Oncogene (2004) 23, 4255–4262 & 2004 Nature Publishing Group All rights reserved 0950-9232/04 $30.00 www.nature.com/onc The 8;21 translocation in leukemogenesis Luke F Peterson1 and Dong-Er Zhang*,1 1Department of Molecular and Experimental Medicine, The Scripps Research Institute, Mail Drop: MEM-L51, La Jolla, CA 92037, USA A common chromosomal translocation in acute myeloid of 13 exons distributed over at least 87 kb (Figure 1) leukemia (AML) involves the AML1 (acute myeloid and also has alternative spliced forms (Wolford and leukemia 1, also called RUNX1, core binding factor Prochazka, 1998). The breakpoints within the AML1 protein (CBFa), and PEBP2aB) gene on chromosome 21 locus are clustered in the classically recognized intron 5, and the ETO (eight-twenty one, also called MTG8) gene with three breakpoint cluster regions (BCR) (Zhang on chromosome 8. This translocation generates an AML1- et al., 2002). On the other hand, the breakpoints are ETO fusion protein. t(8;21) is associated with 12% of de clustered in two introns of ETO (Tighe et al., 1993; novo AML cases and up to 40% in the AML subtype M2 Tighe and Calabi, 1995), with one BCR in intron 1a of the French–American-British classification. Further- (Tighe and Calabi, 1995) and three BCRs in intron 1b more, it is also reported in a small portion of M0, M1, and (Xiao et al., 2001; Zhang et al., 2002). These breakpoints M4 AML samples. Despite numerous studies on the create the same chimeric transcript of AML1-ETO due function of AML1-ETO, the precise mechanism by which to the lack of any splice acceptor site in exon 1b of ETO the fusion protein is involved in leukemia development (Kozu et al., 1993; de Greef et al., 1995; Xiao et al., is still not fully understood. In this review, we will discuss 2001). Interestingly, the clusters of AML1 and ETO structural aspects of the fusion protein and the accumu- breakpoints are associated with topoisomerase II DNA lated knowledge from in vitro analyses on AML1-ETO cleavage and DNase I hypersensitive sites (Zhang et al., functions, and outline putative mechanisms of its leuke- 2002). Such phenomenon is also present in translocation mogenic potential. breakpoints of other genes of leukemia-associated Oncogene (2004) 23, 4255–4262. doi:10.1038/sj.onc.1207727 translocations (Felix, 1998). In addition, complex translocations involving chromosomes 8 and 21 also Keywords: translocation; myeloid leukemia; RUNX; express the same AML1-ETO fusion gene, suggesting AML; ETO; MTG8 that the breakpoint regions are the same (Kozu et al., 1993; de Greef et al., 1995). Overall, no translocation has been identified between AML1 and ETO that occurs outside these intron boundaries that produce the AML1-ETO fusion protein. Structure of acute myeloid leukemia 1 (AML1)-eight- AML1 is a member of the RUNX (PEBP2a, twenty one (ETO) core binding factor protein (CBFa), AML) family, which includes AML1 (RUNX1, PEBP2aB, CBFa2), The 8;21 translocation with breaks at 8q22 and 21q22.3 AML2 (RUNX3, PEBP2aC, CBFa3), and AML3 was first reported by Dr Janet Rowley in 1973 during (RUNX2, PEBP2aA, CBFa1). All RUNX proteins the analysis of a leukemia patient sample (Rowley, contain a runt homology DNA binding domain, 1973). Two genes (AML1 and ETO) located at these which is highly homologous to the Drosophila Runt breakpoints were finally identified in the early 1990s by protein. Runt is involved in segmentation and sexual several groups (Gao et al., 1991; Miyoshi et al., 1991; determination (Kania et al., 1990; Duffy and Gergen, Erickson et al., 1992, 1994; Nisson et al., 1992; Shimizu 1991). Besides the DNA binding runt homology et al., 1992). As shown in Figure 1, the classical domain, AML1 also contains a transactivation domain recognized organization of the AML1 gene on chromo- (Meyers et al., 1995; Kurokawa et al., 1996), nuclear some 21 consists of nine exons and produces at least matrix attachment signal (NMTS) (Zeng et al., 1998), three proteins (Miyoshi et al., 1995). Further fine and two inhibitory domains (Imai et al., 1998; Kanno mapping revealed that AML1 gene locus spans 260 kb et al., 1998; Levanon et al., 1998). Furthermore, AML1 and there are several additional potential exons (Leva- forms a heterodimer with the CBFb facilitating efficient non et al., 2001). The expression of AML1 is regulated DNA binding (Ogawa et al., 1993). In the 8;21 by two independent promoters (Ghozi et al., 1996) and translocation, AML1 contributes the N-terminal region, alternative splicing (Miyoshi et al., 1995; Levanon et al., including runt homology domain, thus making AML1- 1996, 2001). The ETO gene on chromosome 8 consists ETO able to bind AML1 target gene promoters (Figure 2). The ETO protein belongs to the ETO family, which *Correspondence: D-E Zhang; E-mail: [email protected] includes ETO (MTG8), ETO2 (MTG16, MTG8-related The 8;21 translocation in leukemogenesis LF Peterson and D-E Zhang 4256 Figure 2 AML1-ETO-interacting proteins. AML1-ETO contains the N-terminal sequences of AML1, including the RHD. Sub- sequent a.a.’s are of ETO, containing the four NHR1–4. Indicated as well are the regions containing the NMTS of ETO (broken line arrows). Known RHD and ETO/AML1-ETO-interacting proteins (or family of proteins) are shown. The dimerization domain of AML1-ETO and ETO family members is located in NHR2 Figure 1 Genomic structure of t(8;21). Chromosome 8 containing the ETO gene is made up of 13 exons spanning approximately 87 kb, which can give four alternative splice forms and is regulated by two promoters. Chromosome 21 contains the AML1 gene with Protein–protein interactions nine exons that give various alternative splice forms and is regulated by two promoters and spans 260 kb. The breakpoint cluster areas are denoted by the crossing lines between ETO and Since AML1-ETO contains the N-terminal portion of AML1. Owing to the absence of a splice acceptor in exon 1b of AML1 including the intact runt homology domain, it ETO, the mRNA of the fusion transcripts does not include this may interact with other transcription regulators known exon. White boxes and black boxes indicate translated and to bind to this region of AML1 (Figure 2). The first and untranslated exon sequences, respectively. Underlined numbers in the best known AML1 binding protein is its heterodimer the AML1-ETO mRNA denote exons contributed by the ETO gene partner CBFb (also called PEBP2b) that promotes efficient DNA binding of AML1 and is essential for the full activity of AML1 transcriptional activation (Ogawa et al., 1993; Wang et al., 1993, 1996a, b; Kanno 2 (MTGR2)), and MTGR1 (Calabi and Cilli, 1998; et al., 1998). Most protein–protein interactions invol- Kitabayashi et al., 1998). All three members of this ving the runt homology domain (RHD) of AML1 were family have four evolutionarily conserved Nervy homo- identified due to the proximity of the AML1 binding site logy domains termed Nervy homology regions (NHR) to the DNA binding sites of these transcription factors, 1–4 based on their homology to the Drosophila including Ets-1, LEF-1, C/EBPa, PU.1, MEF, Pax5, protein Nervy, which is involved in neuronal develop- and GATA1 (Zhang et al., 1996; Petrovick et al., 1998; ment during segmentation of Drosophila embryos Kim et al., 1999; Libermann et al., 1999; Mao et al., (Feinstein et al., 1995). NHR1 has homology with the 1999; Goetz et al., 2000; Gu et al., 2000; Elagib et al., Drosophila TATA-box-associated factor 110 (TAF110) 2003; Li et al., 2004). For example, AML1 binding sites and other TAFs (Erickson et al., 1994). The NHR2 overlap with Ets-1 sites and are close to LEF-1 sites in domain contains a hydrophobic amino-acid (a.a.) enhancers of several T-cell receptors (Hsiang et al., 1993; heptad repeat and is required for homo- and hetero- Wotton et al., 1994; Giese et al., 1995; Mayall et al., dimerization between the ETO family members 1997). AML1 associates with these adjacent transcrip- (Gelmetti et al., 1998; Kitabayashi et al., 1998; Zhang tion factors and enhances their interaction to DNA (Sun et al., 2001). NHR3 contains a predicted coiled-coil et al., 1995). Furthermore, C/EBPa and PU.1 bind to structure (Minucci et al., 2000). NHR4 is a myeloid- the adjacent sites of AML1 consensus sequence in Nervy-DEAF1 homology domain with two predicted human M-CSF receptor promoter. They interact with zinc-finger motifs (CxxCCxxC and a CxxCHxxC) the RHD of AML1 with differing potentials in the (Erickson et al., 1994; Gross and McGinnis, 1996). synergistic activation of the M-CSF receptor (Zhang These zinc-finger motifs of ETO are involved in protein– et al., 1996; Petrovick et al., 1998). protein interaction, but not protein–DNA interaction. The first ETO-interacting protein identified was its AML1-ETO contains almost the entire ETO protein family member MTGR1 (Kitabayashi et al., 1998). (Figure 2). MTGR1 shares 61% identity with ETO and also Oncogene The 8;21 translocation in leukemogenesis LF Peterson and D-E Zhang 4257 contains four Nervy homology domains. The NHR2 Accumulating data indicate that NMTS of various domain plays a critical role in the interaction between transcription factors may influence the specific localiza- ETO and MTGR1. Another ETO family member, tion and local concentration of transcription factors and ETO2, also interacts with ETO via NHR2 domain provide additional specificities for these factors in gene (Davis et al., 1999). Using ETO as bait or silencing expression (Davie, 1997). In addition, many of the mediator for retinoic acid receptor and thyroid hormone corepressor proteins that interact with AML1-ETO are receptor (SMRT) as bait in two independent yeast two- associated with such nuclear matrix foci (Davie et al., hybrid protein interaction studies and based on the 1999; Wood et al., 2000).