Oncogene (2002) 21, 3232 ± 3240 ã 2002 Nature Publishing Group All rights reserved 0950 ± 9232/02 $25.00 www.nature.com/onc The leukemia-associated transcription repressor AML1/MDS1/EVI1 requires CtBP to induce abnormal growth and dierentiation of murine hematopoietic cells Vitalyi Senyuk1, Soumen Chakraborty1, Fady M Mikhail1,2, Rui Zhao1, Yiqing Chi1 and Giuseppina Nucifora*,1 1Department of Pathology and The Cancer Center, University of Illinois at Chicago, Chicago, Illinois, USA; 2Clinical Pathology Department, Faculty of Medicine, University of Alexandria, Alexandria, Egypt The leukemia-associated fusion gene AML1/MDS1/ (ME) genes, is a product of the (3;21)(q26;q22) EVI1 (AME) encodes a chimeric transcription factor translocation that is associated with de novo and that results from the (3;21)(q26;q22) translocation. This therapy-related myelodysplastic syndrome/acute mye- translocation is observed in patients with therapy-related loid leukemia (t-MDS/AML) patients, and to a lesser myelodysplastic syndrome (MDS), with chronic myelo- extent with chronic myelogenous leukemia (CML) genous leukemia during the blast crisis (CML-BC), and patients during the blast crisis of their disease (Rubin with de novo or therapy-related acute myeloid leukemia et al., 1990; Nucifora et al., 1993). The AML1 gene (AML). AME is obtained by in-frame fusion of the (also known as CBFA2, PEBP2 or RUNX1) encodes AML1 and MDS1/EVI1 genes. We have previously the DNA-binding subunit of the core-binding tran- shown that AME is a transcriptional repressor that scription factor (CBF). AML1 interacts with the other induces leukemia in mice. In order to elucidate the role subunit of CBF, CBFb, which has no DNA-binding of AME in leukemic transformation, we investigated the domain. AML1 consists of a N-terminal DNA-binding interaction of AME with the transcription co-regulator domain with partial homology to the product of CtBP1 and with members of the histone deacetylase Drosophila segmentation gene runt, and a C-terminal (HDAC) family. In this report, we show that AME activation domain that interacts with transcription co- physically interacts in vivo with CtBP1 and HDAC1 and regulators (Ogawa et al., 1993a, b). AML1 is essential that these co-repressors require distinct regions of AME for de®nitive murine hematopoiesis (Wang et al., 1996; for interaction. By using reporter gene assays, we Okuda et al., 1996) and is involved in chromosomal demonstrate that AME represses gene transcription by abnormalities associated with human leukemias (Zent CtBP1-dependent and CtBP1-independent mechanisms. et al., 1997; Roulston et al., 1998) including T-cell Finally, we show that the interaction between AME and leukemia (Mikhail et al., 2002). ME is a DNA-binding CtBP1 is biologically important and is necessary for zinc-®nger transcription factor and a longer form of growth upregulation and abnormal dierentiation of the the leukemia-associated protein EVI1 (Fears et al., murine hematopoietic precursor cell line 32Dc13 and of 1996; Soderholm et al., 1997). ME contains a murine bone marrow progenitors. conserved N-terminal region, called PR domain, two Oncogene (2002) 21, 3232 ± 3240. DOI: 10.1038/sj/ sets of DNA-binding Cys2His2-type zinc-®nger do- onc/1205436 mains, a proline-rich central domain, and an acidic C- terminal domain. The evolutionarily conserved PR Keywords: AML1; MDS1; EVI1; t(3;21); CtBP1; domain was ®rst identi®ed in the RIZ1 protein and in HDAC1 the PRDI-BF1/BLIMP1 transcriptional repressor (Huang, 1994; Fears et al., 1996). ME is expressed in several tissues but is not detected in normal hemato- poietic cells (Fears et al., 1996). Both AML1 and ME Introduction are transcriptional activators (Zhang et al., 1996; Soderholm et al., 1997), whereas AME is a transcrip- The AML1/MDS1/EVI1 (AME) fusion gene, obtained tional repressor (Zent et al., 1996). AME consists of by in frame fusion of the AML1 and MDS1/EVI1 the DNA-binding domain of AML1 fused to almost the entire ME including the PR domain (Nucifora et al., 1994). AME potentially possesses the ability to bind both the AML1- and ME-target promoters. *Correspondence: G Nucifora, Department of Pathology and The Recently, it was shown that AME induces leukemia Cancer Center, Molecular Biology Research Building, M/C 737, in mice transplanted with syngeneic bone marrow cells University of Illinois at Chicago, 900 South Ashland Avenue, that express AME (Cuenco et al., 2000). The pathways Chicago, Illinois, IL 60607, USA; E-mail: [email protected] Received 21 November 2001; revised 15 February 2002; accepted through which AME induces cell transformation and 21 February 2002 leukemia are still unknown. It is possible that AME AME de-regulates growth and differentiation via CtBP V Senyuk et al 3233 could inappropriately repress the expression of genes to a much tighter chromatin structure which is that are primary targets of AML1 and ME by presumably inaccessible to transcription factors (Gro- inappropriate recruitment of transcription repressors zinger et al., 1999). Two distinct classes of HDACs at the promoter site, as it was proposed for AML1/ have been identi®ed in mammalian cells. HDAC1, ETO and PML/RARa (Lin and Evans, 2000; Minucci HDAC2, HDAC3, and HDAC8 belong to class I and et al., 2000). It was shown recently that EVI1 interacts are homologous to yeast Rpd3 and Xenopus HDACm with the co-repressor CtBP (Izutsu et al., 2001; Palmer proteins, with a single deacetylase domain at the N- et al., 2001; Chakraborty et al., 2001). CtBP was ®rst terminus. HDAC4, HDAC5, HDAC6, HDAC7, and identi®ed as a protein that binds to the C-terminal HDAC9 are members of class II and are similar to region of the adenovirus E1A protein (Schaeper et al., yeast Hda1 with one or two catalytic domains in their 1995) and it is a well-characterized transcriptional co- C-terminal regions (Bertos et al., 2001). In order to repressor (Turner and Crossley, 2001). CtBP also binds elucidate the potential role of HDACs in AME- to BKLF, FOG, AREB6 (Turner and Crossley, 1998), mediated transcription repression, we tested the Knirps, Kruppel, Snail (Nibu et al., 1998; Keller et al., interaction between HA-AME and Flag-tagged 2000), ZEB (Postigo and Dean, 1999), polycomb HDAC1, 74, 75, and 76. The results are shown in proteins (Sewalt et al., 1999) and other transcription Figure 3. We found that AME strongly interacts with co-regulators with the consensus motif PXDLX to HDAC1 (Figure 3, lane 2), but has a weak or no enhance transcription repression. In this study, we interaction with HDAC4, HDAC5, and HDAC6 investigate the interaction of AME with the transcrip- (Figure 3, lanes 3 to 5). tion co-regulator CtBP1 and with members of the To identify the region of AME that is necessary for histone deacetylase (HDAC) family. We show that the interaction with HDAC1, we used the AME AME interacts with CtBP1 and HDAC1 through deletion mutants shown in Figure 1 in co-IP assays. dierent domains and represses gene transcription by The results show that whereas the full length AME and CtBP1-dependent and CtBP1-independent mechanisms. the deletion mutants AME-1184, AME-994, and AME- More importantly, we also show that the interaction 704 were co-immunoprecipitated by anti-Flag anti- between AME and CtBP1 is functionally important bodies that recognize the Flag-tagged HDAC1 (Figure and is required for maximum cell growth upregulation 4, lanes 2 to 5), the shortest deletion mutant AME-374 by AME and for abnormal dierentiation and was not co-immunoprecipitated by anti-Flag antibodies immortalization of murine bone marrow progenitors. (Figure 4, lane 11). These results suggest that the interaction between AME and HDAC1 requires the aa region 374 to 704. This region includes the entire proximal zinc-®nger domain and a small part of the Results PR domain from ME. It was reported that HDAC1 interacts with CtBP1 in vivo (Sundqvist et al., 1998). To AME interacts with the co-repressor CtBP1 con®rm that AME does not require CtBP1 for CtBP1 binds to the amino acid (aa) pentapeptide motif HDAC1 interaction, we tested the ability of AME C- PXDLX. The AME fusion protein has two potential mutant to co-IP with HDAC1. The results, shown in CtBP1 binding sites: PFDLT (aa 1036 to 1040) and Figure 5, indicate that full length AME (lane 2 and 4) PLDLS (aa 1067 to 1071). To determine whether AME and AME-C (lanes 3 and 5) interact with HDAC1, and associates with CtBP1 at these consensus motifs we therefore further con®rm that the interaction between tested the full length AME and the AME point AME and HDAC1 does not require CtBP1. However, mutants A, B, and C (Figure 1) in co-immunoprecipi- our data do not indicate whether the interaction is tation (co-IP) assays. The A-mutant has a DL to AS direct. substitution in the proximal site (PFDLT to PFAST); the B-mutant has the same substitution in the distal AME co-localizes in the nucleus with CtBP1 or HDAC1 site (PLDLS to PLASS); the C-mutant has both sites mutated. As shown in Figure 2, the wild type AME The interaction between AME and either CtBP1 or (lane 2) and the A-mutant (lane 3) interact with CtBP1. HDAC1 was also con®rmed by confocal microscopy In contrast, the B- and C-mutants, which have the DL analysis of 293 cells that were co-transfected with HA- to AS substitution in the distal site (PLDLS, aa 1067 AME and either Flag-CtBP1 or Flag-HDAC1. As to 1071), were not co-immunoprecipitated with CtBP1 shown in Figure 6, AME (Figure 6, top panels in red (Figure 2, lanes 4 and 5). These results indicate that the color) co-localizes with either CtBP1 or HDAC1 distal PLDLS site has a very strong anity for CtBP1 (Figure 6, middle panels in green color) in the cell and that the mutations DL to AS in this site abrogate nucleus.