Tumor Gene 1 Protein Represses the Expression of the Tumor Suppressor Interferon Regulatory Factor 8 in Human Hematopoietic Progenitors and in Leukemic Cells

Home , Neurofibromin 1, Zinc finger inhibitor

Wilms’ tumor gene 1 protein represses the expression of the tumor suppressor interferon regulatory factor 8 in human hematopoietic progenitors and in leukemic cells

K Vidovic1, E Svensson1, B Nilsson2, B Thuresson1, T Olofsson1, A Lennartsson3 and U Gullberg1

1Department of Hematology, Lund University, Lund, Sweden; 2Cancer Program, Broad Institute, Cambridge, MA, USA and 3Department of Biosciences and Medical Nutrition, Karolinska Institutet, Huddinge, Sweden

Wilms’ tumor gene 1 (WT1) is a transcription factor involved in often interferes with differentiation.15–17 Thus, a large amount of developmental processes. In adult hematopoiesis, only a small data indicates a leukemogenic role of WT1. Despite the fact that portion of early progenitor cells express WT1, whereas most some target genes of WT1 have been suggested (reviewed leukemias show persistently high levels, suggesting an onco- 18 genic role. We have previously characterized oncogenic BCR/ in Yang et al. ), the mechanisms through which WT1 exerts its ABL1 tyrosine kinase signaling pathways for increased WT1 leukemogenic effects remain unclear. expression. In this study, we show that overexpression of BCR/ Interferon regulatory factor 8 (IRF8), also designated ICSBP ABL1 in CD34 þ progenitor cells leads to reduced expression (interferon consensus sequence-binding protein), belongs to the of interferon regulatory factor 8 (IRF8), in addition to increased IRF family of transcription factors and is normally expressed WT1 expression. Interestingly, IRF8 is known as a tumor suppressor in some leukemias and we investigated whether in hematopoietic cells, including monocytes, macrophages WT1 might repress IRF8 expression. When analyzed in four and subsets of lymphocytes, showing inducibility by a- and 19 leukemia mRNA expression data sets, WT1 and IRF8 were g-interferon (reviewed in Tamura et al ). In late myeloid anticorrelated. Upon overexpression in CD34 þ progenitors, as differentiation, IRF8 participates in the commitment to macro- well as in U937 cells, WT1 strongly downregulated IRF8 phage differentiation and is important for the function of mature expression. All four major WT1 splice variants induced repres- macrophages, and several IRF8 target genes in developing sion, but not the zinc-finger-deleted WT1 mutant, indicating 19 dependence on DNA binding. A reporter construct with the IRF8 macrophages have been identified. In contrast to the almost promoter was repressed by WT1, dependent on a putative WT1- ubiquitously high expression of WT1 in leukemias, the expres- 20 response element. Binding of WT1 to the IRF8 promoter was sion of IRF8 is very low or absent in most leukemias. demonstrated by chromatin immunoprecipitation. Our results Interestingly, elimination of IRF8 in transgenic mice results identify IRF8 as a direct target gene for WT1 and provide a in a systemic expansion of granulocytes with later transition possible mechanism for oncogenic effects of WT1 in leukemia. into blast crisis, a development very similar to human chronic Leukemia (2010) 24, 992–1000; doi:10.1038/leu.2010.33; 21 published online 18 March 2010 myeloid leukemia (CML). Similar hematopathological distur- Keywords: WT1; IRF8; promoter; repression bances, although in milder forms, were also observed after heterozygous elimination of IRF8, indicating haploid insuffi- ciency of IRF8.21 The importance of reduced levels of IRF8 in the pathogenesis of CML is further underscored by the observation that forced expression of IRF8 counteracts the BCR/ Introduction ABL1-induced leukemic phenotype in vivo22 and in vitro.23,24 Besides its association with the development of CML, a role Wilms’ tumor gene 1 (WT1) encodes a DNA-binding zinc- of IRF8 is also implicated in acute leukemogenesis; in murine finger transcription factor necessary for the development models for acute leukemia, IRF8 acts as a tumor suppressor, of various organs and structures in the fetus (reviewed in 1 2 the loss of which cooperates with leukemogenic fusion genes Lee and Haber and in Scharnhorst et al. ). Expression of WT1 in NUP98-TOP1 or AML1-ETO,25,26 with the oncogenic SHP2 a small fraction of early progenitor cells in human and murine 27 28 3–8 tyrosine phosphatase and with loss of neurofibromin 1. adult hematopoiesis, as well as analysis of WT1-deficient Antiproliferative mechanisms downstream of IRF8 may include murine progenitors showing impaired reconstitution of hemato- 29 9 upregulation of p15INK4b, but the mechanism by which the poiesis in vivo, indicate a role for WT1 in maintaining proli- level of IRF8 is regulated within the context of leukemia is feration and/or survival of hematopoietic cells. Consistent with unknown. this notion, overexpression of WT1 alone in myeloid progenitors 10 In this study, we report that the WT1 protein, frequently induced a myeloproliferative-like disorder in mice, and overexpressed in leukemia, can directly repress transcription coexpression of WT1 and the fusion protein AML1-ETO resulted 10 of the IRF8 gene, thus providing a novel mechanism by which in a rapid development of acute leukemia, lending further WT1 might contribute to leukemogenesis. support for the role of WT1 as an oncogene in leukemia. High levels of WT1 are detected in most cases of chronic and acute human leukemia,11–14 as well as in most leukemic Materials and methods cell lines. Whereas endogenous WT1 is downregulated after induced differentiation of cell lines, overexpression of WT1 Cell culture Umbilical cord blood was, after ethical approval and informed Correspondence: Dr U Gullberg, Department of Hematology, Lund consent, collected from mothers giving birth to normal full-term University, BMC, C14, Klinikgatan 28, Lund SE-221 84, Sweden. E-mail: [email protected] infants, from which mononuclear cells were isolated by Received 24 September 2009; revised 21 December 2009; accepted separation on Lymphoprep (Nycomed Pharma, Oslo, Norway), 7 January 2010; published online 18 March 2010 and CD34 þ cells were enriched by labeling with magnetic WT1 protein represses IRF8 K Vidovic et al 993 beads (CD34 Progenitor Cell Isolation Kit, Miltenyi Biotec, Bisulfite sequencing Bergisch-Gladbach, Germany) according to the manufacturer’s DNA-methylation status of the IRF8 promoter in U937 cells and instructions. CD34 þ cell purity was always 490% as deter- CD34 þ cells was analyzed by bisulfite sequencing using the EZ mined by flow cytometry. 293T/17 (American Type Culture DNA Methylation-Direct kit (Zymo Research, Orange, CA, USA) Collection, Manassas, VA, USA) cells were maintained in according to the manufacturer’s protocol, using ZymoTaq DNA Dulbecco’s modified Eagle’s medium supplemented with 10% polymerase (Zymo Research) in the PCR reaction. The conver- fetal calf serum. U937 cells were obtained from the German sion rate was 494%. A 361-bp sequence in the IRF8 promoter Collection of Microorganisms and Cell Cultures (Braunschweig, was sequenced after PCR amplification with the following Germany). The CML cell lines KU812, LAMA-84 and JK-1 cells primer pair: 50-TAGTGGTTTTTTTTGAAGTTGGG-30 and 50-TA were kindly donated by Dr Thoas Fioretos (Clinical Genetics, CAAAAAAACTTTCCCAAAAATTC-30 amplifying À564 to À278 Lund University, Lund, Sweden). Both U937 and CML cell lines of the IRF8 promoter. The TOPO TA Cloning Kit (Invitrogen, were maintained in RPMI 1640 supplemented with 10% fetal Carlsbad, CA, USA) and the BigDye Terminator v1.1 Sequencing calf serum. Kit (Applied Biosystems) with M13 forward and M13 reverse primers were used for cloning and sequencing, respectively.

RNA isolation, reverse transcription and RT-qPCR To analyze the expression of WT1 and IRF8 mRNA, total RNA Cloning and site-directed mutagenesis of IRF8 promoter was isolated using the RNeasy Mini Kit or RNeasy Micro Kit Cloning of human IRF8 promoter genomic DNA was performed according to the manufacturer’s protocol (Qiagen Gmbh, Hilden using the GC-Rich PCR Kit (Roche Diagnostics Gmbh, Germany). A total of 200 ng of total RNA was reverse transcribed Mannheim, Germany). The cloned promotor construct con- using High-Capacity cDNA Reverse Transcription Kit (Applied tained 970 bp upstream of the transcription start and 74 bp Biosystems Inc., Foster City, CA, USA) with random hexamer downstream of the transcription start. The following primers primers according to the manufacturer’s instructions. In some were used for cloning: Sense: outer 50-GGTCCCTGAAAAGCTCAACGCA-30 experiments with BCR/ABL1-transduced CD34 þ cells, 500 0 0 GFP þ cells were sorted directly (by a FACS Aria, BD, Franklin andnested5-TATACGCGTGCCCCCATGTGTGATTCTCTAC-3 ; antisense: outer 50-GGGGACATTACGGTAGGTAGTTCC-30 Lakes, NJ, USA) into PCR tubes containing cell lysis buffer, after 0 0 which reverse transcription (RT) reaction was performed using the and nested 5 -GATAAGCTTCCGCCCACTGTGCCTACCT-3 . Sensiscript RT kit (Qiagen) as described elsewhere.30 Quantitative The PCR product was cloned into the pGL3 Basic firefly PCR (qPCR) was carried out using TaqMan probe-based chemistry luciferase reporter vector (MluI/HindIII) (Promega, Madison, (Applied Biosystems). The probes for the WT1 (Hs00240913_m1), WI, USA). Core nucleotides of a potential WT1-binding site in IRF1 (Hs00971960_m1), IRF4 (Hs00277069_m1) IRF8 the promoter were mutated by PCR: À45/À38 CCCCA were (Hs00175238_m1), ISG15 (interferon-stimulated gene 15) mutated to AATAT. (Hs01921425_s1) and the endogenous controls 18S (Hs99999901_s1) and GAPDH (glyceraldehyde-3-phosphate Transient transfections and luciferase-reporter assays dehydrogenase, Hs99999905_m1) were purchased as Assay-on- cDNA encoding WT1( þ 17 amino acid (AA)/ÀKTS) (KTS Demand (Applied Biosystems). The amplification reactions were denoting lysine, threonine, serine) was amplified by PCR from all performed in triplicates in an ABI Prism 7000 Sequence pCMV-CB6 þ /WT1( þ 17AA, ÀKTS) (kindly provided by Dr F Detection System (Applied Biosystems) according to the manu- Rauscher III, Philadelphia, PA, USA) and cloned into pcDNA3 facturer’s instructions, except when analyzing gene expression (Invitrogen) (EcoRI/EcoRV). In triplicates, 100 ng of the IRF8 from only 500 cells, in which case cDNA in 5 ml of the RT promoter reporter construct and 500 ng of pcDNA3 or pcDNA3/ reaction was used in each 25 ml PCR reaction, as described WT1 ( þ 17AA/ÀKTS) were cotransfected into 250 000 293T/17 30 elsewhere. Data were collected and analyzed using the cells in 24-well plates using the Polyfect Kit (Qiagen) according to Sequence Detector v.1.1 software (Applied Biosystems). Relative the manufacturer’s instructions. After 48 h, the cells were lysed 31 quantitative data based on the DDCT method was calculated. and subjected to luciferase assay. Cell lysis and luciferase assay Normalization: DCT ¼ CT (sample)ÀCT(18S, b2–microglobulin or were performed using the Dual luciferase reporter assay kit GAPDH); DDCT ¼ DCT (sample 1)ÀDCT (sample 2). Relative (Promega) and a Glomax 20/20 luminometer (Turner Designs, ÀDDCT quantification ¼ 2 . Efficacy of the PCR amplifications of Sunnyvale, CA, USA) according to the manufacturers’ instructions. controls and tests was identical; parallelism of standard curves of Normalization of firefly luciferase values for transfection effi- the control and test was confirmed. ciency could not be performed because of systematic stimulatory effects exerted by WT1 on Renilla luciferase expression. Western blot analysis Western blotting was performed using the following primary Chromatin immunoprecipitation analysis antibodies: WT1(180) sc-846, ICSBP(c-19) sc-6058, Actin(c-2) Chromatin immunoprecipitation (ChIP) was performed using sc-8432, GAPDH (6C5) sc-32233, all from Santa Cruz Biotech- Low cell ChIP kit from Diagenode (Sparta, NJ, USA) according nology (Santa Cruz, CA, USA). All antibodies were diluted in a to manufacturer’s instructions. Briefly, CD34 þ cells were ratio of 1:500. Starting block blocking buffer (Pierce, Rockford retrovirally transduced with a vector expressing WT1( þ /À)or IL, USA) was used in the incubation steps with primary and empty MIG vector. Forty-eight hours after transduction, 300 000 secondary antibodies. The following secondary antibodies cells were fixed with formaldehyde and used in the subsequent were used: goat-anti-rabbit-horseradish peroxidase 172–1019, immunoprecipitation. In all, 2 mg of WT1 antibody (C-19, sc-192 goat-anti-mouse-horseradish peroxidase 170–6516 and rabbit or F-6, sc-7385, Santa Cruz Biotechnology), or no antibody as anti-sheep-horseradish peroxidase 172–1017, all from Bio-Rad negative control, was used in each precipitation. The precipi- Laboratories (Hercules, CA, USA). Each lane was loaded with tated DNA was analyzed in triplicate with qPCR (Applied material corresponding to 300 000 CD34 þ cells or 30 000 Biosystems 7500), using Power SyBRgreen chemistry (Applied U937 cells. Biosystems). Relative quantitative data based on the DDCT

Leukemia WT1 protein represses IRF8 K Vidovic et al 994 31 method. were calculated. Normalization: DCT ¼ CT (WT1 þ /À expressing cells)ÀCT(input); DDCT ¼ DCT (WT1 þ /À)ÀDCT (control). Relative quantiﬁcation ¼ 2ÀDDCT. Primers overlapping the putative WT1-binding site at À52 to À38 in the IRF8 promoter (forward: 50-TTCTCGGAAAGCAGAGCACTTC-30; reverse: 50-GCCTTAAAAAGGGTCGTGGG-30) and primers binding 3 kbp downstream of the WT1-binding site (forward: 50-TTTTTTTTTGGGTGGGCGTC-30; reverse: 50GACATCCTTTG GAAGAACACTCTCAC-03) were used for ampliﬁcation.

Microarray data sets We used microarray data from the study by Valk et al.32 (adult acute myeloid leukemia (AML), mixed cytogenetic subtypes, n ¼ 232, from NCBI GEO, accession no. GSE1159), the study by Metzeler et al.33 (adult AML, normal karyotype, n ¼ 162, from NCBI GEO, accession no. GSE12417), the study by Verhaak et al.34 (adult AML, mixed cytogenetic subtypes, n ¼ 461, from NCBI GEO, accession no. GSE6891) and the study by Zheng et al.35 (chronic and blast phase CML, n ¼ 11, from ArrayExpress, accession no. E-MEXP-480). All data were trans- formed to the log2 scale. To compute the correlation between WT1 and IRF8, we used the standard Pearson correlation coefﬁcient.

Results

Expression of the BCR/ABL1 fusion in CD34 þ increases expression of WT1 and decreases expression of IRF8 We have previously characterized a signaling pathway from Control BCR/ABL1 BCR/ABL1 leading to increased transcription of the WT1 gene 55 kDa and increased levels of WT1 protein, conferring increased WT1 resistance to imatinib-induced cell death.36 Interestingly, the tumor suppressor IRF8 is downregulated in a murine model of 55 kDa IRF8 BCR/ABL1-induced myeloproliferative disease.22 We therefore investigated whether the expression of IRF8, in addition to WT1, Actin was affected in human progenitors overexpressing BCR/ABL1. 40 kDa Human CD34 þ cord blood cells were retrovirally transduced 36 Figure 1 Overexpression of BCR/ABL1 in CD34 þ cells increases the with human BCR/ABL1 as described previously, and the expression of WT1 and decreases the expression of IRF8. CD34 þ expression of IRF8 was analyzed by RT-qPCR at 24 h after cord blood cells were transduced with BCR/ABL1 or empty vector transduction. Indeed, the levels of IRF8 mRNA (Figure 1a) and (control), and 24 h after transduction, GFP þ cells were sorted and IRF8 protein (Figure 1b) were strongly reduced in BCR/ABL1- levels of WT1 and IRF8 were analyzed. Transduction efficiency was expressing progenitor cells. As previously reported,36 WT1 56% (mean, control cells) and 14% (mean, BCR/ABL1 cells). (a) RT- qPCR analysis of the WT1 and IRF8 mRNA expression. 18S mRNA was mRNA and protein was increased in response to BCR/ABL1 used as calibrator and the RT-qPCR reactions were performed in (Figures 1a and b). These data confirm and extend previous triplicates in each experiment (mean values±s.e.m.). Control is set to 1. results22,36,37 and made us hypothesize that WT1 downmodu- Two independent experiments are shown. (b) Western blot analysis of lates the expression of IRF8. WT1 (WT1(180) sc-846) and IRF8 (ICSBP(c-19) sc-6058) protein levels. Actin (Actin(c-2) sc-8432) is shown as equal loading control. Molecular weight markers are shown in the left. One representative blot is shown. The data for WT1 expression in response to BCR/ABL1 A negative correlation between WT1 and IRF8 are previously reported in the study by Svensson et al.36 expression Although WT1 is strongly expressed in most leukemias,11–14 levels of IRF8 are reported to be low or absent.20 To gain support myeloid leukemias (as described in the ‘Materials and methods’ for the notion of a WT1-induced downmodulation of IRF8 section). In total, we retrieved 866 Affymetrix U133A or expression, we analyzed mRNA levels in different leukemic cell U133A þ B gene expression profiles (Affymetrix, Santa Clara, lines. Most cell lines were expressing WT1, whereas the IRF8 CA, USA) from three studies on adult AML and one study on expression was low or undetectable, especially in the CML CML. In the AML data sets, we observed a highly significant cell lines K562, KU812, LAMA-84 and JK-1, and also in the AML anticorrelation between the expression of WT1 and cell line HEL. Furthermore, the lymphoma cell lines Namalwa IRF8 (r ¼À0.36, P ¼ 5.8 Â 10À9 in GSE1159; r ¼À0.48, and Ramos expressed moderate IRF8 mRNA levels but had no P ¼ 2.6 Â 10À28 in GSE6891; and r ¼À0.43, P ¼ 3.7 Â 10À9)in detectable WT1 mRNA levels (data not shown). Next, we GSE12417. An anticorrelation was also present in the CML data extended the analysis of the correlation between WT1 and IRF8 set (r ¼À0.72, P ¼ 0.00019). We conclude that the expression of to primary AML and CML samples by analyzing the expression WT1 and IRF8 are anticorrelated both in cell lines and in pattern of WT1 and IRF8 mRNA in microarray data sets. We primary human leukemias, compatible with the notion of WT1 compiled a compendium of gene expression profiles of primary as a repressor of IRF8.

Leukemia WT1 protein represses IRF8 K Vidovic et al 995 Forced expression of WT1 inhibits IRF8 expression strongest effects, reducing IRF8 expression to B40% of that in To investigate whether WT1 can repress mRNA expression of control (Figure 3a). WT1 þ / þ , and in particular WT1À/À, were IRF8, human CD34 þ progenitor cells were retrovirally trans- less effective. WT1delZ did not affect levels of IRF8, indicating duced with WT1 as described previously,36 after which IRF8 dependence on DNA binding for repression of IRF8 (Figure 3a). expression was determined. As shown in Figure 2a, expression We conclude that WT1 represses IRF8 expression in normal of WT1 reduced IRF8 expression to 20–30% of that observed in progenitors and in leukemic cells, and that the strongest control cells. The reduction of IRF8 mRNA correlated with a repression is mediated by the WT1 þ /À splice variant. decrease in IRF8 protein levels, as determined by immunoblot- ting with an anti-IRF8 antibody (IRF8(c-19), Santa Cruz Biotechnology) (Figure 2b). The decrease of IRF8 protein was The repression of IRF8 is not due to general inhibition of also verified with another IRF8 antibody (AF5117, R&D Systems interferon signaling Inc., Minneapolis, MN, USA) (data not shown). There are four IRF8 expression is strongly induced in response to interferon-g, major isoforms of WT1, including or excluding a 17 amino-acid and also to a lesser extent by interferon-a.22 Any interference insert (17AA) in the amino-terminal part, and a 3 amino-acid with interferon signaling thus could have a negative impact on insert (KTS) between zinc-fingers 2 and 3 in the carboxyterminal IRF8 expression. Interestingly, a functional interaction between part of the protein. Although the KTS insert can affect DNA STAT3, involved in interferon signaling, and WT1 has been binding, the 17AA insert may modulate the transactivation reported.38 We therefore wanted to investigate whether WT1 exerted by WT1 (reviewed in Lee and Haber1 and in Scharnhorst induced a general repression of interferon-responsive genes. et al2). To investigate whether the repressive effect of WT1 on To that end, we determined the expression of IRF1, IRF4 and IRF8 expression was specific for certain isoforms of WT1, we ISG15, all well characterized interferon-induced genes overexpressed WT1 À17AA/ÀKTS (WT1À/À), WT1 þ 17AA/ (reviewed in Tamura et al.19), in response to WT1 þ /À in ÀKTS (WT1 þ /À), WT1 À17AA/ þ KTS (WT1À/ þ ) or WT1 U937 cells. Interestingly, none of these genes responded with þ 17AA/ þ KTS (WT1 þ / þ ) in leukemic U937 cells lacking repression in response to WT1. Rather, some WT1 isoforms endogenous WT1 protein, but with high expression of IRF8. upregulated the expression of IRF1, IRF4 and ISG15 (Figures 3b WT1delZ, lacking the entire carboxyterminal zinc-finger and c). To assure that similar amounts of WT1 protein were domain and thus with interrupted DNA-binding capacity, was expressed in the cells, a western blot analysis of WT1 was used as a negative control. WT1 þ /À and WT1À/ þ showed the performed, showing comparable levels of the different isoforms (Figure 3e). Upon expression of WT1 þ /À, a band corresponding to WT1delZ occurred (Figure 3e), the identity of which is not 150 clear, but it could hypothetically indicate processing of WT1 þ /À into the delZ form. Importantly, it does not affect the main conclusion that WT1 represses IRF8, as repression is 100 a common effect among the full-length WT1 variants. We conclude that the WT1-induced repression is not a result of a general inhibition of interferon-responsive genes, but rather is specific for IRF8. 50

WT1 expression does not affect the methylation of the 0 IRF8 promoter Relative amount of IRF-8 mRNA Relative Recent data indicate that IRF8 expression may be silenced by promoter methylation in carcinomas.39–41 In AML, 7 out of 42 WT1 (+/-) 21 analyzed samples showed signs of promoter methylation. MIG (control) However, high expression of IRF8 in the presence of methylated promoter has been reported for leukemic cell lines.43 To investigate whether WT1-induced repression of IRF8 correlated with increased methylation of the IRF8 gene, we performed Control WT1 (+/-) 55 kDa bisulﬁte sequencing of DNA from human progenitor cells transduced with WT1. A 256 bp CpG-rich region of the promoter, À564 to À278, was analyzed. This region was chosen as it is 40,41 IRF8 methylated in some solid cancers. As shown in Figure 4, the analyzed part of the IRF8 promoter was almost completely unmethylated both in the absence or presence of WT1, in spite of reduced IRF8 levels (Figure 2a). In contrast, U937 cells 43 36 kDa showed 494% methylation, consistent with a previous report. GAPDH Although we cannot exclude that long-lasting repression of Figure 2 WT1 represses the expression of IRF8. CD34 þ cord blood the IRF8 gene may lead to methylation, we conclude that cells were retrovirally transduced for 48 h with WT1 þ /À or with immediate WT1-induced repression of IRF8 is not due to empty control vector (MIG), after which GFP þ cells were sorted and increased methylation of the IRF8 gene. analyzed for IRF8 mRNA and protein. Transduction efﬁciency was 50% (mean, control cells) and 41% (mean, WT1 cells). (a) RT-qPCR analysis of IRF8 mRNA levels (mean values±s.e.m, n ¼ 4). Control is set to 100%. (b) Western blot analysis of IRF8 (ICSBP(C-19) sc-6058). WT1 represses IRF8 promoter activity GAPDH (GAPDH(6C5) sc-32233) is shown as equal loading. Next, we investigated whether WT1 affects IRF8 promoter Molecular weight markers are shown in the left. One representative activity. In human hematopoietic cells, one major transcription blot is shown. start site is used.43 We investigated the human promoter and its

Leukemia WT1 protein represses IRF8 K Vidovic et al 996

Figure 3 Interferon-responsive genes are not generally downmodulated by WT1. U937 cells were retrovirally transduced with different WT1 isoforms: WT1À/À,WT1þ /À,WT1À/ þ ,WT1þ / þ and also with the zinc-ﬁnger-deleted WT1 protein variant, WT1delZ. GFP þ cells were sorted after 48 h and analyzed for IRF8 mRNA by RT-qPCR. (a–d) mRNA levels of IRF8, IRF1, IRF4 and ISG15, relative levels as compared with control cells (mean values±s.e.m, n ¼ 4). (e) Western blot analysis of different WT1 isoforms (WT1(180) sc-846) in the transduced U937 cells. GAPDH (GAPDH(6C5) sc-32233) is shown as equal loading. One representative blot is shown.

possible regulation by WT1 through cloning of the proximal repression from overexpression of WT1 (Figure 5), supporting sequence upstream of the transcription start site (À970 to þ 74) the conclusion that WT1 can bind to the nonmutated promoter into the luciferase reporter pGL3 as described in the ‘Materials and repress transcription. and methods’ section, after which reporter transfections were performed in 293T/17 cells, in which the activity of the promoter in response to overexpressed WT1 was determined. WT1 binds to the IRF8 promoter Cotransfection with WT1 þ /À resulted in a modest repression To gain evidence for binding of WT1 to the indicated part of (23%) of promoter activity (Figure 5), indicating binding of the promoter, we performed ChIP analysis. CD34 þ cells were WT1. Inspection of the promoter sequence using MatInspector transduced with WT1 þ /À and 48 h after transduction ChIP was (Genomatix, Munich, Germany; http://www.genomatix.de) in- performed, followed by qPCR amplification of the promoter dicated a possible WT1-binding site at position À52 to À38. region containing the WT1-binding site (À52 to À38) (WT1 site). This site was mutated to abolish WT1 binding, as described in qPCR amplification of a region 3 kbp downstream of the binding the ‘Materials and methods’ section, after which the promoter site (3 kb d.s.) was used as negative control. Two distinct WT1- constructs were used in reporter transfection assays. Impor- specific antibodies were used separately in the immunopreci- tantly, mutated promoter was completely unresponsive to pitation step, to verify the specificity of the ChIP analysis.

Leukemia WT1 protein represses IRF8 K Vidovic et al 997 CD34+ control

-564 -278

CD34+ WT1 +/-

-564 -278

U937

-564 -278

Figure 4 Methylation of the IRF8 promoter is not affected by WT1 overexpression. A schematic picture of the methylation status in the 50- CpG-30dinucleotides in the À564 to À278 sequence of the IRF8 proximal promoter in transduced CD34 þ cells and in nontransduced U937 cells is shown. Filled squares (’) indicate methylated cytosines and unfilled squares (&) indicate unmethylated cytosines. All 38 50- CpG-30dinucleotides, in the region À564 to À278 of the IRF8 proximal promoter, were analyzed. Four distinct clones from each cell type were analyzed. The overall bisulfite conversion efficiency rate was 494%.

Figure 6 WT1 binds to the IRF8 promoter. CD34 þ cord blood cells were retrovirally transduced with WT1( þ /À) or with empty control vector. GFP þ cells were sorted after 48 h and WT1 binding to the promoter was analyzed by ChIP, followed by qPCR amplification of promoter DNA containing a WT1-binding site (WT1 site), in triplicate in each experiment (mean values±s.e.m.). qPCR amplification of a region 3 kbp downstream of the transcription start site (3 kb d.s.) was used as a negative control. The results of three distinct immuno- precipitations, using two different WT1 antibodies (C-19/sc-192 or F-6/sc-7385, both from Santa Cruz Biotechnology) are shown. Fold DNA enrichment in WT1 þ /À expressing cells (filled bars, ’), relative to control cells (open bars, &).

Figure 5 WT1 represses transcription from the IRF8 promoter. 293T/ 17 cells were transfected with a promoter luciferase reporter without Discussion (control) or with cotransfected WT1 þ /À, after which luciferase activity was analyzed. The effect of WT1 on IRF8 promoter (IRF8 prom) and on promoter with mutated WT1-binding site (IRF8 prom The main conclusion in this study is that the transcription factor mut) is shown. The relative luciferase activity is shown, with the WT1 can bind to the promoter of the IRF8 gene and repress its activity from IRF8 promoter in control cells set to 1, mean transcription, thus reducing expression of IRF8. This conclusion values±s.e.m, n ¼ 7. is based on several ﬁndings that: (1) expression of WT1 resulted in a decrease of IRF8 mRNA and protein levels in CD34 þ progenitor cells, as well as in leukemic U937 cells; (2) Precipitation with either antibody resulted in a three- to eight- repression was dependent on DNA binding of WT1, as the fold enrichment of IRF8 promoter DNA containing the WT1- WT1 (delZ) mutant, with no DNA-binding capacity, showed no binding site (Figure 6). In contrast, no enrichment of control effect on IRF8 expression; (3) that a promoter reporter was DNA 3 kbp downstream of the transcription start was detected repressed by WT1, dependent on a potential WT1-binding site (Figure 6). No DNA ampliﬁcation signal could be detected after in the promoter; and (4) WT1 binds to the promoter, as judged negative control immunoprecipitation using no antibody (data by ChIP. Moreover, statistic analysis of expression levels of WT1 not shown). and IRF8 in leukemia data sets showed an anticorrelation

Leukemia WT1 protein represses IRF8 K Vidovic et al 998 between WT1 and IRF8 in AML and CML samples, both with represses an IRF8-promoter luciferase reporter supports this and without known genetic aberrations, with a robust negative notion. However, in the luciferase assay performed, only a correlation, r ¼À0.36 to r ¼À0.72 depending on the magnitude rather weak repression (23%) on the proximal IRF8 promoter of the data sets. It is noteworthy that when compared with WT1, was observed. This might indicate that other regions, outside the IRF8 was among the most anticorrelated genes in the evaluated analyzed 974 bp of the proximal promoter, contributes to gene lists: 50th of 54 614 gene probes (GSE6891), 138th of repression of the endogenous gene. We found WT1 to systema- 44 693 (GSE12417) and 249th of 22 216 (GSE1159). We also tically increase the expression of both SV40 (simian virus 40)- found that high WT1 levels anticorrelated with low amounts of and TK (thymidine kinase)-driven renilla expression. Therefore, IRF8 transcripts in clinical CML samples.35 Thus, experimental normalization of luciferase results would result in a much data, and expression patterns in primary leukemia samples, stronger apparent WT1-mediated repression of IRF8. As a support the conclusion that WT1 can mediate repression consequence of this, we chose not to normalize the results, of IRF8. but rather to conduct a large number of independent experiments What is the molecular mechanism for WT1-mediated repres- (n ¼ 7). Further experimental support for our identification of the sion of IRF8? The transcription factor STAT1 has been linked WT1-binding site in the promoter, was given by our results from to interferon-induced expression of IRF8, mediated by a STAT1- ChIP, performed with two distinct WT1 antibodies, showing an responsive element in close proximity of the currently identified enrichment of the IRF8 promoter, as compared with the negative WT1-response element.41 A functional interaction between control sequence. Taken together, our data indicate a model in WT1 and STAT3 is reported,38 giving rise to the question which WT1 itself directly binds to the IRF8 promoter and represses whether WT1 affects IRF8 expression by interference with transcription, directly or through some corepressor molecule. STAT1. However, we found that the effect of WT1 is specific for Early reports of overexpression of WT1 in a high proportion of IRF8 among interferon-responsive genes, as the levels of various leukemias suggested a role for WT1 in leukemo- IRF1, IRF4 and ISG15 in transduced cells were not decreased. genesis.11–14 Consistently, overexpression of WT1 interferes with Rather, some isoforms of WT1 induced increased expression of differentiation of leukemic cells and WT1 can signal for survival these genes, arguing against an inhibition of STAT1 signaling and proliferation.15,46 Strong support for an oncogenic role of as the mechanism for WT1-mediated repression of IRF8. The WT1 has recently been presented; WT1 induces a myeloproli- indication that WT1 might upregulate IRF1, IRF4 and ISG15 ferative disease in mice and cooperates with a leukemic fusion was however not corroborated when expression levels were protein in the development of acute leukemia.10 The mechanism investigated in leukemic cell lines or in leukemic mRNA by which WT1 contributes to leukemogenesis is unknown, but expression data sets (data not shown), indicating that upregula- the presence of WT1 in several subtypes of chronic and acute tion of these genes by WT1 is not a general phenomenon. leukemias suggests common mechanisms operating in several In some cases of carcinoma, the tumor suppressor IRF8 is forms of leukemic cells. There is accumulating evidence for IRF8 silenced by strong methylation of the promoter,39,41 leading to as a tumor suppressor gene in myeloid neoplasms. IRF8-deficient perturbed response to INFg.44 The methylation status of the mice develop a disease resembling CML with an expansion of the proximal promoter near the transcription start site of IRF8 was myeloid lineage,22 and decrease of IRF8 levels is also implicated previously investigated in U937 cells. In this cell line, active in acute leukemia.25–27 transcription of IRF8 occurs from a methylated IRF8 promoter,43 In conclusion, we show that WT1 can mediate repression of questioning methylation of the proximal IRF8 promoter as an IRF8 expression. Given the known antileukemic effects of IRF8, important transcriptional regulation mechanism in leukemic we propose that this mechanism may be a common way by cells. Consistently, we found that primary progenitor cells with which WT1 contributes to leukemogenesis. overexpressed WT1, and therefore reduced IRF8 expression, showed no detectable methylation of the investigated CpG islands (À278 to À564 bp). Therefore, we conclude that Conflict of interest WT1-mediated repression of IRF8 does not depend on promoter methylation, although we cannot exclude that methylation may The authors declare no conflict of interest. evolve with time after long-standing repression. Although most reports of WT1 as a transcription factor Acknowledgements emphasize the transactivation potential of WT1, growth factor receptors such as the epidermal growth factor receptor and We thank Dr Thoas Fioretos (Lund, Sweden) for providing us with insulin growth factor 1 receptor (IGFR-1), and also cell-cycle the KU812, LAMA-84 and JK-1 cell lines and Dr F Rauscher III regulators such as cyclin E and ornithine decarboxylase and the (Philadelphia, PA) for the kind gift of the pCMV-CB6 þ /WT1 enzyme telomerase RT are reported as repression targets of WT1 18 ( þ 17aa, -KTS) plasmid. This study was supported by grants from (reviewed in Yang et al. ). Among the WT1 splice variants, the the Medical Faculty of Lund (ALF), the Swedish Cancer Society, ÀKTS variant of the zinc-finger domain most often shows the 45 the Swedish Research Council, the Swedish Children’s Cancer strongest DNA binding, and transcriptionally repressive Foundation, the Gunnar Nilsson Cancer Foundation, the O¨ ster- domains in the amino-terminal of WT1 have been defined lund Foundation and funds from Lund University Hospital. (reviewed in Lee and Haber1, Scharnhorst et al.2 and in Yang et al.18). In the present investigation, the WT1 þ 17AA/ÀKTS variant showed the strongest effects on IRF8 expression, but References the other splice variants also, with the possible exception for WT1À17AA/ÀKTS, showed repressive effects on IRF8. 1 Lee SB, Haber DA. Wilms tumor and the WT1 gene. Exp Cell Res Therefore, from present data, we cannot determine which domains 2001; 264: 74–99. in WT1 are critical for repression of IRF8. However, our finding 2 Scharnhorst V, van der Eb AJ, Jochemsen AG. WT1 proteins: functions in growth and differentiation. Gene 2001; 273: 141–161. that a WT1 mutant completely lacking the DNA-binding domain 3 Fraizer GC, Patmasiriwat P, Zhang X, Saunders GF. Expression of (WT1delZ) was without effect, strongly indicates that repression the tumor suppressor gene WT1 in both human and mouse bone is dependent on DNA binding of WT1. Our finding that WT1 marrow. Blood 1995; 86: 4704–4706.

Leukemia WT1 protein represses IRF8 K Vidovic et al 999 4 Baird PN, Simmons PJ. Expression of the Wilms’ tumor in differentiating myeloid progenitor cells. Blood 2003; 102: gene (WT1) in normal hemopoiesis. Exp Hematol 1997; 25: 4547–4554. 312–320. 24 Burchert A, Cai D, Hofbauer LC, Samuelsson MK, Slater EP, 5 Maurer U, Brieger J, Weidmann E, Mitrou PS, Hoelzer D, Duyster J et al. Interferon consensus sequence binding protein Bergmann L. The Wilms’ tumor gene is expressed in a subset of (ICSBP; IRF-8) antagonizes BCR/ABL and down-regulates bcl-2. CD34+ progenitors and downregulated early in the course of Blood 2004; 103: 3480–3489. differentiation in vitro. Exp Hematol 1997; 25: 945–950. 25 Schwieger M, Lohler J, Friel J, Scheller M, Horak I, Stocking C. 6 Menssen HD, Renkl HJ, Entezami M, Thiel E. Wilms’ tumor gene AML1-ETO inhibits maturation of multiple lymphohematopoietic expression in human CD34+ hematopoietic progenitors during lineages and induces myeloblast transformation in synergy with fetal development and early clonogenic growth. Blood 1997; 89: ICSBP deficiency. J Exp Med 2002; 196: 1227–1240. 3486–3487. 26 Gurevich RM, Rosten PM, Schwieger M, Stocking C, Humphries 7 Hosen N, Sonoda Y, Oji Y, Kimura T, Minamiguchi H, Tamaki H RK. Retroviral integration site analysis identifies ICSBP as a et al. Very low frequencies of human normal CD34+ haemato- collaborating tumor suppressor gene in NUP98-TOP1-induced poietic progenitor cells express the Wilms’ tumour gene WT1 at leukemia. Exp Hematol 2006; 34: 1192–1201. levels similar to those in leukaemia cells. Br J Haematol 2002; 116: 27 Konieczna I, Horvath E, Wang H, Lindsey S, Saberwal G, Bei L 409–420. et al. Constitutive activation of SHP2 in mice cooperates with 8 Hosen N, Shirakata T, Nishida S, Yanagihara M, Tsuboi A, ICSBP deficiency to accelerate progression to acute myeloid Kawakami M et al. The Wilms’ tumor gene WT1-GFP knock-in leukemia. J Clin Invest 2008; 118: 853–867. mouse reveals the dynamic regulation of WT1 expression 28 Koenigsmann J, Rudolph C, Sander S, Kershaw O, Gruber AD, in normal and leukemic hematopoiesis. Leukemia 2007; 21: Bullinger L et al. Nf1 haploinsufficiency and ICSBP deficiency 1783–1791. synergize in the development of leukemias. Blood 2009; 113: 9 Alberta JA, Springett GM, Rayburn H, Natoli TA, Loring J, 4690–4701. Kreidberg JA et al. Role of the WT1 tumor suppressor in murine 29 Schmidt M, Bies J, Tamura T, Ozato K, Wolff L. The interferon hematopoiesis. Blood 2003; 101: 2570–2574. regulatory factor ICSBP/IRF-8 in combination with PU.1 up- 10 Nishida S, Hosen N, Shirakata T, Kanato K, Yanagihara M, regulates expression of tumor suppressor p15(Ink4b) in murine Nakatsuka S et al. AML1-ETO rapidly induces acute myeloblastic myeloid cells. Blood 2004; 103: 4142–4149. leukemia in cooperation with Wilms’ tumor gene, WT1. Blood 30 Edvardsson L, Dykes J, Olsson ML, Olofsson T. Clonogenicity, 2005; 107: 3303–3312. gene expression and phenotype during neutrophil versus erythroid 11 Miwa H, Beran M, Saunders GF. Expression of the Wilms’ differentiation of cytokine-stimulated CD34+ human marrow cells tumor gene (WT1) in human leukemias. Leukemia 1992; 6: in vitro. Br J Haematol 2004; 127: 451–463. 405–409. 31 Ginzinger DG. Gene quantification using real-time quantitative 12 Miyagi T, Ahuja H, Kubota T, Kubonishi I, Koeffler HP, Miyoshi I. PCR: an emerging technology hits the mainstream. Exp Hematol Expression of the candidate Wilm’s tumor gene, WT1, in human 2002; 30: 503–512. leukemia cells. Leukemia 1993; 7: 970–977. 32 Valk PJ, Verhaak RG, Beijen MA, Erpelinck CA, Barjesteh van 13 Inoue K, Sugiyama H, Ogawa H, Nakagawa M, Yamagami T, Waalwijk van Doorn-Khosrovani S, Boer JM et al. Prognostically Miwa H et al. WT1 as a new prognostic factor and a new marker useful gene-expression profiles in acute myeloid leukemia. N Engl for the detection of minimal residual disease in acute leukemia. J Med 2004; 350: 1617–1628. Blood 1994; 84: 3071–3079. 33 Metzeler KH, Hummel M, Bloomfield CD, Spiekermann K, 14 Menssen HD, Renkl HJ, Rodeck U, Maurer J, Notter M, Schwartz S Braess J, Sauerland MC et al. An 86-probe-set gene-expression et al. Presence of Wilms’ tumor gene (wt1) transcripts and the WT1 signature predicts survival in cytogenetically normal acute myeloid nuclear protein in the majority of human acute leukemias. leukemias. Blood 2008; 112: 4193–4201. Leukemia 1995; 9: 1060–1067. 34 Verhaak RG, Wouters BJ, Erpelinck CA, Abbas S, Beverloo HB, 15 Inoue K, Tamaki H, Ogawa H, Oka Y, Soma T, Tatekawa T et al. Lugthart S et al. Prediction of molecular subtypes in acute myeloid Wilms’ tumor gene (WT1) competes with differentiation-inducing leukemia based on gene expression profiling. Haematological signal in hematopoietic progenitor cells. Blood 1998; 91: 2009; 94: 131–134. 2969–2976. 35 Zheng C, Li L, Haak M, Brors B, Frank O, Giehl M et al. 16 Svedberg H, Chylicki K, Baldetorp B, Rauscher 3rd FJ, Gullberg U. Gene expression profiling of CD34+ cells identifies a molecular Constitutive expression of the Wilms’ tumor gene (WT1) in the signature of chronic myeloid leukemia blast crisis. Leukemia 2006; leukemic cell line U937 blocks parts of the differentiation 20: 1028–1034. program. Oncogene 1998; 16: 925–932. 36 Svensson E, Vidovic K, Lassen C, Richter J, Olofsson T, Fioretos T 17 Tsuboi A, Oka Y, Ogawa H, Elisseeva OA, Tamaki H, Oji Y et al. et al. Deregulation of the Wilms’ tumour gene 1 protein (WT1) by Constitutive expression of the Wilms’ tumor gene WT1 inhibits BCR/ABL1 mediates resistance to imatinib in human leukaemia the differentiation of myeloid progenitor cells but promotes their cells. Leukemia 2007; 21: 2485–2494. proliferation in response to granulocyte-colony stimulating factor 37 Cilloni D, Messa F, Gottardi E, Fava M, Arruga F, Defilippi I et al. (G-CSF). Leuk Res 1999; 23: 499–505. Sensitivity to imatinib therapy may be predicted by testing Wilms 18 Yang L, Han Y, Suarez Saiz F, Minden MD. A tumor suppressor tumor gene expression and colony growth after a short in vitro and oncogene: the WT1 story. Leukemia 2007; 21: 868–876. incubation. Cancer 2004; 101: 979–988. 19 Tamura T, Yanai H, Savitsky D, Taniguchi T. The IRF family 38 Rong Y, Cheng L, Ning H, Zou J, Zhang Y, Xu F et al. Wilms’ transcription factors in immunity and oncogenesis. Annu Rev tumor 1 and signal transducers and activators of transcription 3 Immunol 2008; 26: 535–584. synergistically promote cell proliferation: a possible mechanism in 20 Schmidt M, Nagel S, Proba J, Thiede C, Ritter M, Waring JF sporadic Wilms’ tumor. Cancer Res 2006; 66: 8049–8057. et al. Lack of interferon consensus sequence binding protein 39 Egwuagu CE, Li W, Yu CR, Che Mei Lin M, Chan CC, Nakamura T (ICSBP) transcripts in human myeloid leukemias. Blood 1998; 91: et al. Interferon-gamma induces regression of epithelial cell 22–29. carcinoma: critical roles of IRF-1 and ICSBP transcription factors. 21 Holtschke T, Lohler J, Kanno Y, Fehr T, Giese N, Rosenbauer F Oncogene 2006; 25: 3670–3679. et al. Immunodeficiency and chronic myelogenous leukemia- 40 Yang D, Thangaraju M, Browning DD, Dong Z, Korchin B, Lev DC like syndrome in mice with a targeted mutation of the ICSBP gene. et al. Repression of IFN regulatory factor 8 by DNA methylation is Cell 1996; 87: 307–317. a molecular determinant of apoptotic resistance and metastatic 22 Hao SX, Ren R. Expression of interferon consensus sequence phenotype in metastatic tumor cells. Cancer Res 2007; 67: binding protein (ICSBP) is downregulated in Bcr-Abl-induced 3301–3309. murine chronic myelogenous leukemia-like disease, and forced 41 McGough JM, Yang D, Huang S, Georgi D, Hewitt SM, Ro¨cken C coexpression of ICSBP inhibits Bcr-Abl-induced myeloproliferative et al. DNA methylation represses IFN-gamma-induced and signal disorder. Mol Cell Biol 2000; 20: 1149–1161. transducer and activator of transcription 1-mediated IFN regulatory 23 Tamura T, Kong HJ, Tunyaplin C, Tsujimura H, Calame K, factor 8 activation in colon carcinoma cells. Mol Cancer Res 2008; Ozato K. ICSBP/IRF-8 inhibits mitogenic activity of p210 Bcr/Abl 6: 1841–1851.

Leukemia WT1 protein represses IRF8 K Vidovic et al 1000 42 Gebhard C, Schwarzﬁscher L, Pham TH, Andreesen R, Mackensen nasopharyngeal, esophageal and multiple other carcinomas. A, Rehli M. Rapid and sensitive detection of CpG-methylation Oncogene 2008; 27: 5267–5276. using methyl-binding (MB)-PCR. Nucleic Acids Res 2006; 34: e82. 45 Laity JH, Dyson HJ, Wright PE. Molecular basis for modulation 43 Tshuikina M, Nilsson K, Oberg F. Positive histone marks are of biological function by alternate splicing of the Wilms’ tumor associated with active transcription from a methylated ICSBP/IRF8 suppressor protein. Proc Natl Acad Sci USA 2000; 97: 11932–11935. gene. Gene 2008; 410: 259–267. 46 Algar EM, Khromykh T, Smith SI, Blackburn DM, Bryson GJ, 44 Lee KY, Geng H, Ng KM, Yu J, van Hasselt A, Cao Y et al. Smith PJ. A WT1 antisense oligonucleotide inhibits proliferation Epigenetic disruption of interferon-gamma response through and induces apoptosis in myeloid leukaemia cell lines. Oncogene silencing the tumor suppressor interferon regulatory factor 8 in 1996; 12: 1005–1014.

Leukemia