with selective HDACi, physiologic roles in differentiation and cell cycle regula- we studied the expression of individual HDAC family tion. Data about HDAC5 and HDAC9 function in cancer members in medulloblastoma and found HDAC5 and however are scarce. In mouse erythroleukemia cells, HDAC9 to be highly expressed in prognostically unfavor- HDAC5 interacts with the transcription factor GATA bind- able subgroups. Of note, because HDAC5 is located on ing protein 1 (GATA1) and shuttles from the nucleus to chromosome arm 17q, the most frequently gained the cytoplasm upon erythroid differentiation of leukemic genomic region in medulloblastoma, this might well cells (51), indicating a role of HDAC5 in differentiation contribute to a higher expression of HDAC5 through a processes of malignant cells as well. Our data suggest gene-dosage effect. Significant upregulation of HDAC5 in that HDAC5 and HDAC9 harbor oncogenic function in tumors harboring a 17q gain as revealed by our study medulloblastoma cells because knockdown inhibits cell underlines the functional relevance of this candidate gene growth and reduces viability in these cells. On the pro- in medulloblastoma biology. With regard to HDAC9,we tein level, HDAC5 and HDAC9 were widely expressed in did not see a significant difference in mRNA levels be- primary medulloblastoma, with a nodular pattern in a tween tumors exhibiting 17q gain and balanced chromo- subset of desmoplastic tumors. This and our data from some 17 status in the second cohort as opposed to the first cell culture experiments indicate that the higher HDAC5 cohort. We believe this is due to the difference in the com- and HDAC9 expression in the nodular area may have an position of the two 17q balanced groups; the first patient antiapoptotic function. cohort contains more patients with activated WNT path- Based on our observation that (a)highHDAC5 and way than the second cohort. Because low HDAC9 expres- HDAC9 expression correlates with poor prognostic sub- sion seems to be mostly attributable to the WNT group, groups and is associated with poor overall survival, (b) the comparison of 17q balanced versus 17q gain in the high expression of HDAC5 and HDAC9 is an independent first cohort shows a difference, whereas the second cohort risk factor, and (c) knockdown of HDAC5 or HDAC9 re- does not. duces cell number, decreases viability, and induces apopto- All classic HDACs (HDAC1-HDAC11) are widely ex- sis in medulloblastoma cells, we propose that HDAC5 and pressed in the vertebrate developing and adult brain as HDAC9 play a major role in medulloblastoma biology. It is has been shown in rats and mice (43–45), with HDAC1 currently under debate whether class IIa HDACs function being expressed in neural stem cells (45). Thus far, only through their own enzymatic activity or display a rather a few studies have systematically examined the expression low enzymatic activity (52, 53) and act in complex only of all 11 classic HDAC family members in primary tumors with class I HDACs (23). The lack of specific HDAC5 or in general and tumors of neural origin in particular. In HDAC9 inhibitors currently prevents the testing of strate- cancers of the gastrointestinal system, high HDAC1, gies involving selective targeting of these HDACs. HDAC2, and HDAC3 expression correlated with poor clin- In summary, we have identified HDAC5 and HDAC9 as ical outcome (29, 30). In neuroblastoma, among all potential novel prognostic markers for medulloblastoma. HDAC family members investigated, only HDAC8 was as- Our functional data furthermore warrants further investi- sociated with advanced-stage disease and poor prognosis gation of selective targeting of HDAC5 and HDAC9 as a (14). Recently, class II and IV HDACs were found down- novel strategy for medulloblastoma treatment. regulated in glioblastoma compared with low-grade astro- cytoma and normal brain (46). Therefore, expression of individual HDAC family members seems to be tumor Disclosure of Potential Conflicts of Interest specific. In our study, we show for the first time that the HDAC5 HDAC9 HDAC class IIa isoenzymes and are No potential conflicts of interest were disclosed. associated with clinical outcome in a malignant disease, not only correlating with survival but furthermore posing Acknowledgments an independent risk factor. Prospective studies will be needed to confirm the prospective value of HDAC5 and We thank Sandra Riedinger, Carina Konrad, Sarah Engelhardt, Cornelia HDAC9 mRNA expression levels in the risk stratification Rütz, and Diana Jäger for the excellent technical assistance. of medulloblastoma patients. Thus far, little is known about the physiologic function Grant Support of HDAC5 and HDAC9 in normal cells and in development. Both HDACs seem to play central roles in modula- The NGFNplus program by a grant of the Bundesministerium für Bil- dung und Forschung, Germany (H.E. Deubzer and O. Witt), the University tion of cardiac stress signals and cardiac development. of Heidelberg through the FRONTIER and the OLYMPIA MORATA pro- Mouse knockout models of HDAC5 and HDAC9 produce grams (H.E. Deubzer), and a grant from the Wilhelm Sander Foundation cardiac phenotypes similar to each other (47, 48), suggest- (T. Milde and I. Oehme). The costs of publication of this article were defrayed in part by the ing similar functions of the two isoenzymes in the physi- payment of page charges. This article must therefore be hereby marked ologic setting. Furthermore, HDAC5 shuttles from the advertisement in accordance with 18 U.S.C. Section 1734 solely to nucleus to the cytoplasm when myoblasts differentiate indicate this fact. (49), and in fibroblasts, HDAC5 repressed the transcrip- Received 02/13/2010; revised 04/14/2010; accepted 04/15/2010; tion of cyclin D3, a cell cycle activator (50), suggesting published OnlineFirst 04/22/2010.

3250 Clin Cancer Res; 16(12) June 15, 2010 Clinical Cancer Research

HDAC5 and HDAC9 in Medulloblastoma

References 1. Kaatsch P, Spix C. German Childhood Cancer Registry—annual re- 23. Haberland M, Montgomery RL, Olson EN. The many roles of histone port 2006/2007 (1980-2006). Mainz (Germany): Deutschen Kinderk- deacetylases in development and physiology: implications for dis- rebsregisters, Institut für Medizinische Biometrie, Epidemiologie und ease and therapy. Nat Rev Genet 2009;10:32–42. Informatik, Universität Mainz, Germany; 2008. 24. Knutson SK, Chyla BJ, Amann JM, Bhaskara S, Huppert SS, Hiebert 2. Taylor RE, Bailey CC, Robinson KJ, et al. Outcome for patients with SW. Liver-specific deletion of histone deacetylase 3 disrupts metastatic (M2-3) medulloblastoma treated with SIOP/UKCCSG metabolic transcriptional networks. EMBO J 2008;27:1017–28. PNET-3 chemotherapy. Eur J Cancer 2005;41:727–34. 25. Witt O, Deubzer HE, Milde T, Oehme I. HDAC family: what are the 3. Packer RJ, Gajjar A, Vezina G, et al. Phase III study of craniospinal cancer relevant targets? Cancer Lett 2009;277:8–21. radiation therapy followed by adjuvant chemotherapy for newly diag- 26. Lane AA, Chabner BA. Histone deacetylase inhibitors in cancer ther- nosed average-risk medulloblastoma. J Clin Oncol 2006;24:4202–8. apy. J Clin Oncol 2009;27:5459–68. 4. Pfister S, Remke M, Benner A, et al. Outcome prediction in pediatric 27. OehmeI,DeubzerHE,LodriniM,MildeT,WittO.Targetingof medulloblastoma based on DNA copy-number aberrations of chro- HDAC8 and investigational inhibitors in neuroblastoma. Expert Opin mosomes 6q and 17q and the MYC and MYCN loci. J Clin Oncol Investig Drugs 2009;18:1605–17. 2009;27:1627–36. 28. Haberland M, Mokalled MH, Montgomery RL, Olson EN. Epigenetic 5. Copeland DR, deMoor C, Moore BD III, Ater JL. Neurocognitive de- control of skull morphogenesis by histone deacetylase 8. Genes Dev velopment of children after a cerebellar tumor in infancy: a longitudi- 2009;23:1625–30. nal study. J Clin Oncol 1999;17:3476–86. 29. Weichert W, Roske A, Gekeler V, et al. Association of patterns of 6. Palmer SL, Gajjar A, Reddick WE, et al. Predicting intellectual out- class I histone deacetylase expression with patient prognosis in come among children treated with 35-40 Gy craniospinal irradiation gastric cancer: a retrospective analysis. Lancet Oncol 2008;9: for medulloblastoma. Neuropsychology 2003;17:548–55. 139–48. 7. Bolden JE, Peart MJ, Johnstone RW. Anticancer activities of histone 30. Weichert W, Roske A, Niesporek S, et al. Class I histone deacetylase deacetylase inhibitors. Nat Rev Drug Discov 2006;5:769–84. expression has independent prognostic impact in human colorectal 8. Deubzer H, Busche B, Ronndahl G, et al. Novel valproic acid deriva- cancer: specific role of class I histone deacetylases in vitro and tives with potent differentiation-inducing activity in myeloid leukemia in vivo. Clin Cancer Res 2008;14:1669–77. cells. Leuk Res 2006;30:1167–75. 31. Northcott PA, Fernandez LA, Hagan JP, et al. The miR-17/92 poly- 9. Lindemann RK, Newbold A, Whitecross KF, et al. Analysis of the ap- cistron is up-regulated in sonic hedgehog-driven medulloblastomas optotic and therapeutic activities of histone deacetylase inhibitors by and induced by N-myc in sonic hedgehog-treated cerebellar neural using a mouse model of B cell lymphoma. Proc Natl Acad Sci U S A precursors. Cancer Res 2009;69:3249–55. 2007;104:8071–6. 32. Pfister S, Janzarik WG, Remke M, et al. BRAF gene duplication con- 10. Komatsu N, Kawamata N, Takeuchi S, et al. SAHA, a HDAC inhibitor, stitutes a mechanism of MAPK pathway activation in low-grade as- has profound anti-growth activity against non-small cell lung cancer trocytomas. J Clin Invest 2008;118:1739–49. cells. Oncol Rep 2006;15:187–91. 33. Solinas-Toldo S, Lampel S, Stilgenbauer S, et al. Matrix-based com- 11. Platta CS, Greenblatt DY, Kunnimalaiyaan M, Chen H. The HDAC parative genomic hybridization: biochips to screen for genomic im- inhibitor trichostatin A inhibits growth of small cell lung cancer cells. balances. Genes Chromosomes Cancer 1997;20:399–407. J Surg Res 2007;142:219–26. 34. Zielinski B, Gratias S, Toedt G, et al. Detection of chromosomal im- 12. Dalgard CL, Van Quill KR, O'Brien JM. Evaluation of the in vitro balances in retinoblastoma by matrix-based comparative genomic and in vivo antitumor activity of histone deacetylase inhibitors for the hybridization. Genes Chromosomes Cancer 2005;43:294–301. therapy of retinoblastoma. Clin Cancer Res 2008;14:3113–23. 35. R: A language and environment for statistical computing. Vienna 13. Deubzer HE, Ehemann V, Westermann F, et al. Histone deacetylase (Austria): R Foundation for Statistical Computing; 2009. inhibitor Helminthosporium carbonum (HC)-toxin suppresses the 36. Gentleman RC, Carey VJ, Bates DM, et al. Bioconductor: open soft- malignant phenotype of neuroblastoma cells. Int J Cancer 2008; ware development for computational biology and bioinformatics. Ge- 122:1891–900. nome Biol 2004;5:R80. 14. Oehme I, Deubzer HE, Wegener D, et al. Histone deacetylase 8 in 37. Huber W, von Heydebreck A, Sultmann H, Poustka A, Vingron M. neuroblastoma tumorigenesis. Clin Cancer Res 2009;15:91–9. Variance stabilization applied to microarray data calibration and to 15. Svechnikova I, Almqvist PM, Ekstrom TJ. HDAC inhibitors effectively the quantification of differential expression. Bioinformatics 2002;18 induce cell type-specific differentiation in human glioblastoma cell Suppl 1:S96–104. lines of different origin. Int J Oncol 2008;32:821–7. 38. Pfister S, Remke M, Toedt G, et al. Supratentorial primitive neuroec- 16. Witt O, Schweigerer L, Driever PH, Wolff J, Pekrun A. Valproic acid todermal tumors of the central nervous system frequently harbor treatment of glioblastoma multiforme in a child. Pediatr Blood Cancer deletions of the CDKN2A locus and other genomic aberrations 2004;43:181. distinct from medulloblastomas. Genes Chromosomes Cancer 17. Furchert SE, Lanvers-Kaminsky C, Juurgens H, Jung M, Loidl A, 2007;46:839–51. Fruhwald MC. Inhibitors of histone deacetylases as potential thera- 39. Schmitt M, Pawlita M. High-throughput detection and multiplex iden- peutic tools for high-risk embryonal tumors of the nervous system of tification of cell contaminations. Nucleic Acids Res 2009;37:e119. childhood. Int J Cancer 2007;120:1787–94. 40. Hothorn T, Lausen B. On maximally selected rank statistics. R News 18. Sonnemann J, Kumar KS, Heesch S, et al. Histone deacetylase in- 2002;2:3–5. hibitors induce cell death and enhance the susceptibility to ionizing 41. Thompson MC, Fuller C, Hogg TL, et al. Genomics identifies medul- radiation, etoposide, and TRAIL in medulloblastoma cells. Int J Oncol loblastoma subgroups that are enriched for specific genetic altera- 2006;28:755–66. tions. J Clin Oncol 2006;24:1924–31. 19. Wegener D, Deubzer HE, Oehme I, et al. HKI 46F08, a novel po- 42. Clifford SC, Lusher ME, Lindsey JC, et al. Wnt/Wingless pathway ac- tent histone deacetylase inhibitor, exhibits antitumoral activity tivation and chromosome 6 loss characterize a distinct molecular against embryonic childhood cancer cells. Anticancer Drugs sub-group of medulloblastomas associated with a favorable progno- 2008;19:849–57. sis. Cell Cycle 2006;5:2666–70. 20. Mann BS, Johnson JR, Cohen MH, Justice R, Pazdur R. FDA ap- 43. Broide RS, Redwine JM, Aftahi N, Young W, Bloom FE, Winrow CJ. proval summary: vorinostat for treatment of advanced primary cuta- Distribution of histone deacetylases 1-11 in the rat brain. J Mol Neu- neous T-cell lymphoma. Oncologist 2007;12:1247–52. rosci 2007;31:47–58. 21. Witt O, Deubzer HE, Milde T, Oehme I. HDAC family: what are the 44. Liu H, Hu Q, Kaufman A, D'Ercole AJ, Ye P. Developmental expres- cancer relevant targets? Cancer Lett 2008. sion of histone deacetylase 11 in the murine brain. J Neurosci Res 22. Batty N, Malouf GG, Issa JP. Histone deacetylase inhibitors as anti- 2008;86:537–43. neoplastic agents. Cancer Lett 2009;280:192–200. 45. MacDonald JL, Roskams AJ. Histone deacetylases 1 and 2 are

www.aacrjournals.org Clin Cancer Res; 16(12) June 15, 2010 3251

Milde et al.

expressed at distinct stages of neuro-glial development. Dev Dyn export of a histone deacetylase regulates muscle differentiation. Na- 2008;237:2256–67. ture 2000;408:106–11. 46. Lucio-Eterovic AK, Cortez MA, Valera ET, et al. Differential expres- 50. Roy S, Shor AC, Bagui TK, Seto E, Pledger WJ. Histone deacetylase sion of 12 histone deacetylase (HDAC) genes in astrocytomas and 5 represses the transcription of cyclin D3. J Cell Biochem 2008;104: normal brain tissue: class II and IV are hypoexpressed in glioblasto- 2143–54. mas. BMC Cancer 2008;8:243. 51. Watamoto K, Towatari M, Ozawa Y, et al. Altered interaction of 47. Chang S, McKinsey TA, Zhang CL, Richardson JA, Hill JA, Olson EN. HDAC5 with GATA-1 during MEL cell differentiation. Oncogene Histone deacetylases 5 and 9 govern responsiveness of the heart to 2003;22:9176–84. a subset of stress signals and play redundant roles in heart develop- 52. Lahm A, Paolini C, Pallaoro M, et al. Unraveling the hidden catalytic ment. Mol Cell Biol 2004;24:8467–76. activity of vertebrate class IIa histone deacetylases. Proc Natl Acad 48. Zhang CL, McKinsey TA, Chang S, Antos CL, Hill JA, Olson EN. Sci U S A 2007;104:17335–40. Class II histone deacetylases act as signal-responsive repressors 53. Jones P, Altamura S, De Francesco R, et al. Probing the elusive cat- of cardiac hypertrophy. Cell 2002;110:479–88. alytic activity of vertebrate class IIa histone deacetylases. Bioorg 49. McKinsey TA, Zhang CL, Lu J, Olson EN. Signal-dependent nuclear Med Chem Lett 2008;18:1814–9.

3252 Clin Cancer Res; 16(12) June 15, 2010 Clinical Cancer Research

HDAC5 and HDAC9 in Medulloblastoma: Novel Markers for Risk Stratification and Role in Tumor Cell Growth

Till Milde, Ina Oehme, Andrey Korshunov, et al.

Clin Cancer Res Published OnlineFirst April 22, 2010.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-10-0395

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2010/04/22/1078-0432.CCR-10-0395.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://clincancerres.aacrjournals.org/content/early/2010/06/01/1078-0432.CCR-10-0395. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.