Activation of TLX3 and NKX2-5 in T(5;14)(Q35;Q32) T-Cell Acute Lymphoblastic Leukemia by Remote 3¶-BCL11B Enhancers and Coregulation by PU.1 and HMGA1

Research Article

Activation of TLX3 and NKX2-5 in t(5;14)(q35;q32) T-Cell Acute Lymphoblastic Leukemia by Remote 3¶-BCL11B Enhancers and Coregulation by PU.1 and HMGA1

Stefan Nagel,1 Michaela Scherr,2 Alexander Kel,3 Klaus Hornischer,3 Gregory E. Crawford,4 Maren Kaufmann,1 Corinna Meyer,1 Hans G. Drexler,1 and Roderick A.F. MacLeod1

1Deutsche Sammlung von Mikroorganismen und Zellkulturen (German Collection of Microorganisms and Cell Cultures), Department of Cell Cultures, Braunschweig, Germany; 2Department of Hematology, Hemostasis and Oncology, Hanover Medical School, Hanover, Germany; 3BIOBASE GmbH, Wolfenbu¨ttel, Germany; and 4Duke University, Institute for Genome Sciences and Policy, Durham, North Carolina

Abstract Introduction In T-cell acute lymphoblastic leukemia, alternative t(5;14)- BCL11B (also known as CTIP2/Rit1) is a Kru¨ppel familyzinc (q35;q32.2) forms effect dysregulation of either TLX3 or finger gene located at 14q32, which is expressed during thymocyte À À NKX2-5 homeobox genes at 5q35 by juxtaposition with maturation from CD4 /CD8 to CD4+/CD8+ stages (1). BCL11B 14q32.2 breakpoints dispersed across the BCL11B downstream mayact either as a tumor-suppressor gene, first detected via genomic desert.Leukemic gene dysregulation by t(5;14) was inactivating mutations or deletions in murine radiation-induced investigated by DNA inhibitory treatments with 26-mer lymphomas (2) and more recently in T-cell acute lymphoblastic double-stranded DNA oligonucleotides directed against can- leukemia (T-cell ALL; ref. 3); or, as an oncogenic activator, first didate enhancers at, or near, orphan T-cell DNase I hypersen- detected byits juxtaposition with the TLX3 (also known as sitive sites located between 3¶-BCL11B and VRK1. NKX2-5 HOX11L2) homeobox gene at 5q35 via the recurrent t(5;14) down-regulation in t(5;14) PEER cells was almost entirely (q35;q32) rearrangement also in T-cell ALL (4–6). A neighboring restricted to DNA inhibitory treatment targeting enhancers related homeobox gene, NKX2-5 (also known as CSX), is similarly within the distal breakpoint cluster region and was dose and activated in T-cell ALL bya variant t(5;14)(q35;q32) or by sequence dependent, whereas enhancers near 3¶-BCL11B t(5;14)(q35;q11.2) involving the T-cell receptor D locus at 14q11.2 regulated that gene only.Chromatin immunoprecipitation (7, 8). Recently, interleukin-2 has been identified as a transcrip- assays showed that the four most effectual NKX2-5 ectopic tional target of BCL11B in both developing and transformed T enhancers were hyperacetylated.These enhancers clustered cells (9). The ectopic expression of Nirenberg-Kim family f1 Mbp downstream of BCL11B, within a region displaying homeobox genes (e.g., TLX3) has been recentlyshown to drive multiple regulatory stigmata, including a TCRA enhancer proliferation of myeloid progenitor cells in a mouse model (10). motif, deep sequence conservation, and tight nuclear matrix Nevertheless, the mechanism underlying leukemic deregulation by attachment relaxed by trichostatin A treatment.Intriguingly, t(5;14) remains obscure, mainlybecause 14q32.2 breakpoints are although TLX3/NKX2-5 promoter/exon 1 regions were hypo- widelyscattered over the f1 Mbp downstream breakpoint cluster acetylated, their expression was trichostatin A sensitive, region. Analogies with T-cell receptor loci and the nature of the implying extrinsic regulation by factor(s) under acetylation ‘‘genomic desert’’ covering 3¶-BCL11B support the existence of PU.1 control.Knockdown of , known to be trichostatin A underlying interactions between TLX3/NKX2-5 and remote TLX3 NKX2-5 responsive and which potentially binds / pro- transcriptional enhancers. moters, effected down-regulation of both homeobox genes. Chromosomal mechanisms governing remote transcriptional Moreover, genomic analysis showed preferential enrichment regulation have latelyattracted increasing interest. In particular, near ectopic enhancers of binding sites for the PU.1 cofactor conserved noncoding elements/regions/sequences identified near NKX2-5 HMGA1, the knockdown of which also inhibited .We metazoan developmental genes show coordinate enhancer activity suggest that HMGA1 and PU.1 coregulate ectopic homeobox when tested in zebrafish embryos (11). Physical juxtaposition gene expression in t(5;14) T-cell acute lymphoblastic leukemia between regulators and their targets has now been shown (12). by interactions mediated at the nuclear matrix.Our data These interactions are believed to occur at chromatin loops tethered document homeobox gene dysregulation by a novel regulatory to nuclear scaffold matrix attachment regions (13–15). DNaseI ¶ BCL11B region at 3 - responsive to histone deacetylase inhibi- hypersensitive sites denote ‘‘active’’ chromatin and are believed to tion and highlight a novel class of potential therapeutic target heighten the accessibilityof nucleosome-depleted regulatoryDNA amid noncoding DNA. [Cancer Res 2007;67(4):1461–71] regions and facilitate loading of transcription factor binding sites. Chromatin relaxation is also associated with histone hyperacetylation (16) and is antagonized byhistone deacetylases.A global survey of regulatoryregions in CD4 + T cells uncovered >5,000 DNaseI Note: Supplementarydata for this article are available at Cancer Research Online hypersensitive sites mostly bracketing active genes (17). A notable (http://cancerres.aacrjournals.org/). exception included an orphan DNaseI hypersensitive site cluster Requests for reprints: Roderick A.F. MacLeod or Stefan Nagel, Deutsche Sammlung von Mikroorganismen und Zellkulturen, Department of Cell Cultures, lying inside the BCL11B gene desert amid a region subsequently Inhoffenstrasse 7B, 38124 Braunschweig, Germany. Phone: 49-531-261-6167; Fax: 49- shown to host an acetylation island in T cells (18). 531-261-6150; E-mail: [email protected] or [email protected]. ¶ I2007 American Association for Cancer Research. To investigate the role of 3 -BCL11B in leukemic cis-activation, doi:10.1158/0008-5472.CAN-06-2615 we adapted DNA inhibitorytreatments byusing matching www.aacrjournals.org 1461 Cancer Res2007; 67: (4). February 15, 2007

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Cancer Research oligonucleotides directed against specific noncoding sequences Chromatin immunoprecipitation assay. Chromatin immunoprecipi- based on previous work on the IgH enhancer (19, 20). Regulatory tation (ChIP) was done as recommended using the ChIP Assaykit (Upstate), 6 potential was further analyzed via nuclear matrix attachment in whereby10 cells were sonicated for 30 seconds, producing fragments A nuclear halo preparations, and bycharacterizing protein acetyla- ranging 100 to 1,000 bp, and precipitated with 10 g of the anti–acetyl- histone H3 antibody. For PCR analysis, 400 ng of DNA (ChIP or complete tion associated with both genes and noncoding regions. Binding genomic DNA) were used as templates (SupplementaryTable S1). site matrices for transcription factors PU.1 and HMGA1, identified PCR. Total RNA was isolated 6 or 24 h after trichostatin A treatment, in the promoter regions of TLX3 and near regulatoryDNaseI either 24 h after electroporation with the individual oligonucleotides, or 4 to hypersensitive sites, respectively, by using the TRANSFAC database 5 days after lentiviral transduction. RNA extraction, cDNA synthesis, and (21) were targeted for inhibition bysmall interfering RNA (siRNA) PCR were done as described previously(7). Real-time quantitative-PCR (RQ- treatment, and their effects on leukemic homeobox gene PCR) was done as described previously(28). Oligonucleotide sequences are expression in t(5;14) cells were measured. listed in SupplementaryTable S1. Our data support the existence of a remote regulatoryDNaseI Oligonucleotide inhibitory treatments. Transient inhibitorytreat- hypersensitive site cluster associated with leukemic homeobox ments were modified from that described byCutrona et al. (20), substituting gene activation in t(5;14) T-cell ALL. TRANSFAC data show this double-stranded 26-mer double-stranded oligonucleotide (DSO) phosphorothioate modified at positions 1-2/25-26 (purchased from MWG) for peptide putative regulatoryenhancer region to be enriched in HMGA1 nucleic acid oligonucleotides. In 10 of 16 cases, DSO sequences (Table 1) binding sites juxtaposed bypartner PU.1 sites at promoter regions were uniquelylocated at, or near ( V1 kbp), orphan T-cell DNaseI of both TLX3 and NKX2-5. We propose that leukemic activation hypersensitive sites (17), whereas the remainder lay either offset by f1 overlays HMGA1-PU.1 interactions mediated cytogenetically by to 10 kbp (DSO.12-14), or remote by>10 kbp (DSO.3/5), from DNaseI transcription factor binding site juxtaposition, highlighting the hypersensitive sites. Target sequences were chosen using TRANSFAC role of noncoding DNA in neoplasia and as a potential therapeutic databanks listing paired transcription factor binding sites displaced by target therein. <10 bp (the so-called composite elements; ref. 21). The selection process is shown in SupplementaryFig. S1 A for DSO.6. DSOs were produced by pooling equimolar amounts of complementarysingle-stranded oligonucleotides dissolved in water, heating to 95jC for 5 min and incubating at room Materialsand Methods temperature for 30 min. For electroporation, 2 Ag pEGFP-N1 reporter Cells and culture. Detailed descriptions of cell lines and culture (Clontech, Heidelberg, Germany) or DSO (20 Amol/L) were used to treat 106 methods are given by Drexler (22). Molecular karyotypes of CCRF-CEM, cells in culture medium using a EPI-2500 impulse generator (Fischer, HPB-ALL, and PEER, all established from pediatric T-cell ALL, have been Heidelberg, Germany) for 10 ms at 350 V. Cells were subsequently published elsewhere (6, 7). Antisense treatment was done as described transferred into 1 mL fresh culture medium and incubated overnight. previously(23). Briefly,5 Amol/L 21-mer oligonucleotide phosphorothioate, Chemically synthesized siRNA directed against HMGA1. Expression modified at positions 1-2/20-21 (MWG, Ebersberg, Germany) and targeting of HMGA1 was knocked down using validated siRNA obtained from Qiagen the start codon, was added to the culture medium at days 0 and 1. Cells (Hilden, Germany). As control, a synthetic siRNA directed against an were harvested on day3 for gene expression analysis. irrelevant sequence was used (BioSpring, Frankfurt, Germany). For treat- Cytogenetic analysis. Harvesting, slide preparation, trypsin G-banding, ments, 1 ALofa20Amol/L stock solution of siRNA was electroporated into and fluorescence in situ hybridization (FISH) are detailed elsewhere (24). PEER cells as described above. Bacterial artificial chromosome (BAC) clones were obtained from BAC/PAC Lentiviral-mediated RNA interference directed against PU.1. DNA Resources (Oakland, CA), RZPD (Berlin, Germany), The Sanger Institute oligonucleotides corresponding to the siRNA nucleotide targeting sequence (Cambridge, United Kingdom), and Drs. Olivier Bernard (Hopital Necker, 720 to 738 bp of the human PU.1 gene (Genbank accession no. NM_003120; Paris, France) and Reiner Siebert (Institut fu¨r Genetik, Kiel, Germany). ref. 29) were chemicallysynthesized with a 5 ¶ BglII overhang and a 3¶ SalI Fosmid clones were obtained from The Sanger Institute. Clone DNA was restriction site for cloning (Biospring, Frankfurt, Germany). The numbering prepared using Princeton Big-PAC kits (EMP, Berlin, Germany) and labeled of the first nucleotide of the short hairpin RNA (shRNA) refers to the ATG with Spectrum Red-dUTP or Spectrum Green-dUTP (Vysis, Bergisch start codon. The shRNA oligonucleotide sequences were as follows: Gladbach, Germany), or Cy3-dUTP [GE Healthcare (formerly Amersham), Munich, Germany], by nick translation (Invitrogen, Karlsruhe, Germany). FP-PU.1-720: 5¶-GATCCCCGAAGCTCACCTACCAGTTCTTCAAGAGA- FISH images were captured and analyzed using Smart Capture 2 software GAACTGGTAGGTGAGCTTCTTTTTTGGAAG-3¶ using a cooled Cohu charge coupled device camera (Applied Imaging, RP-PU.1-720: 5¶-TCGACTTCCAAAAAAGAAGCTCACCTACCAGTTC- Newcastle, United Kingdom) configured to an Axioskop 2 microscope TCTCTTGAAGAACTGGTAGGTGAGCTTCGGG-3¶. (Zeiss, Go¨ttingen, Germany). Preparation of DNA fibers. FISH was done on DNA fibers following The noncomplementary9-nt loop sequences are underlined, and the the method described byMejia et al. (25). Briefly,10 AL cell suspension sense oligonucleotide harbors six T’s as a polymerase III transcription adjusted to 1 Â 106 in PBS was placed immediatelyunder the frosted part termination signal. The oligonucleotides were annealed and inserted 3¶ of of a cleaned microscope slide preheated to 50jC and dried. The slide was the H1-RNA promoter into the BglII/SalI–digested pBlueScript-derived then placed near vertically, and DNA fibers were extruded by rinsing with pH1-plasmid to generate pH1-PU.1-720 as described byScherr et al. (30). 150 AL lysis buffer applied dropwise (0.05 mol/L NaOH in 28.6% ethanol); The sequence was confirmed bysequencing analysis.Two plasmids, pH1- fixed bya second rinse using 200 AL methanol; and thereafter dehydrated shRNA controls 1 and 2, directed against irrelevant sequences, were used as in ethanol (70%, 90%, and 100%, each for 2 min), air-dried, and processed controls (30). for FISH. pdc-SR was used to generate lentiviral transgenic plasmids containing Nuclear matrix attachment. Nuclear halos were prepared as previously H1-siRNA expression cassettes located in the U3 region of the 3¶ long described (26, 27). After FISH analysis, the nuclear locations of z30 nuclear terminal repeat. To generate the lentiviral plasmid pdcH1-siRNA-SR, the signals from each of 20 BACs were recorded and subsequentlyclassed as pH1-PU.1-720 plasmid was digested with SmaI and HincII, and the resulting follows: predominantlymatrix (score 2), perimatrix or both matrix and halo DNA fragments (360 nt) were blunt-end ligated into the SnaBI site of the (score 1), and predominantlyhalo (score 0). Nuclear matrix attachment pdc-SR to generate pdcH1-PU.1-SR plasmid. scores for each clone were taken as the means of these values. For histone VSV.G pseudotyped lentiviral particles were generated by calcium deacetylase inhibition, cells were treated with 10 Amol/L trichostatin A phosphate cotransfection of 293T cells, and viral supernatants were (Upstate, Lake Placid, NY) dissolved in ethanol. concentrated and titered as previouslydescribed, whereas recombinant

Cancer Res2007; 67: (4). February 15, 2007 1462 www.aacrjournals.org

Table 1. DSO sequences, efficacy, and distances from DNaseI hypersensitive sites and transcription factor binding matrices

DSOs Transcription factor binding site matrices

No. Efficacy Coordinates Sequences DNaseI-HS distances (bp) PAX4 SP1 HMGA1 PU.1

Single (17) Multiple

1 — 96507219 AGAGGGAAATGCCCCCACTTTTGGGA 410 2,333 2 — 96868259 CCAGGGTAAACCCACTTCCCTTTCTG 40 5,095 + 3 NKX2-5 97280412 TGGCACCATGTAAGGTCTCTAAGTGT 16,279 16,279 + 4 — 97480312 TAGAAGCCCTAGCCAGGACTAGCACA 1,970 1,970 5 — 97580716 TGATGTCACCAGCCCAAAGGCCAACA 155,100 20,259 + + 6 NKX2-5 97734681 GGAACTGGTGGTTAGTTTTGTGGTCA 130 56 + + 6-mut1 — — GGAACTGGTGGTTAGTTTAGTGGTCA Not applicable 6-mut-2 — — GGAACTGGTGGTTAGTTTAATGGTCA 6-mut-3 — — GGAACTGGTGGTTAGTTTAAAGGTCA 7 NKX2-5 97734779 GGCCCGAGCTCTCTCAGTGGCCACCT 30 154 + + + + 8 NKX2-5 97745902 CATGTGGTTACTTACATTCAAAGTAA 420 2,168 + + 9 NKX2-5 97746132 GTGGTCCGCCCCATCTCTCTTTTCCT 180 2,398 + + 10 NKX2-5 97864528 TCAGGAAGGGATCTGCACCGCCCTGG 320 4,558 + + + + 11 — 97918334 AACTGATTGGGATGGTACCCACATGG 360 1,662 12 NKX2-5 98130412 CAAGCTTGCTCTGGGTAAGTCAATGA 10,091 10,091 + + + + 13 NKX2-5 98180494 GGGACCTCGAATACCCAACAAACTTC 3,168 3,168 + + 14 NKX2-5 98330912 TATCAAAGGATGGAGTACTTGAGGAC 1,465 1,465 15 BCL11B 98670324 AGGGAACTCCCAAGCAGGAAATCAAA 1,050 1,427 16 BCL11B 98722026 TTCACTTGTTCCTTCCTCCCTTCCTT 540 1,098 +

NOTE: Data show sequences of DSOs used for DNA inhibitorytreatments and their efficacies in inhibiting NKX2-5 or BCL11B, or neither, in PEER cells; distances from nearest single and multiple DNaseI hypersensitive sites; and presence (+) of transcription factor binding sites predicted by TRANSFAC search (21). Coordinates are based on National Center for BiotechnologyInformation Build 36.1. Abbreviation: DNaseI-HS, DNaseI hypersensitive sites.

lentiviral supernatants were used to transduce 106 PEER cells with a 1127d7, significantlylacking the interstitial region covered by multiplicityof infection of f2 as described previously(31). clones 1140p17 and 1082a3 (Fig. 1A). The results of fiber FISH Mapping DNaseI hypersensitive sites. DNaseI hypersensitive sites were showed that the insertion point of 3¶-BCL11B material at 5q35 in mapped bymethods previouslydescribed (17, 32). Briefly,intact nuclei from DND-41 cells directlyjuxtaposes TLX3 in cells (Fig. 1A). This human CD4+ T cells were treated with small amounts of DNaseI (Roche, insertion is presumablyresponsible for activating TLX3 in DND-41 Indianapolis, IN), and blunt-ended fragments were ligated to a biotinylated ¶ linker. Approximately230,000 sequences were generated from the linked cells (33) byintroducing ectopic (3 -BCL11B) regulatoryenhancers ends using massivelyparallel signature sequencing. 5 Because this level of and thus resembles that reported in a T-cell ALL patient (34). After sequencing depth onlyidentified 20% of all DNaseI hypersensitive sites, we fine-mapping using fosmid clones (Fig. 1B), the 14q32 breakpoint have recentlysequenced more than 3 million tags using the 454 Life in PEER, first mapped using BAC clones (7), was revised Sciences GS-20 sequencer (Life Sciences).6 Multiple sequences from the (telomerically) to 97.81 Mbp, f900 kbp downstream of BCL11B, DNase-treated librarythat map within a 250 base window have been shown and that of CCRF-CEM to 98.43 Mbp in close agreement with the to be highly accurate at identifying valid DNaseI hypersensitive sites. previous BAC data (7), whereas 3¶-BCL11B itself is located at 98.71 Mbp (Fig. 1C). The configuration of ins(5;14) in DND-41 implies a Results target telomeric of BAC 74h1 within that part of the breakpoint Cytogenetic analysis for further delineation of BCL11B cluster region covered bythe insertion. Furthermore, remaining breakpoint cluster region. The consensus karyotype of the instances of t(5;14) cell lines CCRF-CEM, HPB-ALL, and PEER all DND-41 pediatric T-cell ALL cell line used to demarcate the involved microinsertions, ins(14;5), raising a question mark over BCL11B breakpoint cluster region is 86-98<4n>XXYY,À9,À15,+20. clinical instances where such complex rearrangements maybe FISH analysis using a 14q32 BAC contig revealed paired (i.e., underascertained (6, 7, 10). To reconcile insertion and breakpoint pretetraploidization) cryptic ins(5;14)(q35.2;q32.2q32.2) rearrange- data, the critical target(s) of t(5;14) rearrangements should lie ments undetectable byG-banding (Fig. 1 A, right inset), involving a within the genomic desert between clone 74h1 and BCL11B itself two-component microinsertion comprising f500 kbp of the 3¶- probablywithin the consensus region as defined bythe centro- BCL11B region at 14q32.2 extending from within clones 61o1 to meric portion of the DND-41 insertion. Analysis of nuclear matrix attachment. Nuclear matrix attachment, which reflects transcriptional recruitment of coding 5 http://research.nhgri.nih.gov/DNaseHS/May2005/ and regulatoryregions (13–15), possiblyrelated to nuclear 6 G.E. Crawford, unpublished observations. territories, was assayed cytologically from 3¶-BCL11B to VRK1 in www.aacrjournals.org 1463 Cancer Res2007; 67: (4). February 15, 2007

Figure 1. Cytogenetic analysis and the BCL11B breakpoint cluster region. A and B, cytogenetic analysis of DND-41 and PEER cells. FISHpreparations were counterstained with 4¶,6-diamidino-2-phenylindole (DAPI). A, cryptic formation of ins(5;14)(q35;q32) in DND-41. Right insets, normal G-banding, involving the discontinuous insertion of material from 3¶-BCL11B at 5q35. Note the absence of flanking clones 15e14 (green) and 2348n10 (purple) from the insertion (detailed in C). Inset (A, left, bottom), fiber FISHconfirming juxtaposition of TLX3 containing clone 117l6 (green) with clone 61o1 (red). B, fine mapping of 14q32 breakpoint in PEER cells to lie between fosmid clones G248P8-5772g10 (orange) and G248P8-9081b5 (red). Positions of these clones and the breakpoint in PEER (C). Note the insertion of material from fosmid G248P8-229B6 (green) at the 14q32 breakpoint. C, available cytogenetic data (breakpoints and insertions) on the far downstream region of BCL11B from t(5;14) T-cell ALL cell lines (large arrows) and patients (small arrows) from Su et al. (33). Coordinates are given in Mbp based on http://genome.ucsc.edu/index.html (National Center for Biotechnology Information Build 36.1). Positions of DSO.1-16 (downward pointing triangles) were shaded according to specificity in DNA inhibitory treatment–mediated down-regulation of NKX2-5 (black), BCL11B (gray), or neither (white). Bottom, DNaseI hypersensitive sites (DNAseI-HS; rockets) targeted by DSO.6-9 in the 97.72-.78 Mbp region, together with comparison data (HSS sites) described by Su et al. (33). Black rockets, activity, whether against NKX2-5 (DSO.6-9) or in heterologous promoter assays (HSS-3/4). Note the absence of promoter activity at HSS-1/2/5 (white rockets). NB, coincident active DNaseI hypersensitive site in both studies. Note also the TCRA enhancer–like motif (AP-1, LEF-1, ETS-1, AML1, and TATA box) at DSO.6/7. Acetylation islands (18) and duplicated noncoding sequences recently posited to host insulators (34) are also shown.

PEER cells (Fig. 2A and B). Illustrating the correlation between chromosome 17 displayed progressively looser nuclear matrix nuclear matrix attachment and gene activityin PEER cells, clones attachment scored 2, 1, and 0, respectively(Fig. 2 A, bottom left). In for NOTCH1 (obligate expression in T cells), PAX5 (silenced in T PEER cells, the highest nuclear matrix attachment was observed cells), and highlyrepetitive centromeric a-satellite DNA from close to BCL11B expression of which is T-cell specific, followed bya

Cancer Res2007; 67: (4). February 15, 2007 1464 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Activation of TLX3 and NKX2-5 in t(5;14)(q35;q32) central plateau (97.3–97.9 Mbp), straddling putative distal regula- 61o1 at 6,895 bp adjoining the breakpoint in PEER cells. toryregions DSO.3 and DSO.6-9 (Fig. 2 A and B). In SC-1 B-cell Expression of BC043585 and VRK1 were unaffected byanyDSO lymphoma cells in which BCL11B and BC043585 are silent, neither tested, including DSO.4 located within intron 4 of BC043585 the BCL11B peak nor the adjacent plateau were discernible (data (Fig. 3D), and DSO.1/2 at 3¶-VRK1. Of 10 DNaseI hypersensitive not shown)—findings consistent with tissue-specific variations in sites closelytargeted (sites <1 kbp), seven were inhibitory(five of nuclear matrix attachment (Fig. 2A, top). Trichostatin A treatment NKX2-5 and two of BCL11B), whereas two of the three noneffica- for 24 h reduced nuclear matrix attachment uniformlyacross the cious sites (DSO.1/2) laymost distant outside the breakpoint distal breakpoint cluster region, including both plateau and cluster region and close to VRK1. Neither non–DNaseI hypersen- neighboring telomeric valleyregions (Fig. 2 A, bottom right, B). sitive sites DSO.4/5 proved inhibitoryof NKX2-5 byRT-PCR or Interestingly, this valley shared coordinates with noninserted/ RQ-PCR. Although the immediate proximityof DSO to DNaseI deleted material in ins(5;14) in DND-41 cells (Fig. 1A and C) and hypersensitive sites did not dictate efficacy, the association of the in a t(5;14) patient reported bySu et al. (33). In overextracted latter with targeting specificitywas nonrandom: eight of nine preparations (8 min), differences in nuclear matrix attachment NKX2-5 inhibitorysites layinside the breakpoint cluster region, of between ins(14;5) responsible for leukemic NKX2-5 activation, which the four most efficacious (DSO.6-9) clustered immediately which exhibited high nuclear matrix attachment, forming a tight centromeric of the 14q32 breakpoint in PEER cells. On the other structure reminiscent of ‘‘active chromatin hubs’’ (35), were evident, hand, both BCL11B-inhibitorysites (DSO.15/16), although inside whereas nonleukemogenic der(5) and der(14) partner chromo- the breakpoint cluster region, laymost distant from the break- somes looped outside the nuclear matrix (Fig. 2A, bottom middle). point and nearest to BCL11B itself (Fig. 1C). One outlying Inhibitory treatments of candidate enhancer regions. To inhibitoryDSO site (DSO.3) located >500 kbp centromeric of the characterize the gene regulatorypotential of noncoding regions PEER breakpoint laywithin the high nuclear matrix attachment identified near DNaseI hypersensitive sites, these were targeted region (Fig. 2B). In summary, the most remarkable feature was the using DNA inhibitorytreatments based on reports describing the cluster of efficacious sites (DSO.6-9) located near the insertion use of short inhibitorypeptide nucleic acid probes matching IgH point of NKX2-5 in PEER cells. regulatorysequences, thus inhibiting oncogene expression driven Multiple DNaseI hypersensitive sites. Multiple DNaseI hyper- byIgH enhancers (19, 20). The efficiencyof DNA transfer by sensitive sites (clustered within 500 bp to minimize noise) are electroporation was checked bya reporter construct (pEGFP) in taken to serve as proxyenhancers (32). Multiple DNaseI several B-cell lines with t(8;14)(q24;32)/t(14;18)(q32;q21) and in hypersensitive sites coordinates derived by deep sequencing inside T-cell lines with t(5;14)(q35;q32), among which 380 and PEER cells, the BCL11B genomic desert are given in SupplementaryTable S3 respectively, displayed the best survival (both >80%) and transfec- and the intersite distances shown in SupplementaryFig. S2. DNaseI tion efficiencies (f60% and f80%, respectively). The 380 B-cell hypersensitive site densities fluctuate along 3¶-BCL11B, usually lymphoma cell line carries a t(14;18)(q32;q21), which juxtaposes peaking near gene loci, notably BCL11B itself, but onlyweaklynear BCL2 with the IgH-EA enhancer (22). Preliminaryinvestigations VRK1, if at all (SupplementaryFig. S2). Densitypeaks often showed that inhibition of BCL2 and NKX2-5 transcription by coincided with the largest DNaseI hypersensitive sites, although complementaryoligonucleotides was ( a) restricted to DSOs by these sometimes occurred within troughs. A similar pattern both reverse transcription-PCR (RT-PCR; SupplementaryFig. S1 B) occurred near acetylation islands. The second highest DNaseI and RQ-PCR (SupplementaryTable S2); ( b) abolished bypoint hypersensitive site peak coincided with the DSO.6-9 region, which mutations, confirming sequence specificity(SupplementaryTable is devoid of known genes. Interestingly, this peak included the S2); and (c) concentration dependent (SupplementaryFig. S1 C). largest concentration of multiple DNaseI hypersensitive sites DNA inhibitorytreatments directed against regulatoryregions was (SupplementaryFig. S2). validated against the IgH-EA enhancer sequence driving BCL2 Histone acetylation analyses. In higher organisms, gene transcription in 380 B-cell lymphoma cells (Supplementary Fig. activityis profoundlyassociated with histone acetylation in S1D), analogous to that used byCutrona et al. (20). promoter/enhancer regions (36). We did ChIP analysis to assay Enhancer inhibition was assayed by measuring BCL11B and histone-H3 acetylation of promoter/exon 1 regions of genes NKX2-5 expression in PEER cells in which these loci are juxtaposed BC043585, BCL11B, NKX2-5, and VRK1, together with DSO target via a f100 kbp genomic insertion of NKX2-5 f1.1 Mbp regions in both PEER (NKX2-5 expressing) and DND-41 cells (TLX3 downstream of BCL11B (Fig. 1B and C; ref. 7). Assayed by RT- expressing). We found both BCL11B and BC043585 to be highly PCR, DSO.3/6-10/12-14 inhibited expression of NKX2-5 (Fig. 3A), acetylated in PEER and moderately in DND-41 cells, whereas VRK1 whereas DSO.15/16 inhibited BCL11B only(Fig. 3 B). Control RT- was moderatelyacetylatedin both cell lines. In contrast, NKX2-5 PCR studies showed that inhibition depended on both double- and TLX3 displayed negligible promoter/exon 1 acetylation in strandedness and sequence fidelity(SupplementaryFig. S1 B). PEER and DND-41 cells (Fig. 4A). Among DNA inhibitorytreatment Quantification of NKX2-5 down-regulation byRQ-PCR yielded target regions, the DSO.6-9 cluster juxtaposing the NKX2-5 broadlysimilar results as well as confirming the dependence of insertion and TLX3 recipient breakpoints were stronglyacetylated inhibition on double strandedness and sequence fidelity(Fig. 3 C; in both PEER and DND-41, whereas more distant DSO targets were SupplementaryTable S2). Gene expression analysisof NKX2-5 and weaklyacetylated despite their efficacy(Fig. 4 B). Thus, the BCL11B distinguished three types of DSO target site in PEER cells: acetylation islands of T-cell ALL cells resemble those of Those that affected down-regulation of NKX2-5 (byDSO.3, DSO.6- nonmalignant T-cells (described in ref. 18) shown in Fig. 1C. 10, DSO.12-14), of BCL11B (byDSO.15/16), or of neither (byDSO.1/ Histone deacetylase inhibitors reactivate genes silenced by 2, DSO.4/5, DSO.11; Fig. 3C). Interestingly, principal components of histone deacetylation. Expression of MYB, NOTCH1, UBF, and the TCRA enhancer comprising binding sites for activator protein- VRK1 (ubiquitouslyexpressed in T cells) were unmodulated by 1, high-mobilitygroup domain containing protein LEF-1, ETS-1, trichostatin A treatment, whereas TLX3, BCL11B,andGATA3 AML1, and a TATA box were identified within the BAC sequence (subject to developmental regulation in Tcells) were down-regulated www.aacrjournals.org 1465 Cancer Res2007; 67: (4). February 15, 2007

Figure 2. Nuclear matrix attachment in PEER cells. The three-color FISHimages in (A) depict nuclear matrix preparations from PEER nuclei and hybridized with contrast-labeled BAC clones. Top panels, nuclear matrix attachment along the noncoding 3¶-BCL11B region: Note the low nuclear matrix attachment of clones 14l13 and 354a24 (yellow) derived from the region immediately centromeric of the BCL11B breakpoint cluster region, contrasting with clones covering the neighboring leukemic regulatory region, 61o1, 3181d5, and 74h1 (red), which exhibited high nuclear matrix attachment. On the other hand, clones from the intervening region closer to BCL11B, 1057p17, and 3104h21 (green) showed intermediate nuclear matrix attachment. In overextracted preparations (8 min), it was possible to distinguish both the correlation of nuclear matrix attachment with gene activity, as shown by the progressive margination of NOTCH1 (red), PAX5 (green), respectively, silent and active in PEER cells, and centromeric-17 (cen 17, yellow, bottom left); and between ins(14;5) responsible for leukemic NKX2-5 activation that exhibited high nuclear matrix attachment, forming a tight structure reminiscent of active chromatin hubs, whereas der(5) and der(14) ‘‘silent’’ partner chromosomes looped outward from the matrix (bottom middle). Trichostatin A (TSA) treatment–affected histone deacetylase inhibition, which induces silencing of NKX2-5 transcription in PEER cells, reduced matrix attachment, as shown by clones 15e14 (red) or 3181d5 (green; bottom right). Preparations were counterstained with DAPI and reversed to yield gray background staining. B, nuclear matrix attachment data (points) over the entire 2.3 Mbp region between VRK1 and BCL11B; bars, SE. Note central plateau of high nuclear matrix attachment including putative enhancer regions (DSO.3, DSO.6-9) and EST gene (BC043585), flanking lowered nuclear matrix attachment zones, and elevated nuclear matrix attachment close to transcribed genes VRK1 and BCL11B. Note the uniform reduction in nuclear matrix attachment along the region tested after trichostatin A treatment to inhibit histone deacetylase.

(Fig. 4C). Similarly, in PEER cells, trichostatin A treatment affected PU.1 and HMGA1 knockdown. Inspection of upstream down-regulation of NKX2-5 by62% (Fig. 4 D; SupplementaryTable promoter regions of both TLX3 and NKX2-5 revealed GAGGAA S2). The trichostatin A sensitivities of both TLX3 and NKX2-5 sequences corresponding to consensus PU.1 binding sites 100 and considered together with the absence of promoter/exon-1 acety- 366 bp upstream of their respective transcription start points. lation, revealed byChIP assay(Fig. 4 A), implies that leukemic TRANSFAC analysis done at DSO sites (F30 bp) to uncover activation of both homeobox genes byt(5;14) is subject to indirect transcription factor binding sites matrices showed the greatest regulation byhistone deacetylase. Obvious intermediates are degree of enrichment for HMGA1 (P < 0.005) at efficacious sites, transcription factors such as PU.1, down-regulated bytrichostatin whereas PU.1 sites were distributed randomly(Table 1). To A in multiple hematopoietic lineages (37), whereas both GATA3 investigate TLX3/NKX2-5 regulation bytranscription factors and PU.1 are repressed bydirect and promoter acetylation, subject to acetylation control, we treated DND-41 cells with respectively(38). antisense oligonucleotides directed against GATA3 and PU.1 and

Cancer Res2007; 67: (4). February 15, 2007 1466 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Activation of TLX3 and NKX2-5 in t(5;14)(q35;q32) subsequentlymonitored expression of TLX3. These experiments breakpoint cluster region that hosts 10 of 11 regulatorysites, showed that antisense PU.1, but not antisense GATA3 treatment, whereas onlyone of five outliers proved regulatory.Regulatory inhibited transcription of TLX3 in DND-41 cells (Fig. 5A). sequences in PEER cells sometimes layclose to DNaseI hypersen- Prompted bythese results, lentivirallymediated RNA interferences sitive sites, although absolute coincidence was not a requirement. directed against PU.1 or chemicallysynthesizedsiRNA against Sites were dedicated to either NKX2-5 or BCL11B, their orientation HMGA1 were done in PEER cells using RQ-PCR of NKX2-5 as end apparentlydictated byproximity.Maximal inhibition of NKX2-5 in point. These experiments showed that knockdown of PU.1 (by50%) PEER cells was affected bytreatments with DSO.6-9, which cluster or HMGA1 (by70%) effected down-regulation of NKX2-5 by80% within a f10 kbp region located f1 Mbp downstream of BCL11B and 70%, respectively(Fig. 5 B; SupplementaryTable S2). Taken and which, in PEER cells, immediatelyjuxtapose NKX2-5.In together, we concluded that PU.1 and HMGA1 are both necessary addition to DNaseI hypersensitive sites, this region displays a to sustain TLX3/NKX2-5 transcription in t(5;14) cells. diverse arrayof known and likely cis-regulatorystigmata, including elevated nuclear matrix attachment, hyperacetylation colocaliza- tion with an orphan T-cell acetylation island (18), abyssal cross- Discussion species homologyrelated to ‘‘five-wayregulatorypotential’’ (39), a We investigated leukemic activation of Nirenberg-Kim family proximate TCRA enhancer–like motif (AP-1, LEF-1, ETS-1, AML1, homeobox genes NKX2-5 and TLX3 involved in alternative t(5;14) and TATA box), and placement at the centromeric breakpoint translocations bycarryingout DNA inhibitorytreatments directed cluster region boundarynear the NKX2-5 insertion in PEER cells. against candidate enhancers within the 3¶-BCL11B breakpoint A recent studyhas also mapped DNaseI hypersensitivesites in cluster region, initiallyfocusing on T-cell single DNaseI hypersen- primaryT-cell ALL (33), a pair of which (HSS-3/4), respectively, sitive sites identified therein byglobal screening (17). It remains to match DSO.6/7 and DSO.8/9 described herein (Fig. 1C). Remark- be determined whether DNA inhibitorytreatment acts directlyon ably, of the five DNaseI hypersensitive sites tested by the authors, DNA substrates by, for example, blocking transcription factor heterologous promoter activitywas solelymanifested byHSS-3/4, access, or bysequestering activatorytranscription factors them- consistent with our finding that DSO.6-9 exerted the greatest selves. Inhibition was largelyrestricted to targets located within the effects on NKX2-5 expression. Taken together, these data provide

Figure 3. DNA inhibitory treatment by complementary DSOs. A and B, inhibition of NKX2-5 or BCL11B expression in PEER cells electroporated with DSO complementary to noncoding sequences at 3¶-BCL11B. Note the complete inhibition of NKX2-5 alone by DSO.3, DSO.6-10, and DSO.12/13 (A), and of BCL11B alone by DSO.15/16 and partially by DSO.14 (B). C, RQ-PCR data (detailed in Supplementary Table S1). Note peak inhibition by DSO.6-9 clustered near 14q32 breakpoint in PEER cells. Note also abrogation of DSO inhibition by use of either single-stranded (SSO.6a or SSO.6b) or mutated oligonucleotides. D, ineffectiveness of DNA inhibitory treatment using various DSO on transcription of BC043585 (top)orVRK1 (bottom), highlighting the specificity of the NKX2-5 inhibition. www.aacrjournals.org 1467 Cancer Res2007; 67: (4). February 15, 2007

Figure 4. Acetylation and leukemic expression of TLX3 and NKX2-5. Equal amounts of ChIP or genomic control DNA, respectively, was used for PCR analysis of promoter/exon 1 regions of genes VRK1, BC043585, NKX2-5, TLX3, and BCL11B (A), and noncoding regions hosting DSOs (B). Comparison of PCR product intensities reveals acetylation of VRK1, BC043585, and BCL11B, contrasting with NKX2-5 or TLX3, which remained unacetylated. Note acetylation of DSO.6-9, present within a known T-cell acetylation island (18) in both DND-41 and PEER, whereas other DSOs remained inconsistently acetylated, weakly acetylated, or nonacetylated. Effects of trichostatin A treatment on the expression of TLX3 in DND-41 assayed by RT-PCR (C) and of NKX2-5 expression in PEER assayed by RQ-PCR (D). TLX3, BCL11B, and GATA3 were down-regulated by trichostatin A treatment of DND-41 cells in contrast to unchanged levels of UBF, VRK1, MYB, and NOTCH1. Similarly, NKX2-5 exhibited trichostatin A–sensitive expression in PEER cells.

compelling evidence for a distal regulatoryenhancer complex f1 deprivation. Unlike the deeplyconserved region surrounding Mbp downstream of BCL11B—rivaling sonic hedgehog in remote- DSO.6-9, locallyexpressed transcripts, including BC043585, are ness from its putative coordinate target gene (40). primate specific, belying any regulatory connection with deeply Remoteness, together with the failure of DSO.6-9 to inhibit conserved sequences also represented in fugu and zebrafish. BCL11B expression, raise the question of whether BCL11B is indeed However, because 3¶-BCL11B breakpoints are variable, the role of the normal subordinate of this enhancer. The clinical homogeneity additional enhancers in the region remains to be addressed in of t(5;14) cases irrespective of breakpoint, together with the failure detail. Further experiments are planned to assess the regulatory of flanking DSO to inhibit BC043585, and indeed the absence of any repertoire of the DSO.6-9 and other candidate enhancers within convincing alternative, leaves BCL11B byfar the most likelytarget. 3¶-BCL11B and the potential of the former as a locus control Furthermore, in PEER cells that exceptionally(among T cells) region, inter alia, contributing to the maintenance of local express verylow levels of BCL11B (7), the distal regulatoryregion hyperacetylation. containing DSO.6-9 is cut off from BCL11B bythe intervening To direct spatiotemporal transcriptional regulation, develop- f100 kbp insertion containing NKX2-5, consistent with regulatory mental genes mayrequire boundaryelements (insulators). In this

Cancer Res2007; 67: (4). February 15, 2007 1468 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Activation of TLX3 and NKX2-5 in t(5;14)(q35;q32) regard, DSOs effectual against NKX2-5 in PEER cells are separated clustered inside an orphan acetylation island recently described in from DSO.15/16 inhibitoryof BCL11B onlybyflanking multiple primaryresting and activated T cells (18), and implythe retention noncoding elements/regions/sequences duplicated 3¶ of the and possible exploitation of an epigenetic alteration bymalignant homologous BCL11A. This type of homology hallmarks transcrip- T cells. Although PU.1 GAGGAA consensus binding site sequences tional insulators (35). The need to exclude transcriptional occur at 5¶-TLX3 (at À100 bp), NKX2-5 (at À366 bp), and BCL11B insulators would help explain the unprecedentedlywidespread (at À61 bp), the TRANSFAC database shows no preferential yet enigmatic formation of BCL11B rearrangements via micro- enrichment of PU.1 matrices near effectual DSO, in contrast to its insertions (6, 7, 34), or complex deletion-associated inversions (3), regulatoryconsort HMGA1 for which matrices occur at all DSO to the possible exclusion of conventional translocations, at least in inhibitoryof NKX2-5 with the sole omission of DSO.15. These cell lines where the most detailed studies have been documented findings serve to encourage wider investigation of the implied hitherto. relationship between the respective expression levels of PU.1/ In addition to confirming the existence in malignant T cells of an HMGA1 and Nirenberg-Kim familyhomeobox genes in cases of acetylation island recently described in their normal human T-cell ALL with t(5;14). PU.1 regulates IgH developmental expres- counterparts (18), but absent or displaced in mice (41), ChIP sion via enhancer binding (42) and interacts with HMGA1 by assays revealed that indigenous loci at 14q32.2 (VRK1, BC043585, increasing its transcriptional binding activity(43). HMGA1 also and BCL11B) were promoter/exon 1 acetylated. This finding modifies DNA structure at the IFNh enhancer to consolidate the contrasts with immigrant TLX3 and NKX2-5,whichwere, enhanceosome (44, 45), whereas HMGA1 acetylation by CBP effects nevertheless, both sensitive to histone deacetylase inhibition by its disruption (46). HMGA1 functions as a transcriptional coac- trichostatin A and knockdown of PU.1 or HMGA1. To resolve these tivator of the IgH-l enhancer in B cells through indirect association apparentlycontradictoryfindings, given the provision of PU.1 with DNA, being but looselybound to enhancer DNA (47). binding site matrices at both TLX3 and NKX2-5 promoters, we Our data showing that PU.1 and HMGA1 are both required to propose that the sensitivities of the latter t(5;14) targets to histone sustain leukemic homeobox gene transcription in T cells, and that deacetylase inhibition may be mediated indirectly by PU.1, the der(14) breakpoints near effectual DSO sites are flanked by PU.1 expression of which is trichostatin A sensitive (37). All four and HMGA1 matrices, suggest a possible nucleochromosomal acetylated DSO sites (DSO.6-9) target DNaseI hypersensitive sites mechanism underpinning the transient regulatoryassociation of

Figure 5. Effects of PU.1 and HMGA1 knockdown on TLX3 and NKX2-5. A, effects of antisense treatment of DND-41 cells using oligonucleotides directed against GATA3 or PU.1 compared with control oligonucleotide (AS-C). RT-PCR analysis shows down-regulation of TLX3 after as-PU.1 treatment of DND-41 cells, indicating the dependence of TLX3 on PU.1. B, quantitative RT-PCR of PU.1, HMGA1, and NKX2-5 after RNA interference–mediated inhibition. Knockdown of PU.1 was done by lentiviral-mediated transfer of shRNA vectors directed against PU.1. PU.1 expression was reduced by 40% and correlated with 30% reduction of NKX2-5. Inhibition of HMGA1 was done by electroporation of synthetic siRNA. Compared with controls, expression of both HMGA1 and NKX2-5 was reduced by 70%, respectively. Data indicate regulation of NKX2-5 expression by PU.1 and HMGA1.

www.aacrjournals.org 1469 Cancer Res2007; 67: (4). February 15, 2007

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Cancer Research these transcription factors as invoked in B cells. A recent studyhas enhanceosome might interact with distal DNaseI hypersensitive shown enrichment of binding sites for ETS familytranscription sites convened at the nuclear matrix embodying ‘‘active chromatin factors (including PU.1) near DNaseI hypersensitive sites implicat- hubs.’’ ed in long-distance regulation of tissue-specific genes in endothe- To summarize, we developed a method (DNA inhibitory lial cells (48). The selectivelytight nuclear matrix attachment treatment) for the targeting of specific noncoding DNA sequences displayed at 14q32, reminiscent of active chromosomal hubs, and that permitted identification of remote leukemic enhancers at its alleviation bytrichostatin A treatments inhibitoryof TLX3/ 3¶-BCL11B in t(5;14) cells. Interestingly, the most effectual NKX2-5, support a model wherebyPU.1/HMGAl interactions are enhancers laywithin a DNaseI hypersensitivesite cluster located mediated nucleochromosomally, possibly via tethering to SATB1, f0.9 to 1.0 Mbp downstream of BCL11B and displayed multiple which is known to act as a substrate of HMGA1 (49). regulatorystigmata. Ectopic NKX2-5 and TLX3 expression were In T cells, SATB1 occupies chromosomal scaffolds to facilitate inhibited byknockdown of either HMGA1 or PU.1, highlighting tissue-specific nuclear matrix attachment and histone modifica- their complicityin leukemic homeobox deregulation in t(5;14). tion, therebylinking chromatin modification and looping to As well as providing novel paradigms for oncogene dysregulation, transcription (50). Our findings that the centromeric boundaryof t(5;14) T-cell ALL is a major clinical entityin which the iden- the 3¶-BCL11B breakpoint cluster region coincides with a steep tification of potential therapeutic targets amid noncoding DNA decline in nuclear matrix attachment, while the DSO.6-9 region sensitive to DNA inhibitorytreatments and histone deacetylase occupies an elevated nuclear matrix attachment plateau, also inhibition merits wider investigation. support a role for nuclear geographyin leukemogenic activation. The reasons underlying deep cross-species conservation, as present Acknowledgments around DSO.6-9, remain obscure but mayreflect combined Received 7/14/2006; revised 11/2/2006; accepted 12/8/2006. structural and functional constraints, such as the need to effect Grant support: Jose´Carreras Leukemia Research Fund grants (R.A.F. MacLeod transient interactions between appropriate transcription factors. In and S. Nagel). The costs of publication of this article were defrayed in part by the payment of page the present case, this might involve interactions between HMGA1 charges. This article must therefore be herebymarked advertisement in accordance bound near enhancers, and PU.1 bound to promoters of homeobox with 18 U.S.C. Section 1734 solelyto indicate this fact. genes subject to a varietyof acetylationcontrols: both direct, by We thank Sandra Goetze and Silke Winkelman for help in establishing nuclear matrix assays, Drs. Olivier Bernard and Reiner Siebert for supplying clones, Greg Elgar opening DNaseI hypersensitive sites, and indirect, mediated via for access to unpublished data, SriLaxmi Kalavalapalli and Sebastian Becker for help HMGA1/PU.1. According to such a model, a putative t(5;14) with data analysis, and Professor Ju¨rgen Bode for criticallyreading the manuscript.

References 10. Su X, Drabkin H, Clappier E, et al. Transforming 22. Drexler HG. The leukemia-lymphoma cell line facts potential of the T-cell acute lymphoblastic leukemia- book. San Diego: Academic Press; 2001. 1. Wakabayashi Y, Watanabe H, Inoue J, et al. Bcl11b is associated homeobox genes HOXA13, TLX1, and TLX3. 23. Nagel S, Scherr M, Quentmeier H, et al. HLXB9 h required for differentiation and survival of a T lym- Genes Chromosomes Cancer 2006;45:846–55. activates IL6 in Hodgkin lymphoma cell lines and is phocytes. Nat Immunol 2003;4:533–9. 11. Woolfe A, Goodson M, Goode DK, et al. Highly regulated byPI3K signalling involving E2F3. Leukemia 2. Wakabayashi Y, Inoue J, Takahashi Y, et al. conserved non-coding sequences are associated with 2005;19:841–6. Homozygous deletions and point mutations of the vertebrate development. PloS 2004;Biol 3:e7. 24. MacLeod RAF, Drexler HG. Cytogenetic character- Rit1/Bcl11b gene in g-rayinduced mouse thymic 12. Carter D, Chakalova L, Osborne CS, Dai YF, Fraser P. ization of tumor cell lines. Methods Mol Med 2004;88: lymphomas. Biochem Biophys Res Commun 2003;301: Long-range chromatin regulatoryinteractions in vivo. 57–76. 598–603. Nat Genet 2002;32:623–6. 25. Mejia JE, Willmott A, LevyE, Earnshaw WC, Larin Z. 3. Przybylski GK, Dik WA, Wanzeck J, et al. Disruption of 13. Bode J, Schlake T, Rios-Ramirez M, et al. Scaffold/ Functional complementation of a genetic deficiency the BCL11B gene through inv(14)(q11.2q32.31) results in matrix-attached regions: structural properties creating with human artificial chromosomes. Am J Hum Genet the expression of BCL11B-TRDC fusion transcripts and transcriptionallyactive loci. Int Rev Cytol1995;162A: 2001;69:315–26. is associated with the absence of wild-type BCL11B 389–454. 26. Vidakovic M, Koester M, Goetze S, et al. Co- transcripts in T-ALL. Leukemia 2005;19:201–8. 14. Bode J, Goetze S, Heng H, Krawetz SA, Benham C. localization of PARP-1 and lamin B in the nuclear 4. Bernard OA, Busson-LeConiat M, Ballerini P, et al. A From DNA structure to gene expression: mediators of architecture: a halo-fluorescence- and confocal-micros- new recurrent and specific cryptic translocation, nuclear compartmentalization and dynamics. Chromo- copystudy.J Cell Biochem 2005;96:555–68. t(5;14)(q35;q32), is associated with expression of the some Res 2003;11:435–45. 27. Goetze S, Huesemann Y, Baer A, Bode J. Hox11L2 gene in T acute lymphoblastic leukemia. 15. VassetzkyY, Hair A, Mechali M. Rearrangement of Functional characterization of transgene integration Leukemia 2001;15:1495–504. chromatin domains during development in Xenopus. patterns byhalo fluorescence in situ hybridization: 5. Satterwhite E, Sonoki T, Willis TG, et al. The BCL11 Genes Dev 2000;14:1541–52. electroporation versus retroviral infection. Biochemis- gene family: involvement of BCL11A in lymphoid 16. Taunton J, Hassig CA, Schreiber SL. A mammalian try2003;42:7035–43. malignancies. Blood 2001;98:3413–20. histone deacetylase related to the yeast transcriptional 28. Eder M, Battmer K, Kafert S, et al. Monitoring of 6. MacLeod RAF, Nagel S, Kaufmann M, Janssen JW, regulator Rpd3p. Science 1996;272:408–11. BCR-ABL expression using real-time RT-PCR in CML Drexler HG. Activation of HOX11L2 byjuxtaposition 17. Crawford GE, Holt IE, Mullikin JC, et al. Identifying after bone marrow or peripheral blood stem cell with 3¶-BCL11B in an acute lymphoblastic leukemia cell gene regulatoryelements bygenome-wide recoveryof transplantation. Leukemia 1999;13:1383–9. line (HPB-ALL) with t(5;14)(q35;q32.2). Genes Chromo- DNase hypersensitive sites. Proc Natl Acad Sci U S A 29. Luo Q, Ma X, Wahl SM, Bieker JJ, CrossleyM, somes Cancer 2003;37:84–91. 2004;101:992–7. Montaner LJ. Activation and repression of interleukin-12 7. Nagel S, Kaufmann M, Drexler HG, MacLeod RAF. The 18. Roh TY, Cuddapah S, Zhao K. Active chromatin p40 transcription byerythroidKruppel-like factor in cardiac homeobox gene NKX2-5 is deregulated by domains are defined byacetylationislands revealed by macrophages. J Biol Chem 2004;279:18451–6. juxtaposition with BCL11B in pediatric T-ALL cell lines genome-wide mapping. Genes Dev 2005;19:542–52. 30. ScherrM,BattmerK,GanserA,EderM. via a novel t(5;14)(q35.1;q32.2). Cancer Res 2003;63: 19. Cutrona G, Carpaneto EM, Ulivi M, et al. Effects in Modulation of gene expression bylentiviral-mediated 5329–34. live cells of a c-myc anti-gene PNA linked to a nuclear deliveryof small interfering RNA. Cell Cycle2003;22: 8. Przybylski GK, Dik WA, Grabarczyk P, et al. The effect localization signal. Nat Biotechnol 2000;18:300–3. 251–7. of a novel recombination between the homeobox gene 20. Cutrona G, Carpaneto EM, Ponzanelli A, et al. 31. Scherr M, Battmer K, Blomer U, et al. Gene transfer NKX2-5 and the TRD locus in T-cell acute lymphoblastic Inhibition of the translocated c-myc in Burkitt’s into peripheral blood-derived CD34+ NOD/SCID-repo- leukemia on activation of the NKX2-5 gene. Haemato- lymphoma bya PNA complementaryto the E A pulating cells. Blood 2002;99:709–12. logica 2006;91:317–21. enhancer. Cancer Res 2003;63:6144–8. 32. Crawford GE, Holt IE, Whittle J, et al. Genome-wide 9. Cismasiu VB, Ghanta S, Duque J, et al. BCL11B 21. Matys V, Kel-Margoulis O, Fricke E, et al. TRANS- mapping of DNase hypersensitive sites using massively participates in the activation of interleukin-2 gene FAC(r) and its module TRANSCompel(r): transcriptional parallel signature sequencing (MPSS). Genome Res 2006; expression in CD4+ T lymphocytes. Blood 2006;108: gene regulation in eukaryotes. Nucleic Acids Res 2006; 16:123–31. 2695–702. 34:D108–10. 33. Su XY, Della-Valle V, Andre-Schmutz I, et al.

Cancer Res2007; 67: (4). February 15, 2007 1470 www.aacrjournals.org

HOX11L2/TLX3 is transcriptionallyactivated through ing cis-regulatorymodules in aligned mammalian 45. Yie J, Merika M, Munshi N, Chen G, Thanos D. The T-cell regulatoryelements downstream of BCL11B as a genome sequences. Genome Res 2005;15:1051–60. role of HMG I(Y) in the assemblyand function of the result of the t(5;14)(q35;q32). Blood 2006;108:4189–202. 40. Lettice LA, HeaneySJ, Purdie LA, et al. A long-range IFN-h enhanceosome. EMBO J 1999;18:3074–89. 34. Su XY, Busson M, Della Valle V, et al. Various types of Shh enhancer regulates expression in the developing 46. Munshi N, Merika M, Yie J, Senger K, Chen G, Thanos rearrangements target TLX3 locus in T-cell acute limb and fin and is associated with preaxial polydactyly. D. Acetylation of HMG I(Y) by CBP turns off IFNh lymphoblastic leukemia. Genes Chromosomes Cancer Hum Mol Genet 2003;12:1725–35. expression bydisrupting the enhanceosome. Mol Cell 2004;41:243–9. 41. Togashi T, Obata M, Aoyagi Y, Kominami R, Mishima 1998;2:457–67. 35. Vavouri T, McEwen GK, Woolfe A, Gilks WR, Elgar G. Y. Two distinct methods analyzing chromatin structure 47. McCarthyKM, McDevit D, Andreucci A, Reeves R, Defining a genomic radius for long-range enhancer using centrifugation and antibodies to modified histone Nikolajczyk BS. HMGA1 co-activates transcription in B action: duplicated conserved non-coding elements hold H3: both provide similar chromatin states of the Rit1/ cells through indirect association with DNA. J Biol Chem the key. Trends Genet 2006;22:5–10. Bcl11b gene. Biochem Biophys Res Commun 2004;313: 2003;278:42106–14. 36. de Laat W, Grosveld F. Spatial organization of gene 489–95. 48. Bernat JA, Crawford GE, Ogurtsov AY, Collins FS, expression: the active chromatin hub. Chromosome Res 42. Nelsen B, Tian G, Erman B, et al. Regulation of Ginsburg D, Kondrashov AS. Distant conserved sequen- 2003;11:447–59. lymphoid-specific immunoglobulin A heavychain gene ces flanking endothelial-specific promoters contain 37. Laribee RN, Klemsz MJ. Loss of PU.1 expression enhancer byETS-domain proteins. Science 1993;261:82–6. tissue-specific DNase-hypersensitive sites and overrep- following inhibition of histone deacetylases. J Immunol 43. Lewis RT, Andreucci A, Nikolajczyk BS. PU.1-mediat- resented motifs. Hum Mol Genet 2006;15:105. 2001;167:5160–6. ed transcription is enhanced byHMG-I(Y)-dependent 49. Yasui D, Miyano M, Cai S, Varga-Weisz P, Kohwi- 38. Huo X, Zhang J. Important roles of reversible structural mechanisms. J Biol Chem 2001;276:9550–7. Shigematsu T. SATB1 targets chromatin remodelling to acetylation in the function of hematopoietic transcrip- 44. Falvo JV, Thanos D, Maniatis T. Reversal of intrinsic regulate genes over long distances. Nature 2002;419:641–5. tion factors. J Cell Mol Med 2005;9:103–12. DNA bends in the IFNh gene enhancer bytranscription 50. Cai S, Han HJ, Kohwi-Shigematsu T. Tissue-specific 39. King DC, Taylor J, Elnitski L, et al. Evaluation of factors and the architectural protein HMG I(Y). Cell nuclear architecture and gene expression regulated by regulatorypotential and conservation scores for detect- 1995;83:1101–11. SATB1. Nat Genet 2003;34:42–51.

www.aacrjournals.org 1471 Cancer Res2007; 67: (4). February 15, 2007

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Activation of TLX3 and NKX2-5 in t(5;14)(q35;q32) T-Cell Acute Lymphoblastic Leukemia by Remote 3 ′-BCL11B Enhancers and Coregulation by PU.1 and HMGA1

Stefan Nagel, Michaela Scherr, Alexander Kel, et al.

Cancer Res 2007;67:1461-1471.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/67/4/1461

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2007/02/13/67.4.1461.DC1

Cited articles This article cites 47 articles, 18 of which you can access for free at: http://cancerres.aacrjournals.org/content/67/4/1461.full#ref-list-1

Citing articles This article has been cited by 11 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/67/4/1461.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/67/4/1461. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.