Oncogene (2011) 30, 2044–2056 & 2011 Macmillan Publishers Limited All rights reserved 0950-9232/11 www.nature.com/onc ORIGINAL ARTICLE Genome-wide screen reveals WNT11, a non-canonical WNT , as a direct target of ETS ERG

LH Mochmann1, J Bock1, J Ortiz-Ta´nchez1, C Schlee1, A Bohne1, K Neumann2, WK Hofmann3, E Thiel1 and CD Baldus1

1Department of and Oncology, Charite´, Campus Benjamin Franklin, Berlin, Germany; 2Institute for Biometrics and Clinical Epidemiology, Charite´, Campus Mitte, Berlin, Germany and 3Department of Hematology and Oncology, University Hospital Mannheim, Mannheim, Germany

E26 transforming sequence-related gene (ERG) is a current therapies in acute patients with poor transcription factor involved in normal hematopoiesis prognosis characterized by high ERG mRNA expression. and is dysregulated in leukemia. ERG mRNA over- Oncogene (2011) 30, 2044–2056; doi:10.1038/onc.2010.582; expression was associated with poor prognosis in a subset published online 17 January 2011 of patients with T-cell acute lymphoblastic leukemia (T-ALL) and acute myeloid leukemia (AML). Herein, a Keywords: ETS-related gene (ERG); WNT11; acute genome-wide screen of ERG target was conducted leukemia; ERG target genes; 6-bromoindirubin-3-oxime by chromatin immunoprecipitation-on-chip (ChIP-chip) in (BIO); morphological transformation Jurkat cells. In this screen, 342 significant annotated genes were derived from this global approach. Notably, ERG-enriched targets included WNT signaling genes: WNT11, WNT2, WNT9A, CCND1 and FZD7. Further- more, chromatin immunoprecipitation (ChIP) of normal Introduction and primary leukemia material also confirmed WNT11 as a target of ERG in six of seven The erythroblastosis virus E26 transforming sequence patient samples. A larger sampling of patient diagnostic (ETS) encodes the ETS-related gene (ERG) that has an material revealed that ERG and WNT11 mRNA were co- important physiological role in hematopoiesis (Loughran expressed in 80% of AML (n ¼ 30) and 40% in T-ALL et al., 2008), angiogenesis (Birdsey et al., 2008), vascular (n ¼ 30) bone marrow samples. Small interfering RNA (Ellett et al., 2009) and bone development (Iwamoto (siRNA)-mediated knockdown of ERG confirmed down- et al., 2007). ERG belongs to the highly conserved ETS transcription factor family that is defined by the ETS regulation of WNT11 transcripts. Conversely, in a tet-on 0 0 ERG-inducible assay, WNT11 transcripts were co- DNA-binding motif 5 -GGA(A/T)-3 (Sharrocks, 2001; stimulated. A WNT pathway agonist, 6-bromoindirubin- Wei et al., 2010). In hematopoietic development, ERG 3-oxime (BIO), was used to determine the effect of cell has long been postulated to function during early stages growth on the ERG-inducible cells. The addition of BIO of T-cell development as its mRNA abundance peaks resulted in an ERG-dependent proliferative growth and diminishes as cells undergo T-lineage commitment advantage over ERG-uninduced cells. Finally, ERG (Anderson et al., 1999). In addition, ectopic ERG induction prompted morphological transformation where- expression was shown to induce megakaryocytic differ- by round unpolarized K562 cells developed elongated entiation in human K562 cells (Rainis et al., 2005) and protrusions and became polarized. This morphological in hematopoietic progenitors ERG promotes expansion transformation could effectively be inhibited with BIO of megakaryocytes (Stankiewicz and Crispino, 2009). and with siRNA knockdown of WNT11. In conclusion, ERG function was further characterized by Loughran ERG transcriptional networks in leukemia converge on et al. in heterozygous mice harboring a missense muta- WNT signaling targets. Specifically, WNT11 emerged as tion that phenotypically displayed mild cytopenia in a direct target of ERG. Potent ERG induction promoted the B-cell compartment and notably had a reduction morphological transformation through WNT11 signals. of progenitor cells by 50%. Furthermore, homozygosity The findings in this study unravel new ERG-directed for the same mutation failed to establish definitive hemato- molecular signals that may contribute to the resistance of poiesis at the embryonic stage. Thus, normal ERG function was necessary to establish and maintain hematopoiesis. ERG dysregulation has been reported in solid tumors and hematological malignancies. Three fusion Correspondence: Dr CD Baldus, Department of Hematology and composed of ERG with TMPRSS2 (Tomlins et al., Oncology, Charite´, Campus Benjamin Franklin, Hindenburgdamm 30, 2005; Klezovitch et al., 2008), EWS (Sorensen et al., 12203 Berlin, Germany. E-mail: [email protected] 1994) or TLS (Kong et al., 1997) create oncogenic Received 14 July 2010; revised 2 November 2010; accepted 23 November proteins. The most frequent chromosomal fusion in 2010; published online 17 January 2011 prostate cancer consists of the 50-untranslated region of ERG genome-wide screen in acute leukemia LH Mochmann et al 2045 TMPRSS2 fused with 30-end ERG. This fusion has ChIP (C20) and with double (C20 and C17) precipitat- multiple variants, of which two of the most common ing antibodies for increased accessibility to multiple variants are associated with poor outcome (Narod et al., ERG epitopes (refer to Materials and methods for a 2008). In leukemia, ERG overexpression is also believed detailed description). Each ChIP with single and double to contribute to the molecular pathogenesis in a sub- precipitating antibodies was carried out in duplicate. set of T-cell acute lymphoblastic leukemia (T-ALL) and The in vivo assay allowed for the enrichment and acute myeloid leukemia (AML) patients. High ERG identification of ERG-bound DNA sequences that were expressers were associated with an inferior clinical subsequently hybridized to a high-resolution human outcome (Marcucci et al., 2005; Baldus et al., 2006). promoter chip. Duplicate ChIP-chips 1 and 3 (single- The pathogenesis of ERG was also observed in antibody ChIP) resulted in 13 070 and 4405 signifi- sublethally irradiated mice transplanted with ERG cant peaks, respectively. Duplicate ChIP-chips 2 and 4 transduced progenitor cells, whereby megakaryoblastic (double-antibody ChIP) resulted in 11 227 and 6630 leukemia developed (Salek-Ardakani et al., 2009). Thus, significant peaks, respectively. As expected, many several clinical and experimental studies indicate that significant peaks were detected at individual gene ERG contributes to the pathogenesis in cancer and promoters, which yielded 1683, 3066, 1304 and 973 leukemia; however, the underlying biological mechan- single genes, in ChIP-chips 1–4, respectively. Pooled isms are not yet fully understood. gene sets from ChIP-chips 1 and 3 are denoted as ChIP- To unravel the molecular function of ERG in acute chip I and pooled gene sets from ChIP-chips 2 and 4 are leukemia, we have conducted a genome-wide screen of denoted as ChIP-chip II. Finally, only overlapping genes ERG target genes. Chromatin immunoprecipitation- from ChIP-chips I and II were combined to condense on-chip (ChIP-chip) analyses of ERG candidate target the significant candidate gene pool to 342 gene annota- genes revealed that ERG may participate in a broader tions (Table 1). DAVID Functional Annotation Tool spectrum of biological signaling than previously des- (Laboratory of Immunopathogenesis and Bioinfor- cribed. ERG loss and gain of function experiments matics, Frederick, MD, USA) was used to characterize directly affected WNT11, a non-canonical WNT path- biological themes (Table 2). Statistically enriched gene way gene. Moreover, a proliferative growth advantage ontology categories suggest a broad functional role for was observed when ERG-induced cells were treated ERG that included developmental processes, multi- with WNT agonist 6-bromoindirubin-3-oxime (BIO) cellular processes, biological adhesion and biological and a glycogen synthase-3b (GSK-3b) inhibitor. Finally, regulation (P-value o0.05). Furthermore, ERG candi- co-expression of ERG and WNT11 stimulated morpho- date genes subjected to Ingenuity Pathway Analyses logical transformation of round hematopoietic cells to revealed an overlap of enriched WNT target genes in polarized cells with protrusions upon ERG induction. In several key developmental pathways (Table 3). addition, the elongation process of ERG-induced cells was effectively inhibited by the addition of BIO and with small interfering RNA (siRNA)-mediated knockdown Validation of enriched promoter regions and selection of WNT11. These findings show that in human leukemia of ERG target genes WNT11 is a direct target of ERG and highlight a role Based on the gene’s described relevance to hemato- for ERG in the . poiesis and leukemia in the scientific literature (versus a random calculated approach), 24 of the 342 enriched promoter regions were selected for further examination. MATCH algorithm was used to determine the signifi- Results cance and location of conserved ETS-binding motif (5-GGAA/T-3) in each of the 24 putative promoter Genome-wide screen of ERG candidate genes regions up to 2 kb from the transcription start site (TSS) ERG transcriptional networks in leukemia are un- (Table 4 and Supplementary Table 1). In all, 17 of 24 known. Thus, in order to construct ERG-related selected targets were confirmed by quantitative PCR, networks, a genome-wide screen by ChIP-chip was with at least twofold enrichment relative to total chro- conducted in a human T-cell leukemia line, Jurkat. ChIP matin in both experimental ChIPs (Figure 1a). Notably, was performed with two ERG-specific antibodies: single of these enriched promoter amplicons, genes WNT2,

Table 1 Overview of selected ERG candidate genes Precipitating antibody ChIP-chip hybridization Peaks above threshold (40.1) Single genes Pooled gene sets Total gene count

C20 ChIP-chip 1 13 070 1683 ChIP-chip I 342 annotated genes ChIP-chip 3 4405 1304 C17/20 ChIP-chip 2 11 227 3066 ChIP-chip II ChIP-chip 4 6630 973

Abbreviations: ChIP-chip, chromatin immunoprecipitation-on-chip; ERG, E26 transforming sequence-related gene. The table gives a description of significantly enriched peak number and single gene count calculated by sliding window analysis (ACME software) for each hybridization (ChIP-chips 1–4) and corresponding ChIP antibodies. The final count of 342 annotated genes was determined by selecting shared genes from ChIP-chip I and ChIP-chip II.

Oncogene ERG genome-wide screen in acute leukemia LH Mochmann et al 2046 WNT9A, WNT11, CCND1 and FZD7 are participating putative genes (refer to ranking list, Table 4) and genes in WNT signaling. Other enriched promoter significant peaks clustered at nearly matching genomic regions also included hematopoietic genes (CD7, EPO coordinates in ChIP-chips 1–4 (Supplementary Figure 1b). and CD14) and transcriptional regulation genes RBPJL, In contrast, ERG, and other randomly selected non- CDK9, TWIST1, HDAC4, RXRA and OLIG2. Also, target genes, had statistically insignificant or undetect- TMPRSS2, a component of TMPRSS2–ERG fusion able signal intensities for ChIP-chips 1–4. Thus, genes oncogene, was significantly enriched in ChIP-chip harboring statistically significant ETS-binding sites in experiments and was validated by PCR, implicating the proximal promoter regions, enrichment of the that ERG may bind the fusion TMPRSS2–ERG proximal promoter regions through quantitative PCR promoter to induce expression. In line with previous and significant multiple peak detection (threshold ERG ChIP analyses, GP1BB, encoding the megakar- limit40.1) in the respective promoter regions collec- yocytic CD42c marker, was a significantly enriched tively determined the 17 candidate genes as putative promoter region (Bastian et al., 1996). ERG targets. A survey of chip hybridization intensity peaks from four hybridizations (ChIP-chips 1–4) for the 17 enriched candidate genes showed several statistically signifi- ERG occupies the WNT11 promoter in primary T-ALL cant peaks mapping to similar genomic coordinates. and AML leukemia blasts For example, an overlay of hybridization intensity sig- Of the 17 candidate genes, WNT11, in parallel with nals in the WNT11 promoter region for ChIP-chips 1 ERG, has a developmental role in hematopoiesis and 2 showed similar intensity signals spanning similar (Brandon et al., 2000) and was implicated in prostate arrayed genomic regions in Figure 1b (refer to Supple- tumor malignancy (Zhu et al., 2004). Therefore, we mentary Figure 1a for an overlay of hybridizations, focused on WNT11 and examined whether ERG bound ChIP-chips 1–4). The overlay of intensity signals assured to the WNT11 promoter in primary hematopoietic cells. ChIP-chip reproducibility of WNT11 promoter enrich- ChIP was conducted with fresh bone marrow cells from ment with single and double ChIP precipitating ERG one normal donor, one T-ALL patient and five AML antibodies. Interestingly, WNT11 promoter had the patients (AML A, B, C, D and E). WNT11 promoter highest number of statistically significant peaks of the 24 enrichment was observed in six of seven ChIP bone marrow donors (normal donor, T-ALL, AML C, AML D and AML E) relative to total chromatin (Figure 2). Table 2 Functional annotation of ERG candidate genes However, a significant linear correlation between ERG Biological term Gene % P-value B–H mRNA expression and WNT11 promoter enrichment count P-value was not observed. We presumed this result was due to Developmental process 98 28 1.5E–6 1.7E–5 the small number of ChIP samples and due to the Multicellular organismal process 113 33 8.2E–7 1.9E–5 molecular heterogeneity of acute leukemia. To gain Biological adhesion 29 9 4.6E–4 3.5E–3 better understanding of ERG and WNT11 relationship, Biological regulation 163 48 5.4E–3 3.0E–2 we additionally measured mRNA expression of ERG and WNT11 in diagnostic bone marrow material that Abbreviation: ERG, E26 transforming sequence-related gene. included a larger cohort of AML (n ¼ 30), T-ALL Candidate ERG targets, 342 annotated genes, were functionally categorized by DAVID Bioinformatics Database 6.7. The gene count (n ¼ 30) and normal bone marrow (n ¼ 30). We observed is given for the number of genes belonging to a biological term out of that ERG mRNA expression was co-expressed with the total number of annotated genes (also shown as a percentage). WNT11 mRNA in a considerable fraction (80%) of Fisher’s exact P-value is listed for significance of biological enrichment unselected AML and normal bone marrow samples and of the total number of candidate genes. The biological classifica- tions suggest a broader spectrum of biological functions for ERG. to a lower extent of co-expression, 40% in T-ALL bone The significant biological processes are ranked by Benjamini– marrow samples (Supplementary Figure 2a). Despite Hochberg (B–H) corrected P-values. a nonlinear correlation, WNT11 promoter enrichment

Table 3 Top canonical pathways of ERG putative genes Significant canonical pathways Gene Total number P-value B–H count of genes P-value

Basal cell carcinoma 7 68 0.0001 0.02 BMP3, FZD5, FZD7, FZD8, WNT2, WNT11, WNT9A Human embryonic stem cell pluripotency 9 148 0.0002 0.02 BMP3, FGF2, FZD5, FZD7, FZD8, NTRK2, WNT2, WNT11, WNT9A Wnt/b-catenin signaling 10 168 0.005 0.02 CCND1, CDH1, FZD5, FZD7, FZD8, SOX3, SOX11, WNT2, WNT11, WNT9A Role of NANOG in mammalian embryonic stem cell pluripotency 8 114 0.005 0.02 BMP3, FZD5, FZD7, FZD8, GATA4, WNT2, WNT11, WNT9A

Abbreviation: ERG, E26 transforming sequence-related gene. A total of 342 annotated genes obtained from ChIP-chip were subjected to ingenuity pathway analysis. Top canonical pathways are listed with gene symbols, gene count, total genes from a specific pathway, enrichment P-values, and Benjamini–Hochberg corrected P-values (B–H P-value). Gene symbols indicated by bold text are listed in multiple pathways.

Oncogene ERG genome-wide screen in acute leukemia LH Mochmann et al 2047 Table 4 Description of selected 24 ChIP-chip enriched promoter regions Accession Gene Gene name I II Number of Number of ETS number symbol ETS motifs motifs o1kb 1–2 kb

BC074791 WNT11 Wingless-type MMTV integration site family, member 11 9.66E þ 17 7.31E þ 12 1 0 NM_014276 RBPJL Recombining binding suppressor of hairless 8.12E þ 17 8.55E þ 12 0 NAa BC029364 ICAM4 Intercellular adhesion molecule 4 6.69E þ 17 8.59E þ 12 0 1 NM_004764 PIWIL1 Piwi-like 1 6.69E þ 17 6.69E þ 12 1 0 AK075419 UBAC2 Phosphoglycerate dehydrogenase like 1 6.14E þ 17 4.38E þ 12 1 NA BC051839 TMPRSS2 Transmembrane protease, serine 2 6.14E þ 17 4.98E þ 12 3 0 NM_000407 GP1BB IB (), b polypeptide 6.69E þ 17 6.69E þ 12 2 NA NM_053056 CCND1 Cyclin D1 4.54E þ 17 6.18E þ 12 0 1 NM_005806 OLIG2 lineage transcription factor 2 4.38E þ 17 4.66E þ 12 2 0 NM_153425 TRADD TNFRSF1A-associated via death domain 5.20E þ 16 7.39E þ 12 3 1 BC013297 CD7 CD7 antigen (P41) 2.00E þ 15 10.0E þ 13 3 0 NM_003507 FZD7 homolog 7 15.0E þ 13 4.54E þ 12 0 1 BC078170 WNT2 Wingless-type MMTV integration site family member 2 13.0E þ 13 7.89E þ 12 0 1b NM_170713 RASSF1 RAS association (RALGDS/AF-6) domain family 1 10.2E þ 12 7.89E þ 12 0 2 NM_000799 EPO Erythropoietin 7.98E þ 12 4.16E þ 12 0 1 BC019270 CDC42EP3 CDC42 effector protein (rho GTPase binding) 3 5.80E þ 12 4.54E þ 12 4 0 U66306 RXRA Retinoid X , a 5.20E þ 12 4.11E þ 12 2 1 NM_000474 TWIST1 Twist homolog 1 (Drosophila) 4.97E þ 12 7.26E þ 12 1 NA BC010507 CD14 CD14 antigen 4.97E þ 12 4.97E þ 12 2 2 NM_003395 WNT9A Wingless-type MMTV integration site family, member 9A 6.94E þ 11 4.66E þ 12 2 2 BC112199 ERBB4 V-ERB-A erythroblastic leukemia viral oncogene 6.68E þ 11 6.63E þ 11 2 0 NM_006037 HDAC4 Histone deacetylase 4 6.68E þ 11 4.66E þ 12 0 2 NM_003790 TNFRSF25 Tumor necrosis factor receptor superfamily, 25 5.48E þ 11 4.38E þ 11 1 0 NM_001261 CDK9 Cyclin-dependent kinase 9 4.38E þ 11 4.38E þ 11 1 1

Abbreviations: ChIP-chip, chromatin immunoprecipitation-on-chip; ERG, E26 transforming sequence-related gene; ETS, E26 transforming sequence. A selection of 24 (of 342 annotated genes) candidate ERG targets from duplicate chips (ChIP-chips I and II) are described and ranked according to chip hybridization Benjamini–Hochberg corrected P-value displayed as -log(P-value). MATCH algorithm was used to determine the frequency of the TRANSFAC conserved ETS-binding motif located in each of the 24 putative promoter regions ranging within 1 kb and between 1 and 2 kb from the transcription start site. Stringent core match and matrix match scores ranged from 0.94 to 1.00 and 0.94 to 0.97, respectively. A cut-off to minimize the sum of both error rates was used. Proximal promoter DNA sequences were obtained from Cold Spring Harbor Laboratory Transcriptional Regulatory Element Database and Swiss Institute of Bioinformatics Eukaryotic Promoter Database. aPredicted promoter sequences 41 kb were not available (NA) at the indicated databases. A single ETS-binding motif was found in the RBPJL proximal promoter region (o1 kb), although not statistically significant according to MATCH algorithm. bETS conserved motif located at 2.2 kb. through ChIP in primary normal and leukemia cells WNT11 transcripts (by 40–50% reduction) (Figure 3c). strengthened that ERG occupied the WNT11 promoter. The lack of mRNA expression of several genes in KG1 and in other leukemia cell lines may be due to aberrant gene promoter methylation, as described for WNT2 Knockdown of ERG expression results in transcriptional (Lelinek et al., 2008) and WNT9 (Shu et al., 2006), or downregulation of ERG target genes requirement of additional factors in leukemia cells. To gain an understanding of relevant targets of ERG in leukemia, siRNA molecules were used to knock down endogenous ERG to determine if mRNA expression Induced ERG expression results in transcriptional of candidate targets was affected. Examination of ERG upregulation of ERG target genes mRNA expression in Jurkat was quantitatively low A controlled ectopic expression of ERG in K562 cells (a compared with other hematopoietic cell lines. Therefore, cell line normally lacking ERG) was used to examine the KG1 cells, having higher ERG mRNA expression, were cellular consequences of ERG overexpression. A tet-on used to knock down ERG mRNA expression. ERG inducible system was selected to induce mRNA expression was efficiently reduced by 70% in ERG expression by the addition of doxycycline (dox). KG1 cells. This was further evidenced by western blot, Stably transfected pTRE-ERG isoform 3 (ERG3) clones whereby a reduction of 60% (for the 70% ERG knock- in K562 cells were examined for elevated ERG mRNA down at the transcript level) of ERG protein was induction (Figure 4a, black bars) and elevated ERG observed (Figures 3a and 3b). Furthermore, we deter- protein levels (Figure 4b). A potent fold induction of mined that 9 of the 17 ERG putative targets were ERG mRNA expression was achieved in individual expressed in KG1 cells. Of the nine target genes clones ranging from 10- to 30-fold. These stable clones consistently downregulated by siRNA-mediated ERG are referred to hereafter as ERG-inducible cells. In the knockdown were RASSF1, TWIST1 and RXRA, and absence of dox, ERG expression was low or not those significantly downregulated were TNFRSF25 and measurable.

Oncogene ERG genome-wide screen in acute leukemia LH Mochmann et al 2048 40 C20 C17/C20 35

30

25

20

15 Fold enrichment 10

5

0

EPO CD7 CDK9 CD14 FZD7 RBPJLWNT2 CDC42OLIG2 WNT11 PIWILI ICAM4 HDAC4UBAC2 ERBB4 TRADD GP1BBWNT9ACCND1 TWIST1 RASSF1 TMPRSS2 TNFRSF25 Gene symbol

2 ChIP-chip 1 ChIP-chip 2

1.5

1

0.5

0

Log2 (Cy5/Cy3) 75594000 75596000 75598000 75600000 75602000 75604000

-0.5 WNT11 -1

-1.5 11 Figure 1 Validation of ERG targets in leukemia. (a) A selection of 24 candidate genes, obtained in duplicate ChIP-chip assays (ChIP-chips I and II), was validated by determining enriched promoter regions via chromatin PCR amplification (ERG-specific immunoprecipitated DNA versus total chromatin) with SYBR Green quantitative PCR. As a control, IgG-immunoprecipitated DNA versus total chromatin was amplified in parallel to ERG-enriched DNA from ChIP-chips I and II and subtracted from fold enrichment values as background. Significant candidate genes with at least twofold enrichment (dotted line) in duplicate ChIPs are shown relative to total chromatin. Each SYBR Green assay was conducted in triplicate wells and values are displayed as averages of three SYBR Green assays. (b) Using SignalMap software from NimbleGen human promoter chip array, fluorescent log 2 ratios display multiple peak detection at the chromosomal location of the WNT11 promoter in ChIP-chip 1 and the transcription start site is indicated (TSS) by an arrow.

To determine the effects of downstream targets with by twofold (Figure 4c). Thus, prolonged growth of the ERG-inducible assay, we measured mRNA expres- ERG-induced cells not only enhanced upregulation of sion of all 17 putative ERG targets. WNT11 was the WNT11 but also influenced TNFRSF25 and GATA-4 only co-induced gene with ERG mRNA after a 72-h transcripts. Other putative targets unaffected by these induction. In ERG-induced cells, WNT11 was co- measures may require additional factors, or simply ERG expressed in multiple individual clones in the 2–5-fold alone cannot directly affect gene expression because range (Figure 4a, gray bars). Furthermore, WNT11 of tight regulation or dysregulation, as is often the case mRNA expression was undetectable in untransfected in . K562 cells; thus, fold changes in WNT11 mRNA To determine if ERG bound the WNT11 promoter expression were entirely ERG-dependent. in vivo, 1.2 kb of the WNT11 proximal promoter was A re-evaluation of 17 putative targets following cloned into a luciferase reporter vector, pGL3-basic. prolonged growth (continuous dox stimulation for 8 The 1.2-kb WNT11 promoter insert (denoted as pGL3– days) of ERG-inducible cells was conducted. ERG WNT11) was designed to include the oligonucleotide mRNA expression doubled and WNT11 mRNA ex- sequence spotted on the chip for ChIP-chip hybridiza- pression was upregulated by 15–19-fold (Figure 4c) tion, the oligonucleotide sequence in the amplified when compared with 72 h dox stimulation (Figure 4a). product for quantitative PCR, and the significantly In addition, candidate target genes GATA-4 was determined ETS-binding motifs. Transfection of pGL3– upregulated 4–6-fold and TNFRSF25 was upregulated WNT11 into ERG-inducible cells reliably activated

Oncogene ERG genome-wide screen in acute leukemia LH Mochmann et al 2049 25 1.4 luciferase activity in response to ERG induction when

1.2 compared with no dox conditions (Supplementary 20 Figure 2b). We conclude that these results indeed 1.0 confirm WNT11 is a direct target of ERG. 15 0.8 mRNA expression 0.6 GSK-3b inhibition promotes growth advantage 10

ERG in ERG-induced cells promoter ChIP enrichment 0.4 Owing to increased WNT11 transcripts prompted by 5 0.2 ERG induction and promoter enrichment of several Relative WNT11 WNT genes, we hypothesized that ERG overexpression 0 0.0 AML A T-ALL ND AML B AML E AML D AML C influenced the WNT pathway. Glycogen synthase WNT11 promoter ChIP enrichment Relative ERG mRNA expression kinase, GSK-3b, a multifaceted kinase central to WNT signaling, functions in a diverse manner through regu- Figure 2 WNT11 promoter enrichment by ChIP in primary bone marrow cells. ChIP assays with ERG-specific antibody C20 and lation of cell proliferation, differentiation and survival control IgG in duplicate ChIPs were conducted in primary normal (Wu and Pan, 2009). GSK-3b inhibitors have become and leukemia bone marrow cells. The bone marrow samples included an attractive therapeutic option, as they has been shown five AML (AML A–E), one T-ALL and one normal donor. Gray to suppress tumorigenesis (Luo, 2009). A small-molecule bars display an average of three SYBR Green assays per sample GSK-3b inhibitor, BIO, can activate canonical WNT showing fold enrichment of the amplified WNT11 promoter region (the sequence amplified is located from -401 to -501 bp relative to signaling (Soda et al., 2008) and suppress leukemia cell TSS) relative to total chromatin. Enrichment of the amplified WNT11 proliferation (Holmes et al., 2008). The ERG-inducible promoter region obtained from ChIP with IgG was subtracted as assay was utilized to study the proliferative growth background from the fold enrichment value. Quantitative PCR was effects in the presence and absence of GSK-3b inhibitor, conducted in triplicate wells. For comparison with ERG abundance, the right axis is shown as a plot of relative ERG mRNA expression BIO. Upon ERG induction, there were no proliferative measured by multiplex quantitative RT–PCR of each of the seven advantages between the presence and absence of dox individual bone marrow samples. in ERG-inducible cells. The addition of BIO to ERG-

120

100 siRNA siRNA Control 80 ERG control

60 ERG

40 β * -actin 20 % Relative mRNA knockdown

0 Control siRNA ERG siRNA

100

80

* 60 *

40

20 % Relative mRNA knockdown

0 GAPDH RASSF1 RXRA TWIST1 WNT11 TNFRSF25 Gene symbol Figure 3 siRNA-mediated knockdown of ERG downregulates gene expression of WNT11 and other ERG putative targets. (a) Multiplex quantitative RT–PCR was conducted to measure relative knockdown of ERG mRNA expression. ERG mRNA knockdown of 70±0.03% is shown. (b) This figure depicts a representative western blot displaying reduced ERG protein levels from siRNA-mediated ERG knockdown in KG1 cells (60% at the transcript level). ERG protein levels of cells transfected with control siRNA luciferase and untransfected KG1 cells remain unchanged. (c) To determine the effects of ERG knockdown on putative target genes, cDNA derived from transfected KG1 cells with ERG-specific siRNA and control siRNA (70% knockdown) were used as templates for the measurement of relative mRNA changes. ERG siRNA-mediated knockdown statistically affected WNT11 and TNFRSF25. Fold changes relative to the control siRNA are displayed as averages of two SYBR Green assays and each assay was conducted in duplicate wells. The asterisk indicates significance by Student’s paired t-test (P-valueo0.05).

Oncogene ERG genome-wide screen in acute leukemia LH Mochmann et al 2050 35 2 ERG WNT11 30 25 1.5 ERG 20 β 15 -actin 1 10 +-dox 5 0.5 Absorbance (450nm) Relative mRNA expression 0 * CL64 CL9 CL4 CL40 CL46 CL41 Individual clones 0 24 48 72 90 Hours ERG WNT11 GATA-4 TNFRSF25 80 BIO dox BIO no dox Control dox Control no dox 70 60 100 dox no dox 50 40 80

expression 30 60 20 cells/ml) 4

Fold induction of mRNA 10 0 40 CL9 CL4 CL46 CL64 Individual clones 20 Figure 4 ERG induction upregulates gene expression of WNT11 Cell count (10 * and other ERG putative targets. (a) K562 cells transfected with pTRE-ERG3 were subsequently used for SYBR Green RT–PCR 0 analysis for determining fold changes in ERG and WNT11 mRNA Control BIO levels in cells treated with ( þ ) and without (À) dox over a 72-h period. Strong co-induction levels of ERG and WNT11 mRNA Figure 5 GSK-3b inhibitor, BIO, promotes a proliferative growth expression are displayed by black and gray bars, respectively. The advantage in ERG-induced cells. (a) Proliferation was measured at induction levels were normalized to uninduced (without dox) cells. 24, 48 and 72 h by WST-1 assay. BIO (1.5 mM)-treated K562 pTRE- A representative SYBR Green RT–PCR assay of pTRE-ERG3 ERG3 cells elicit a growth advantage in the ERG-induced (solid individual stable clones with strong ERG and WNT11 mRNA line, black circles) cells over the BIO-treated uninduced cells (solid induction is shown. (b) K562 cells transfected with pTRE-ERG3 line, white circles) at 72 h. BIO untreated K562 pTRE-ERG3 cells show elevated ERG protein levels upon dox induction. with ERG induction (dashed line, black squares) and without ERG A representative western blot of using ERG antibody C20 in the induction (dashed line, white squares) do not significantly differ in tet-on ERG-inducible assay is shown. ERG protein levels are proliferation. The proliferative growth advantage elicited by the induced with the addition of dox ( þ ) and uninduced without dox addition of BIO to ERG-induced cells compared with BIO-treated (À) in cells normally lacking ERG. (c) Long-term culture (8 days of uninduced cells was observed in three independent stable clones dox stimulation) of ERG-induced cells resulted in an increase in over a 3-day period. The absorbance measurement values are WNT11, GATA-4 and TNFRSF25 mRNA expression in four averages (with standard error bars) of five wells per experimental individual clones. Relative mRNA expression levels were deter- treatment. Paired Wilcoxon’s rank sum test was used to determine mined by duplicate SYBR Green assays in duplicate wells. the statistical differences. (b) Determination of viable cell numbers by trypan blue exclusion with BIO treatment at 72 h of K562 pTRE-ERG3 cells under dox-induced (black bars) and uninduced (gray bars) conditions. The viable cell numbers are displayed as averages with standard error bars. Student’s paired t-test was used induced cells, however, resulted in a proliferative growth to determine the statistical difference. Asterisk (*) indicates advantage, whereas treatment of ERG-uninduced statistically significant differences (P-valueo0.05). cells with BIO effectively suppressed proliferation (Figure 5a). The proliferative advantage was further evi- denced by the greater number of viable cells in determine the functional consequences of ERG and BIO-treated cells with ERG induction versus prolifera- WNT11 overexpression. ERG-inducible clones were tive suppression of ERG-uninduced cells (Figure 5b). cultured for a prolonged time in the presence and Thus, overexpression of ERG provokes a proliferative absence of dox stimulation. As early as 4 days of culture, resistance to WNT agonist, BIO. a considerable fraction of ERG-induced cells underwent morphological shape changes from small round K562 cells to cells that were polarized and elongated with ERG overexpression induces morphological bidirectional protrusions (Figure 6a). These shape transformation changes only occurred in the presence of dox, whereas WNT11 is well known for promoting morphological in the absence of dox, the ERG-uninduced cells retained changes during early cell differentiation (Eisenberg the native round and unpolarized morphology. The et al., 1997) and morphological transformation of elongated cells with bidirectional protrusions and terminally differentiated cells (Uysal-onganer et al., attached morphology were further verified by strong 2010). This knowledge provided a focus point as to fluorescence of the ERG-induced cells (cloned into a

Oncogene ERG genome-wide screen in acute leukemia LH Mochmann et al 2051 + dox - dox

10X 20X 10X

60 60

50 50

40 40 30 30 20 20 10 10 with protrusions / field with protrusions / field Number of polarized cells Number of polarized cells 0 + dox - dox + dox - dox + dox - dox + dox - dox 0 + dox - dox + dox - dox + dox - dox control BIO control BIO DAY 4 DAY 6 DAY 8 Day 4 Day 6

100 30

80 25

mRNA 20 60 * 15

WNT11 40

knockdown (%) 10

siRNA 20

with protrusions / field 5 Number of polarized cells 0 siRNAcontrol siRNA WNT11 0 WNT11 siRNA control siRNA Figure 6 Co-expression of ERG and WNT11 induces morphological transformation. (a) Potent ERG-inducible clones were cultured with continuous dox stimulation for 8 days. Morphological shape changes from round unpolarized cells to polarized cells with protrusions began to appear as early as day 4 in the presence of dox ( þ dox), whereas in the absence of dox (Àdox) cellular morphology remained unchanged. Visualization of the cellular shape changes at  10 and  20 magnification of both dox-treated ( þ dox) and dox-untreated (Àdox) cells are shown. Due to the red fluorescence traceable tet-on system, ERG-induced shape changes could be visualized by strong fluorescence of the elongated cells as opposed to the absence of fluorescence of ERG-uninduced cells. (b) ERG-inducible cells with and without dox were seeded at an equal cell number on a 2-mm-grid six-well dish for enumeration of cells polarized with protrusions (at  10 magnification). Prolonged growth of ERG-induced cells peaked at day 8 having 50–60 polarized cells with protrusions per field, whereas uninduced cells did not undergo shape changes. Bar graph is representative of experiments conducted with three K562 pTRE-ERG3 clones and depicts the density of polarized cells with protrusions in 10 fields on the x-axis, and the height depicts number of cells polarized with protrusions per field. (c) ERG-induced and -uninduced cells were treated with and without BIO (1.5 mM) and analyzed in a similar manner as described above. BIO-treated cells strongly inhibited the elongation process of ERG-induced cells when compared with BIO-untreated cells. (d) To determine if cell morphological transformation was due to ERG induction of WNT11, WNT11 siRNA was utilized to knock down WNT11 transcript levels. Duplicate transfections of WNT11 and control siRNA were conducted with two individual K562 pTRE-ERG3 stable clones. Pre-stimulated cells (3 days dox stimulation) were transfected with WNT11 and control siRNA. Knockdown of WNT11 transcripts was reduced by 70%. Asterisk (*) indicates statistically significant differences (P-valueo0.05). (e) Duplicate transfections of WNT11 and control siRNA were seeded equally by cell number onto a 2-mm-grid six-well dish. The number of polarized cells with protrusions was counted as described above. siRNA- mediated WNT11 knockdown significantly reduced the number of cells polarized with protrusions per field (mean of 4.5 cells/field) when compared with control siRNA (mean of 6.3 cells/field). Bar graph is representative of experiments conducted with two K562 pTRE-ERG3 clones and depicts the density of polarized cells with protrusions in 40 fields (80 fields in total were analyzed) on the x-axis, and the height depicts number of cells polarized with protrusions per field. Statistical significance was determined by the non- parametric Wilcoxon’s rank sum test.

Oncogene ERG genome-wide screen in acute leukemia LH Mochmann et al 2052 tet-on red fluorescence tracking vector) versus absence of WNT11 is often coupled with the upregulation of of fluorescence of the ERG-uninduced cells. ERG-inducible GATA-4 in cardiomyocytes (Flaherty and Dawn, 2008) cells were then seeded in equal cell numbers onto a six- and the co-expression is mirrored in the ERG-induced well dish with grids for enumeration of cells polarized cells; thus, further studies for validating this association with protrusions. Cells polarized with protrusions are necessary. peaked at 50–60 cells per field after 8 days of dox In previous studies, avian embryonic stem cell stimulation (Figure 6b). This observation was also differentiation into multiple cell phenotypes was evident in seven of the eight stably pTR-ERG3 shown to be dependent on WNT11 expression (Brandon transfected clones. et al., 2000). Thus, similar to ERG function, WNT11 We then tested the effect of BIO on morphological has an early developmental role in normal hematopoi- changes. The addition of BIO led to a strong inhibition esis. Moreover, emerging evidence reveals that genes of the elongation process in the ERG-induced cells. belonging to the WNT pathway, both canonical and ERG-uninduced cells treated with BIO retained the round non-canonical WNT genes, are aberrantly expressed morphology. This result implied that the presence of BIO in leukemia due to mutations (Reya and Clevers, 2005) inhibited elongation (Figure 6c), while eliciting a growth or aberrant methylation (Gomez-Roman et al., 2007; advantage of ERG-induced cells, as shown in Figure 5. Martin et al., 2009) that result in WNT cascade errors. Finally, to test whether WNT11 directed morpholo- Interestingly, synonymous to ERG, WNT11 mRNA gical transformation of ERG-induced cells, WNT11 expression and protein were elevated in high-grade siRNA was introduced into ERG-induced cells pre- prostate tumors (Zhu et al., 2004; Uysal-onganer et al., stimulated for 72 h. Overall, the transfection procedure 2010) as well as implicated in breast cancer progression reduced dramatically the number of polarized cells with (Lin et al., 2007). A random sampling of a larger cohort protrusions compared with cells that did not undergo of AML, T-ALL and normal bone marrow diagnostic electroporation. Despite this effect, siRNA mediated material in this study demonstrated a nonlinear quanti- knockdown of WNT11, mRNA knockdown of 70% tative relationship between ERG mRNA expression and (Figure 6d) reduced the number of cells polarized with WNT11 mRNA expression, although co-expression of protrusions per field when compared with the control both genes was evident in AML and in normal samples. siRNA at day 6 of dox stimulation (Figure 6e). The Moreover, the nonlinear correlation between ERG mean of cells polarized with protrusions per field was 4.5 ChIP enrichment and WNT11 gene expression also and 6.3 cells per field for WNT11 and control siRNA, appeared to follow different dynamics. As one study respectively (two-sided P-value ¼ 0.007) in a total of 80 proposed, gene expression has a linear relationship with fields counted. These results demonstrate that WNT11 is ChIP enrichment for actively transcribed genes, but the necessary to undergo elongation and attachment in the linear model does not fit a considerable fraction of non- morphological transformation of ERG-induced cells. actively transcribed genes. This may be a limitation in the ChIP method, or more likely due to the complex intermediate steps in accessing the chromosome Discussion (Guenther et al., 2007). Recent in vitro and in vivo ChIP-seq binding studies of ERG in prostate cells Genetic aberrations of transcription factors required for have suggested that chromatin accessibility is a major normal hematopoiesis are frequently found in leukemia determining factor for promoter occupancy, although and contribute to transformation of hematopoietic the correlation to gene expression was not addressed progenitors through deregulated signaling (Look and (Wei et al, 2010). Thus, we stipulate from these results O’Neil, 2007; Chiang et al., 2009). The ETS transcrip- that ERG overexpression, coupled with co-expression of tion factor ERG is overexpressed in subtypes of newly WNT11, may potentially have a role in T-ALL and diagnosed T-ALL and AML patients. Overexpression of AML through deregulated non-canonical WNT ERG mRNA expression was further correlated with signals. poor prognosis; however, the downstream effects of Next, we modulated a central junction of the WNT deregulated mRNA expression are not yet understood pathway through GSK-3b inhibition. GSK-3b inhibi- (Baldus et al., 2007). Interestingly, patients with com- tion through BIO is of particular clinical interest, as the plex cytogenetics attribute aberrant ERG expression to effects can preserve and support proliferation of normal recurring chromosomal abnormalities such as amplifica- hematopoietic stem cells through canonical WNT tion at regions of the ERG loci (Baldus et al., 2004). signaling (Sato et al., 2004). In contrast, GSK-3b Thus, ERG deregulated expression might have a pivotal inhibition by BIO suppressed proliferation of various role, as an upstream regulator, in acute leukemia. leukemia cell lines, including K562 (Holmes et al., 2008). Herein, we explored the potential signaling pathways Interestingly, WNT11 has an opposing effect, for that ERG could regulate by ChIP-chip screening. This instance, in murine embryonic stem cells a repression whole-genome screen identified WNT11, a non-canoni- of canonical WNT signals is reported with exogenous cal WNT gene, as a direct downstream target of ERG. addition of WNT11 (Singla et al., 2006; Anton et al., Furthermore, enrichment of the WNT11 promoter was 2007). In our results, the addition of BIO elicited a confirmed from ChIP of primary normal and leukemia growth advantage in ERG-induced cells. This result may bone marrow cells. Other putative targets that remain to be due to ERG/WNT11 function overriding BIO effects be explored are TNFRSF25 and GATA-4. Upregulation (presumably canonical WNT signals) through stimula-

Oncogene ERG genome-wide screen in acute leukemia LH Mochmann et al 2053 tion of WNT non-canonical signals. Those signals, we Patient material propose, possibly occur through the Ca2 þ -dependent To conduct ChIP assays, we obtained untreated bone marrow MAPK or PI3K-AKT pathways, as both pathways were aspirates from adult patients with newly diagnosed AML shown to downregulate GSK-3b activity (Bikkavilli (n ¼ 5), T-ALL (n ¼ 1) and one healthy donor. Randomly et al., 2008; Hu et al., 2009). selected diagnostic AML (n ¼ 30), T-ALL (n ¼ 30) and normal bone marrow (n ¼ 30) samples were obtained for the measure- Finally, the potent morphological transformation of ment of ERG and WNT11 mRNA expression. AML patients ERG-induced cells was unanticipated, yet plausible, were treated at Charite´, Campus Benjamin Franklin, Depart- according to a recent report describing significant ment of Hematology and Oncology, Berlin, Germany. T-ALL association of ERG-positive tumors with WNT pathway specimens were obtained from patients enrolled in the German signaling genes and epithelial-to-mesenchymal transition Acute Lymphoblastic Leukemia Multicenter Study Group genes (Gupta et al., 2010). In our study, ERG-induced 07/03. Written informed consent was obtained from all morphological shape changes also complement the patients and healthy volunteers, and the local ethics committee initial global ChIP-chip analysis that determined cell approved the study. adhesion as a significantly enriched biological theme (Table 2). Moreover, there is surmounting evidence of ChIP-chip WNT11 as a morphogenic and cytoskeletal effector ChIP-chip was carried out as described in the NimbleGen molecule (Zhou et al., 2007; Flaherty and Dawn, 2008; Systems, Inc. (Madison, WI, USA) and Farnham (2009) Lai et al., 2009; Uysal-onganer et al., 2010). Morpho- methods. Briefly, Jurkat cells (1 Â 109 cells) were crosslinked logical transformation involving ERG gene activation of with formaldehyde. Following this, cells were suspended in WNT11 may have a role in the bone marrow micro- lysis solution and DNA was fragmented to sizes ranging from 200 to 1000 bp. The ERG-bound DNA was enriched with two environment. In the case of leukemia, high ERG- ERG-specific antibodies targeting two different epitopes of the expressing cells producing morphogen WNT11 may protein. The antibodies used for immunoprecipitation were create a niche in the bone marrow for resistant cells. rabbit Erg-1/2/3 C20 (epitope at C-terminus) and an equimolar Further investigations are warranted to determine the mix of rabbit C20 and rabbit C17 (epitope corresponding consequences of high ERG and WNT11 expression in to internal ERG protein) and non-specific IgG antibody as the bone marrow microenvironment. control (Santa Cruz Biotechnologies, Santa Cruz, CA, USA). In conclusion, this genome-wide screen revealed Reverse crosslinking, DNA purification and ligation-mediated WNT pathway as a major transcriptional network PCR (LM-PCR) amplification enabled to obtain the following of ERG. Specifically, WNT11 emerged as the most experimental groups: enriched DNA by antibodies C20, C17/ prominent candidate of ERG. GSK-3b inhibition C20, IgG and total chromatin. Four promoter array hybridi- zations were conducted as follows: C20 enriched DNA (Cy5- provoked a proliferative growth advantage in cells with label) was paired with total chromatin (Cy3-label) and referred ERG abundance and co-expression of WNT11, implying to as ChIP-chip 1. The second hybridization (ChIP-chip 3) ERG’s involvement in WNT signaling. This result consisted of C20 enriched DNA (Cy5-label) and was paired suggests that for high-risk acute leukemia patients, with IgG-enriched DNA (Cy3-label). The second pair of characterized by high ERG mRNA expression, GSK-3b hybridizations (ChIP-chips 2 and 4) include C17/C20 enriched inhibitors as potential therapeutic options may be DNA (Cy5-label) paired with total chromatin (Cy3-label) and insufficient due to intrinsic ERG-induced resistance. C17/C20 enriched DNA (Cy5-label) paired with IgG-enriched Finally, morphogenic transformation of ERG-induced DNA (Cy3-label). The hybridization was performed with a cells mediated by co-expression with WNT11 implicates high-resolution promoter tiled array consisting of 50–75mer a potential role for ERG- and WNT11-activating probes (770 000) that represents 29 000 annotated human transcripts (NimbleGen Systems, Inc.). Algorithms for Calcu- genetic programs that promote polarization and protru- lating Microarray Enrichment, ACME (Fred Hutchinson sions. This type of cellular response is a biological signal Cancer Research Center, Seattle, WA, USA), was used to for cell migration and cell invasion by which these determine potential genomic enriched regions, by sliding pathogenic mechanisms have already been proposed window analysis (using a window size of 2000 bp) with Lowess for ERG in prostate cancer (Klezovitch et al., 2008). normalized log 2 (Cy5/Cy3) ratios. The calculated P-values Herein, these data collectively provide novel molecular were Benjamini–Hochberg corrected and a threshold was set at components of ERG-directed cellular signals as a model 90% (ACME bioinformatics software). For statistical analysis, for acute leukemia patients with high ERG expression significant gene annotations derived from hybridizations, and poor prognosis. ChIP-chips 1 and 3 were combined and are referred to as ChIP-chip I. Significant gene annotations from hybridizations and ChIP-chips 2 and 4 were combined and are referred to as ChIP-chip II. Only overlapping gene sets from ChIP-chips I and II were considered as candidate genes (Table 1). A detailed Materials and methods ChIP-chip description and complete NimbleGen promoter array data are available from the GEO database accession Cell culture and chemicals number GSE21495. Jurkat, KG1 and K562 cell lines were obtained from the German Resource Center for Biological Material, DSMZ (Braunschweig, Germany), and grown in RPMI media with Databases 10% fetal bovine serum. All cell lines were cultured at 37 1Cin Databases used in our analyses were as follows: for gene a5%CO2 humidified chamber. BIO, a GSK-3b inhibitor, was promoter analyses, the Cold Spring Harbor Laboratory purchased from Stem Cell Technologies (Vancouver, BC, Transcriptional Regulatory Element Database, Swiss Institute Canada) and dissolved in DMSO. of Bioinformatics Eukaryotic Promoter Database and Gene

Oncogene ERG genome-wide screen in acute leukemia LH Mochmann et al 2054 Regulation MATCH program were used to predict proximal Cologne, Germany) as per the manufacturer’s instructions, promoter regions and predict conserved ETS-binding motif. with ERG-specific siRNA (sih5ERG), or control siLuciferase DAVID Bioinformatics Resources 6.7 (Dennis et al., 2003; (both siRNAs were kind gifts from Silence Therapeutics, Huang et al., 2009) and Ingenuity Pathway Analysis programs Berlin-Buch, Germany), or On-Target plus siRNA WNT11 were used to categorize biological pathways/themes of (Dharmacon, Denver, CO, USA). RNA knockdown levels candidate gene lists. were monitored by quantitative PCR. ERG protein levels were additionally monitored by standard western blotting techni- ques using antibody C20 for ERG detection. ChIP of primary leukemia blasts Primary bone marrow samples with sufficient material (cell numbers) to conduct ChIP assays were selected. Freshly Inducible expression constructs prepared bone marrow aspirates from newly diagnosed AML ERG3 cDNA was synthesized by reverse transcription using patients, a T-ALL patient and a normal bone marrow donor Superscript III (Roche Diagnostics GmbH) with ERG iso- were enriched for the mononuclear fraction by Ficoll-Hypaque form-specific primers as previously characterized (Bohne et al., density gradient (Amersham Pharmacia Biotech, Uppsala, 2009). A commercial ‘tet-on’ two-plasmid system (Clontech, Sweden). The following was performed for each donor: cell Mountain View, CA, USA) was used to induce ERG count was determined and cells were crosslinked with for- expression in trans under the control of a tetracycline-inducible maldehyde according to NimbleGen protocols. Immunopreci- promoter. One vector, ptet-onAdvanced (pTet), expresses the pitation was performed using 3 mg of either C20 ERG antibody transactivator protein only in the presence of an analog of or control anti-IgG antibody to precipitate ERG-bound DNA tetracycline, doxycycline (referred to hereafter as dox, pur- complexes. A total of 2 mg of protein per ChIP was used to chased from Clontech). An insert of ERG3 isoform was perform two independent ChIPs for each antibody. Precipitated subcloned into a second vector, pTRE-Tight-BI-DsRed- DNA templates were purified and amplified as described above. Express (pTRE), which includes a RED fluorescent gene with a bidirectional promoter consisting of binding sites for transactivator protein. Vectors pTet (along with a linear Quantitative PCR of promoter regions puromycin marker) and pTRE-ERG3 were sequentially Quantitative PCR was performed to validate enrichment of transfected into K562 cells using electroporation (Lonza candidate target promoter regions obtained from the ChIP- Cologne AG, Cologne, Germany). K562 cells lack ERG chip genome-wide screen. Proximal promoter primers within mRNA expression and therefore ERG mRNA detection is 2 kb from the TSS were designed to include at least one of entirely dependent on dox induction. Transfected K562 cells the estimated conserved ETS-binding motifs, 5-GGAA/T-3 were maintained with puromycin at 1.25 mg/ml (Merck, and are described in Supplementary Table 1. SYBR Green Nottingham, UK) and G418 at 800 ng/ml (GIBCO, Invitrogen (Invitrogen GmbH, Karlsruhe, Germany) based quantitative GmbH) to maintain double stable clones. Transfected bulk PCR was utilized to measure the relative abundance of enriched cells were then seeded to expand individual stable clones and DNA to total chromatin. The cut-off for enriched target genes examined for ERG induction by the addition of dox (1 mg/ml). was set at twofold enrichment in both I and II ChIP groups ERG induction was determined by red fluorescence detection versus controls IgG-enriched DNA and total chromatin. and standard western blotting (using ERG-specific C20 antibody), and measurement of relative ERG mRNA expres- Quantitative PCR of gene expression sion by RT–PCR. For experimental controls, dox-untreated Total RNA was purified from RNeasy plus (Qiagen GmbH, cells were cultured in parallel to dox-treated cells as uninduced Hilden, Germany). First-strand complementary DNA (cDNA) status throughout each assay described in this study. Stable synthesis was performed with AMV reverse transcriptase kit transfected clones displaying strong ERG mRNA expression (Roche Diagnostics GmbH, Mannheim, Germany). Quantita- upon dox induction were selected for further analyses. Stably tive real-time–PCR (RT–PCR) was carried out in duplicate transfected K562 cell derivatives in this study are referred to using SYBR Green master mix (Invitrogen GmbH). Amplifi- pTRE-ERG3. Stable clones are enumerated as CL1, CL2, CL3 cations were performed using the following conditions: 95 1C, and so on. 10 min; 40 cycles of 95 1C, 30 s; 55 1C, 1 min; 72 1C, 30 s. The Luciferase reporter assays were conducted in the ERG- comparative CT equation (2ÀDDCT) was used to determine the inducible expression assay. Essentially, WNT11 proximal pro- relative expression levels of all genes used in this study under moter spanning þ 193 to -1042 bp relative to TSS (B1.2 kb) was control and experimental conditions (primer pairs described in cloned into pGL3-basic vector (Promega, Madisson, WI, USA). Supplementary Table 2). Glyceraldehyde 3-phosphate dehy- pRL-TK vector (kind gift from H Fechner lab, Department of drogenase (GAPDH) was utilized as an internal control Cardiology, Charite´, Berlin, Germany) was used for the normal- throughout this study. Multiplex RT–PCR to determine ization of transfection efficiency and as control measurement of ERG mRNA expression was previously described (Bohne renilla luciferase activity. Vectors pGL3–WNT11 and pRL-TK et al. 2009). WNT11 mRNA expression primer pairs and probe were co-transfected into cells stimulated with and without dox. As are described in Supplementary Information. Quantitative control, pGL3-basic and pRL-TK were transfected individually in measurements of both genes in patient material were done dox-stimulated cells. Firefly and renilla luciferase activities were using TaqMan Gene Expression Assays in duplicate wells measured with Dual-Luciferase Reporter Assay System (Promega). (Applied Biosystems Inc., Foster City, CA, USA). WNT11 transcripts in patient material were considered ‘present’ with Cell proliferation and enumeration CT values 27, were co-amplified with GUS as the house p Cell proliferation was measured with WST-1 reagent according keeping gene and values shown are relative to WNT11 mRNA to the manufacturer’s instructions (Roche Diagnostics expression in KG1 cells. GmbH). Briefly, cells for each time point were seeded in a 96-well plate with 3 Â 104/well. K562 pTRE-ERG3 stable siRNA-induced ERG knockdown clones were cultured with and without dox for a 24-h period. KG1 cells and K562 pTRE-ERG3-inducible clones were Following this, pTRE-ERG3-transfected cells were treated transfected according to AMAXA (Lonza Cologne AG, with BIO for 3 days. Absorbance (450 nm) of the reaction of

Oncogene ERG genome-wide screen in acute leukemia LH Mochmann et al 2055 WST-1 conversion to formazon dye was measured after 2 h Conflict of interest incubation with WST-1 at 24, 48 and 72 h. The trypan blue dye exclusion test was used to determine the number of viable cells The authors declare no conflict of interest. by cell counting with a hemocytometer. Counting of polarized cells with protrusions involved seeding ERG-induced and uninduced cells at equal cell numbers onto a 2-mm-gridsix-welldish(NalgeNunc International, Rochester, NY, USA). Following dox stimulation up to 8 days, the suspended cell Acknowledgements culture was gently removed for enumeration of cells polarized with protrusions at  10 magnification. Experiments conducted This study was supported by a grant from the Deutsche for siRNA-mediated knockdown of WNT11 were counted in Krebshilfe (Max Eder Nachwuchsfo¨rderung) to CDB. We duplicate. To count cells in an even distribution of the culture thank Dr Martin Neumann (Department of Hematology and dish surface area, culture dishes were divided into 3–4 sections, Oncology, Charite´, Berlin, Germany) for his critical reading of and 10 fields per section were counted by two persons. In the manuscript. We also thank Dr Anja Ku¨hl (Research addition, a blinded system was used for cell enumeration to Center ImmunoSciences, Charite´, Berlin, Germany) for helpful ensure reproducibility. advice and fluorescence microscopy assistance.

References

Anderson MK, Hernandez-Hoyos G, Diamond RA, Rothenberg EV. Ellett F, Kile BT, Lieschke GJ. (2009). The role of the ETS factor (1999). Precise developmental regulation of Ets family transcription in zebrafish vasculogenesis. Mech Dev 126: 220–229. factors during specification and commitment to the T cell lineage. Farnham PJ. (2009). Insights from genomic profiling of transcription Development 126: 3131–3148. factors. Nat Rev Genet 10: 605–616. Anton R, Kestler HA, Kestler HA, Ku¨hl M. (2007). Beta-catenin Flaherty MP, Dawn B. (2008). Noncanonical Wnt11 signaling and signaling contributes to stemness and regulates early differentiation cardiomyogenic differentiation. Trend Cardiovas Med 18: 260–268. in murine embryonic stem cells. FEBS Lett 581: 5247–5254. Gomez-Roman J, Cordeu L, Agirre X, Jime´nez-Velasco A, San Jose´- Baldus C, Burmeister T, Martus P, Schwartz S, Go¨kbuget N, Eneriz E, Garate L et al. (2007). Epigenetic regulation of Bloomfield CD et al. (2006). High expression of the ETS Wnt-signaling pathway in acute lymphoblastic leukemia. Blood transcription factor ERG predicts adverse outcome in acute 109: 3462–3469. T-lymphoblastic leukemia in adults. J Clin Oncol 24: 4714–4720. Guenther MG, Levine SS, Boyer LA, Jaenisch R, Young RA. (2007). Baldus CD, Liyanarachchi S, Mro´zek K, Auer H, Tanner SM, A chromatin landmark and transcription initiation at most Guimond M et al. (2004). Acute myeloid leukemia with complex promoters in human cells. Cell 130: 77–88. karyotypes and abnormal : amplification discloses Gupta S, Iljin K, Sara H, Mpindi JP, Mirtti T, Vainio P et al. (2010). overexpression of APP, ETS2, and ERG genes. Proc Natl Acad Sci FZD4 as a mediator of ERG oncogene—induced WNT signaling USA 101: 3915–3920. and epithelial-to-mesenchymal transition in human prostate cancer Baldus CD, Mro´zek K, Marcucci G, Bloomfield CD. (2007). Clinical cells. Cancer Res 70: 6735–6745. outcome of de novo acute myeloid leukaemia patients with normal Holmes T, O0Brien TA, Knight R, Lindeman R, Shen S, Song E et al. cytogenetics is affected by molecular genetic alterations: a concise (2008). Glycogen synthase kinase-3beta inhibition preserves hema- review. Br J Haematol 137: 387–400. topoietic stem cell activity and inhibits leukemic cell growth. Bastian LS, Yagi M, Chan C, Roth GJ. (1996). Analysis of the Stem Cells 26: 1288–1297. megakaryocyte glycoprotein IX promoter identifies positive and Hu D, Bi X, Fang W, Han A, Yang W. (2009). GSK3beta is involved negative regulatory domains and functional GATA and Ets sites. in JNK2-mediated beta-catenin inhibition. PloS One 4: e6640. J Biol Chem 271: 18554–18560. Huang DW, Sherman BT, Lempicki RA. (2009). Systematic and Bikkavilli RK, Feigin ME, Malbon CC. (2008). p38 mitogen-activated integrative analysis of large gene lists using DAVID bioinformatics protein kinase regulates canonical Wnt-beta-catenin signaling by resources. Nat Prot 4: 44–57. inactivation of GSK3beta. J Cell Sci 121: 3598–3607. Iwamoto M, Tamamura Y, Koyama E, Komori T, Takeshita N, Birdsey GM, Dryden NH, Amsellem V, Gebhardt F, Sahnan K, Williams JA et al. (2007). Transcription factor ERG and joint and Haskard DO et al. (2008). Transcription factor Erg regulates articular cartilage formation during mouse limb and spine angiogenesis and endothelial apoptosis through VE-. Blood skeletogenesis. Dev Biol 305: 40–51. 111: 3498–3506. Klezovitch O, Risk M, Coleman I, Lucas JM, Null M, True LD et al. Bohne A, Schlee C, Mossner M, Thibaut J, Heesch S, Thiel E et al. (2008). A causal role for ERG in neoplastic transformation of (2009). Epigenetic control of differential expression of specific ERG prostate epithelium. Proc Natl Acad Sci USA 105: 2105–2110. isoforms in acute T-lymphoblastic leukemia. Leuk Res 33: 817–812. Kong XT, Ida K, Ichikawa H, Shimizu K, Ohki M, Maseki N et al. Brandon C, Eisenberg LM, Eisenberg CA. (2000). WNT signaling (1997). Consistent detection of TLS/FUS-ERG chimeric transcripts modulates the diversification of hematopoietic cells. Blood 96: in acute myeloid leukemia with t(16;21)(p11;q22) and identification 4132–4141. of a novel transcript. Blood 90: 1192–1199. Chiang J, Majeti R, Becker MW, Tian Q, Lee TL, Yan X et al. (2009). Lai S, Chien AJ, Moon RT. (2009). Wnt/Fz signaling and the Dysregulated gene expression networks in human acute myelogen- cytoskeleton: potential roles in tumorigenesis. Cell Res 19: 532–545. ous leukemia stem cells. Proc Natl Acad Sci USA 106: 3396–3401. Lelinek J, Kuang S, Tong W, Yang H, Lin W, Lee MK et al. (2008). Dennis GJ, Sherman BT, Hosack DA, Yang J, Gao W, Lane HC et al. Genome-wide identification of aberrantly methylated promoter (2003). DAVID: database for annotation, visualization, and associated CpG islands in acute lymphocytic leukemia. Leukemia integrated discovery. Genome Biol 4:3. 22: 1529–1538. Eisenberg CA, Gourdie RG, Eisenberg LM. (1997). Wnt-11 is Lin Z, Reierstad S, Huang C, Bulun SE. (2007). Novel estrogen expressed in early avian mesoderm and required for the differen- receptor-alpha binding sites and estradiol target genes identified by tiation of the quail mesoderm cell line QCE-6. Development 124: chromatin immunoprecipitation cloning in breast cancer. Cancer 525–536. Res 67: 5017–5024.

Oncogene ERG genome-wide screen in acute leukemia LH Mochmann et al 2056 Look A, O0Neil J. (2007). Mechanisms of transcription factor deregula- Singla DK, Schneider DJ, LeWinter MM, Sobel BE. (2006). Wnt3a tion in lymphoid cell transformation. Oncogene 26: 6838–6849. But Not Wnt11 supports self-renewal of embryonic stem cells. Loughran SJ, Kruse EA, Hacking DF, de Graaf CA, Hyland CD, Biochem Biophys Res Commun 345: 789–795. Willson TA et al. (2008). The transcription factor Erg is essential for Soda M, Willert K, Kaushansky K, Geddis AE. (2008). Inhibition of definitive hematopoiesis and the function of adult hematopoietic GSK-3beta promotes survival and proliferation of megakaryocytic stem cells. Nat Immunol 9: 810–819. cells through a beta-catenin-independent pathway. Cellular Signal Luo J. (2009). Glycogen synthase kinase 3beta (GSK-3b) in 20: 2317–2323. tumorigenesis and cancer chemotherapy. Cancer Lett 273: 194–200. Sorensen PH, Lessnic, Stephen L, Lopez-Terrada D, Liu XF, Triche Marcucci G, Baldus CD, Ruppert AS, Whitman SP, Archer KJ, TJ et al. (1994). A second Ewing’s sarcoma translocation, t(21;22), Marcucci G et al. (2005). Overexpression of the ETS-related gene, fuses the EWS gene to another ETS-family transcription factor, ERG, predicts a worse outcome in acute myeloid leukemia with ERG. Nat Genet 6: 146–151. normal karyotype: cancer and leukemia Group B study. J Clin Stankiewicz MJ, Crispino JD. (2009). ETS2 and ERG promote Oncol 23: 9234–9242. megakaryopoiesis and synergize with alterations in GATA-1 to Martin V, Valencia A, Agirre X, Cervera J, San Jose-Eneriz E, immortalize hematopoietic progenitor cells. Blood 113: 3337–3347. Vilas-Zornoza A et al. (2009). Epigenetic regulation of the non- Tomlins SA, Rhodes DR, Perner S, Dhanasekaran SM, Mehra R, canonical Wnt pathway in acute myeloid leukemia. Cancer Sci 3:1–8. Sun XW et al. (2005). Recurrent fusion of TMPRSS2 and Narod S, Seth A, Nam R. (2008). Fusion in the ETS gene family and ETS transcription factor genes in prostate cancer. Science 310: prostate cancer. Br J Cancer 99: 847–851. 644–648. Rainis L, Toki T, Pimanda JE, Rosenthal E, Machol K, Strehl S et al. Uysal-onganer P, Kawano Y, Caro M, Walker MM, Diez S, Darrington (2005). The proto-oncogene ERG in megakaryoblastic leukemias. RS et al. (2010). Wnt-11 promotes neuroendocrine-like differentiation, Cancer Res 65: 7596–7602. survival and migration of prostate cancer cells. Mol Cancer 9: 55. Reya T, Clevers H. (2005). Wnt signalling in stem cells and cancer. Wei G, Badis G, Berger MF, Kivioja T, Palin K, Enge M et al. (2010). Nature 434: 843–850. Genome-wide analysis of ETS-family DNA-binding in vitro and Salek-Ardakani S, Smooha G, De Boer J, Sebire NJ, Morrow M, in vivo. EMBO J 29: 2147–2160. Rainis L et al. (2009). ERG is a megakaryocytic oncogene. Cancer Wu D, Pan W. (2009). GSK3: a multifaceted kinase in Wnt signaling. Res 69: 4665–4673. Trend Biochem Sci 722: 1–8. Sato N, Meijer L, Skaltsounis L, Greengard P, Brivanlou AH. (2004). Zhou W, Lin L, Majumdar A, Li X, Zhang X, Liu W et al. (2007). Maintenance of pluripotency in human and mouse embryonic stem Modulation of morphogenesis by noncanonical Wnt signaling cells through activation of Wnt signaling by a pharmacological requires ATF/CREB family-mediated transcriptional activation of GSK-3-specific inhibitor. Nat Med 10: 55–63. TGFb. Nat Genet 39: 1225–1234. Sharrocks AD. (2001). The Ets-domain transcription factor family. Zhu H, Mazor M, Kawano Y, Walker MM, Leung HY, Armstrong K Nat Rev Mol Cell Biol 2: 827–837. et al Shu J, Jelinek J, Chang H, Shen L, Qin T, Chung W et al. (2006). . (2004). Analysis of Wnt gene expression in prostate cancer: Silencing of bidirectional promoters by DNA methylation in mutual inhibition by WNT11 and the . Cancer tumorigenesis. Cancer Res 66: 5077–5084. Res 64: 7918–7926.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene