Oncogene
SEARCH     advanced search my account e-alerts subscribe register
Journal home
Advance online publication
Current issue
Archive
Press releases
For authors
For referees
Contact editorial office
About the journal
For librarians
Subscribe
Advertising
naturereprints
Contact NPG
Customer services
Site features
NPG Subject areas
Access material from all our publications in your subject area:
Biotechnology Biotechnology
Cancer Cancer
Chemistry Chemistry
Dentistry Dentistry
Development Development
Drug Discovery Drug Discovery
Earth Sciences Earth Sciences
Evolution & Ecology Evolution & Ecology
Genetics Genetics
Immunology Immunology
Materials Materials Science
Medical Research Medical Research
Microbiology Microbiology
Molecular Cell Biology Molecular Cell Biology
Neuroscience Neuroscience
Pharmacology Pharmacology
Physics Physics
Browse all publications
 
10 September 2001, Volume 20, Number 40, Pages 5695-5707
Table of contents    Previous  Article  Next   [PDF]
Molecular mechanisms of leukemogenesis mediated by MLL fusion proteins
Paul M Ayton and Michael L Cleary

Department of Pathology, Stanford University Medical Center, 300 Pasteur Drive, Stanford, California, CA 94305, USA

Correspondence to: P M Ayton, Department of Pathology, Stanford University Medical Center, 300 Pasteur Drive, Stanford, California, CA 94305, USA

Abstract

The MLL (Mixed Lineage Leukemia) gene is a common target for chromosomal translocations associated with human acute leukemias. These translocations result in a gain of MLL function by generating novel chimeric proteins containing the amino-terminus of MLL fused in-frame with one of 30 distinct partner proteins. Structure/function studies using an in vitro myeloid progenitor immortalization assay have revealed that at least four nuclear partner proteins contribute transcriptional effector properties to MLL to produce a range of chimeric transcription factors with leukemogenic potential. Mouse models suggest that expression of an MLL fusion protein is necessary but not sufficient for leukemogenesis. Interestingly, whilst all MLL fusion proteins tested so far phenocopy each other with respect to in vitro immortalization, the latency period required for the onset of acute leukemia in vivo is variable and partner protein dependent. We discuss potential mechanisms that may account for the ability of distinct MLL fusion proteins to promote short or long latency leukemogenesis. Oncogene (2001) 20, 5695-5707.

Keywords

MLL; leukemia

Genes that encode transcription factors are commonly targeted by chromosomal translocations in acute leukemias. One consequence is the in-frame joining of two genes to generate unique fusion proteins of novel function (Cleary, 1991; Rabbitts, 1994). The MLL family of oncogenic fusion proteins represents one such class of deregulated transcriptional regulators (Ayton and Cleary, 2001). An unusual feature of MLL fusion proteins is the large number and diversity of heterologous proteins that are fused with MLL. These are observed in leukemias of either lymphoid or myeloid lineage derivation and are particularly prevalent in infant leukemias and treatment-related secondary leukemias. In general, they are associated with an extremely poor prognosis (Mitelman et al., 1994; Secker-Walker, 1998). In this article, we review recent advances derived from in vitro and in vivo mouse models that are beginning to yield insights regarding the molecular events that regulate initiation and progression of acute myeloid leukemias (AML) associated with MLL fusion proteins.

Role of MLL in mammalian development

MLL function is required for mid-gestational development of the mouse (Yu et al., 1995; Yagi et al., 1998). Embryonic lethality at E10.5 is associated with multiple patterning defects in neural crest-derived structures of the branchial arches, cranial nerves and ganglia (Yu et al., 1995, 1998). Although correct initiation of expression of several examined Hox genes (a7, c8 and c9) is observed in homozygous embryos at E9.5, subsequent maintenance of Hox gene expression is lost (Yu et al., 1998). These studies suggest that MLL functions as a mammalian counterpart of Drosophila trx, which positively maintains the expression of multiple Hox genes during development.

MLL mutant mice also exhibit hematopoietic abnormalities (Hess et al., 1997). Yolk sac progenitors from MLL nullizygous mutants at E10.5 generated definitive CFU-M and CFU-GEMM colonies in hematopoietic assays, however the colonies were consistently smaller, fewer in number, and exhibited a slower growth rate compared to colonies generated from wild type littermates. A second mutant MLL allele resulted in embryonic lethality later in gestation (E12.5-13.5) with mutant embryos exhibiting edema and bleeding, similar to the phenotype described for loss of AML-1 function (Okuda et al., 1996; Yagi et al., 1998). With this particular MLL mutant, the size and kinetics of myeloid colony production were significantly reduced and associated with decreased expression of a number of Hox genes (Hox-a7, a9, a10, c4) in the MLL mutant fetal liver (Yagi et al., 1998).

Despite myeloid progenitor defects associated with both MLL mutant alleles, all mature cell types were generated, suggesting that MLL function is not absolutely required for terminal myeloid differentiation, but may influence the proliferation and/or survival of multipotent progenitors. Loss of MLL function was not associated with defective erythroid colony formation, consistent with the lack of MLL expression in the erythroid compartment (Hess et al., 1997). Normal MLL function within the hematopoietic system bears some similarity to that previously reported for GATA-2 regulation of multi-potential progenitor development (Tsai et al., 1994). Interestingly, GATA-2 expression can be detected in MLL homozygote mutant embryos, suggesting that MLL lies either downstream or on a parallel pathway to GATA-2 (Hess et al., 1997). The early embryonic lethality of homozygote MLL mutants has thus far prevented analysis of the role of MLL function during lymphoid development.

Domain structure of MLL fusion proteins

Heterologous protein fusions constitute the most prevalent class of MLL mutants associated with acute leukemia. Cytogenetic analysis of complex three-way MLL translocations identified the absolute conservation of the derivative chromosome 11 product (Rowley, 1992), which encodes the amino-terminal 1400 residues of MLL fused to one of several possible partner proteins (Figure 1). Fusion proteins consistently retain the AT hook, SNL 1 and 2 motifs, and the CxxC domain (see below) of MLL. Conversely, the PHD, transactivation and SET domains of MLL are consistently replaced by partner protein sequences. All MLL fusion proteins maintain a productive open reading frame, suggesting that this feature is strongly selected for and therefore essential for the transformation properties of the respective fusion proteins.

Two other rarer classes of leukemogenic MLL mutants have also been described (Figure 1). These include: (1) various partial amino-terminal duplications (PTD) of MLL; and (2) deletions of MLL exon 8 that encodes critical cysteine residues within the first PHD finger of MLL (Schichman et al., 1994, Caligiuri et al., 1996; Lochner et al., 1996; reviewed in Ayton and Cleary, 2001). These alternative classes of MLL mutants presumably promote leukemogenesis without the concomitant need for fusion of MLL to a partner protein, suggesting that alteration of MLL function alone, in some circumstances, is sufficient to promote leukemogenesis. Similar to MLL fusion proteins, PTD mutants maintain the MLL open reading frame, suggesting that both classes of mutants may act as dominant alleles, although it is currently unclear whether they exert positive or negative effects on MLL function.

The amino terminus of MLL

The amino terminal portion of MLL that is consistently retained in all MLL fusion proteins contains several conserved motifs that appear essential for function (Figure 1). Three short motifs, termed AT hooks (Tkachuk et al., 1992), are thought to mediate specific binding to the minor groove of AT-rich DNA, as originally described for the transcription factor HMGI/Y (Reeves and Nissen, 1990). The solution structure of HMGI/Y in complex with an in vivo binding site in the beta interferon enhancer has been recently solved (Huth et al., 1997). Each AT hook motif consists of a central RGR motif that directly binds the DNA minor groove. The central core of the motif is flanked at its amino terminus by a KR dipeptide and an array of nine conserved residues at its carboxy terminus. Both flanking modules mediate extensive hydrophobic and polar contacts with the target site. Interestingly, the three AT hooks present in the amino terminus of MLL lack the carboxy-terminal residues that mediate extensive phosphodiester backbone contacts, suggesting that MLL may possess decreased affinity for the minor groove relative to HMGI/Y (Huth et al., 1997). MLL may therefore utilize its minor groove binding activity to indirectly stabilize protein-DNA interactions by inducing conformational changes, such as bends in DNA of the target binding site, which may in turn facilitate the binding of specific transacting factors that regulate target gene transcription. Alternatively, upon binding to the minor grove, MLL may mediate protein-protein interactions that allow distinct factors to efficiently interact with each other and the basal transcription machinery.

The AT hook motifs of HMGI/Y are subject to cell cycle dependent post-translational modification. In vivo, AT hook phosphorylation occurs during mitosis on the same residue that is phosphorylated in vitro by cdc2 kinase (Nissen et al., 1991; Reeves et al., 1991). AT hook phosphorylation was associated with a significant reduction in DNA binding activity. A consensus cdc2 kinase site is also present in the most amino-terminal of the three AT hooks of MLL, suggesting that the ability of this motif to bind to target sites may also may be regulated by cell cycle dependent phosphorylation events.

MLL contains a second cysteine-rich region, termed the CxxC motif, which exhibits sequence homology with two proteins implicated in the epigenetic regulation of transcription via methylation. These proteins include mammalian DNA methytransferase (DMT) and methyl binding domain protein 1 (MBD1) (Domer et al., 1993; Cross et al., 1997; Hendrich and Bird, 1998). The latter proteins contain a conserved methyl-binding domain, distinct from CxxC, that recognizes and binds methylated target promoters. Although the functional significance of the CxxC motif is currently unknown, it may constitute a protein-protein interaction interface that contributes to the transcriptional effector properties of these proteins. In support of this possibility, transcriptional repression activity has been localized to the CxxC motif of MLL (Zeleznik-Le et al., 1994; Prasad et al., 1995).

Positioned between the AT hook motifs and the CxxC domain lie two short conserved sequences in the amino portion of MLL that direct its localization to discrete sub-nuclear domains (Yano et al., 1997). These two regions, termed SNL1 and SNL2, mediate speckled sub-nuclear localization. Both regions are highly conserved with Drosophila trx and represent the only structural homology between trx and MLL that is retained in MLL fusion proteins. This suggests that the ability of MLL/trx to localize to discrete sub-nuclear sites has been conserved through evolution and may therefore play a critical role in regulating MLL/trx function. Furthermore, these results suggest that at least some trx downstream target genes/signaling pathways are also conserved in mammals and the normal regulation of these targets may well be disrupted during MLL-mediated leukemogenesis.

The MLL partner proteins

Almost 30 partner genes that participate with MLL in reciprocal chromosomal translocations associated with acute leukemia have been reported. All of the MLL partner proteins that account for the vast majority of ALL/AML have been identified, but several novel MLL partner proteins have been recently isolated from rare, infrequent translocations. These include AF3p21, AF5q31, AF9q34, AF15q14, LAF4, GAS-7, GRAF, GMPS, GEPHYRIN, and LARG (Taki et al., 1999; von Bergh et al., 2000; Borkhardt et al., 2000; Hayette et al., 2000; Kourlas et al., 2000; Megonigal et al., 2000; Pegram et al., 2000; Sano et al., 2000; Kuwada et al., 2001). Although the clinical impact of these rare fusions may well be minimal, their further characterization will undoubtedly promote further mechanistic insights into MLL-mediated leukemogenesis.

Most MLL partner genes are widely expressed in a variety of adult tissues, including hematopoietic cells. A number of the predicted proteins exhibit considerable homology within functional domains and therefore can be regarded as members of novel protein families. MLL partner proteins appear to fall into two functional categories: signaling molecules that normally localize to the cytoplasm/cell junctions or nuclear factors implicated in various aspects of transcriptional regulation. A list of the first 17 MLL partner proteins to be identified, including their putative functional domains and close structural homologues has been recently reviewed (Ayton and Cleary, 2001).

The transcriptional effector functions of MLL fusion partners are essential for leukemogenesis

Several laboratories have embarked upon structure/function studies to delineate the critical regions of MLL partner proteins necessary for leukemogenesis. An in vitro myeloid transformation assay (MTA) has been particularly useful for this purpose (Lavau et al., 1997). A cartoon summarizing the MTA is depicted in Figure 2. This assay delivers expression of MLL fusion proteins via retroviral transduction to primary murine hematopoietic progenitors, which form a variety of myeloid colonies when cultured in semi-solid methylcellulose medium supplemented with a cocktail of myeloid cytokines. Immortalized myeloid progenitors maintain their self-renewal and clonogenic potential, repeatedly forming subsequent immature colonies after serial replating. In contrast, normal progenitors display a loss of clonogenic potential after a single round of replating due to induction of terminal differentiation.

MLL-ENL: Structure/function analysis of MLL-ENL established that specific functional domains of both MLL and ENL were necessary for myeloid transformation (Slany et al., 1998). Deletion analysis suggested that amino-terminal regions of MLL including the first two AT hooks, as well as a region including the CxxC motif were required for immortalization. For ENL, most of the protein was dispensable except for a carboxy-terminal region of 84 residues (Figure 3), predicted to form two amphipathic alpha helices, that was necessary and sufficient for immortalization by MLL-ENL. Most importantly, the short carboxy-terminal region of ENL also mediated transactivation when fused to the Gal4 DNA binding domain. Interestingly, the transactivation domain of ENL is conserved with Anc1, a stoichiometric component of two yeast basal transcription complexes (TFIIF and TFIID) and the SWI/SNF chromatin remodeling complex (Cairns et al., 1996). These findings suggest a direct role for ENL in transcription, although ENL has not yet been reported in the comparable mammalian transcriptional complexes. Nevertheless, the ENL moiety clearly donates transcriptional effector properties to the MLL-ENL fusion protein. Therefore, the mechanism by which MLL-ENL mediates transformation could involve a dominant gain of MLL function whereby MLL-ENL acts to constitutively maintain MLL target gene expression.

MLL-ELL: Three members of the ELL family have been identified and shown to share a modular domain structure (Figure 3) (Thirman et al., 1994; Shilatifard et al., 1997a; Miller et al., 2000). ELL was isolated biochemically due to its ability to act as a transcriptional elongation factor for RNA polymerase II (Shilatifard et al., 1996). The central R2 domain of ELL suppresses transient pausing of RNA polymerase II along the DNA template, thereby increasing the catalytic rate of transcription elongation. In addition, further analysis determined a role for the most amino-terminal 50 residues of ELL, termed the R1 domain, in inhibiting promoter-specific transcription by RNA polymerase II, although the R1 domain is not retained by MLL-ELL fusion proteins (Shilatifard et al., 1997b). ELL also possesses two conserved carboxy-terminal domains of unknown function: a short lysine-rich R3 domain and the R4 domain. Although fusion of MLL to ELL does not impair the transcriptional elongation activity of the R2 domain, structure/function studies show that the latter is not required for myeloid immortalization by MLL-ELL (DiMartino et al., 2000). Rather, the previously uncharacterized carboxy-terminal R4 domain is necessary and sufficient for myeloid immortalization by MLL-ELL (Figure 3) (DiMartino et al., 2000). Similar to ENL, the R4 domain of ELL exhibits transactivation potential in transient transcriptional assays. In a parallel study, retroviral transduction of MLL-ELL into hematopoietic stem cells, followed by transplantation into lethally irradiated recipients, led to the development of AML after an extended latency period (Lavau et al., 2000b). The R4 domain of ELL interacts with a novel protein named EAF1 (ELL associated factor 1), which itself bears similarity to the central transactivation domain of proteins in the AF4 family (Lavau et al., 2000c). Thus, EAF1 may facilitate transactivation via the ELL R4 domain. In support of this possibility, direct fusion of EAF1 to MLL is sufficient to immortalize myeloid progenitors in vitro and generate AML in vivo (Lavau et al., 2000c).

MLL-CBP: CBP is another MLL fusion partner with known roles in transcriptional regulation. Regions of CBP necessary and sufficient for both in vitro immortalization and in vivo development of AML were localized to those containing both the bromo-domain and the neighboring histone acetyl transferase (HAT) domain (Figure 3) (Lavau et al., 2000a). When fused to MLL, the bromo-domain or HAT domain alone is unable to immortalize myeloid progenitors. The HAT domain of CBP has previously been shown to acetylate the lysine tails on all four core histones, thereby inducing chromatin accessibility (Bannister and Kouzarides, 1996). Interestingly, the CBP HAT domain alone retains promoter specific transactivation activity when fused to the Gal4 DNA binding domain, yet this region of CBP is not sufficient to induce immortalization in vitro (Martinez-Balbas et al., 1998). This is the first example of a nuclear MLL partner protein where transactivation properties do not directly correlate with immortalization. What does the bromodomain contribute to immortalization by MLL-CBP? The solution structure of the bromo-domain of the CBP-associated protein P/CAF shows that highly conserved residues within the bromo-domain mediate binding to acetylated lysine residues of histones H3 and H4 (Dhalluin et al., 1999). Residues within the bromo-domain of yeast Gcn5 that mediate acetyl-lysine binding are not required for histone acetylation, but instead play a vital role in subsequent nucleosome remodeling via recruitment of the SWI/SNF complex (Syntichaki et al., 2000). Whilst the mechanism by which MLL-CBP immortalizes myeloid progenitors is presently unclear, the available data suggest that it most likely involves epigenetic deregulation of MLL function, perhaps by indirectly promoting chromatin accessibility within the regulatory regions of MLL target genes that subsequently allow entry of the basal transcriptional machinery.

MLL-AF10: AF10 belongs to a small, evolutionary conserved family of proteins (Chaplin et al., 1995a). In Drosophila, genetic analysis suggests that the AF10 orthologue Dalf, is required for the maintenance of Even-skipped expression in the embryonic CNS (Bahri et al., 2001). Dalf exhibits neural-specific function but, similar to many human MLL partner proteins, is ubiquitously expressed. All members of the AF10 family possess an amino terminal PHD finger and a central motif with features of a leucine zipper (LZ). Most MLL-AF10 fusions occur near the amino-terminus of AF10, disrupting its single PHD domain (Chaplin et al., 1995b). However, one patient was identified with an MLL fusion to the central region of AF10, immediately upstream of the conserved LZ motif. Structure/function studies revealed that a region spanning the LZ is necessary for MLL-AF10 mediated immortalization in vitro (DiMartino et al., submitted). However, the minimal LZ region alone is not sufficient to mediate immortalization by MLL-AF10. Rather, the LZ motif in combination with a flanking conserved octapeptide sequence (EQLLERQW) is necessary and sufficient for myeloid immortalization (Figure 3). Furthermore, when fused to MLL, both motifs in tandem are able to activate an MLL responsive Hoxa7 reporter gene, a property that neither individual motif possesses. Transactivation in this assay was weaker than that for MLL-ENL or MLL-ELL. Transplantation of immortalized MLL-AF10 cells into SCID recipient mice resulted in the development of rapid AML within 60 days (DiMartino et al., submitted).

Taken together, the structure/function studies strongly suggest that a unifying mechanism for the activation of MLL by nuclear fusion partners is the gain of transcriptional effector properties. For two partner proteins, ENL and ELL, the ability to immortalize myeloid progenitors correlates well with their ability to transactivate reporter genes in transient transcriptional assays. For MLL-CBP, the HAT domain possesses transactivation potential when fused to a heterologous protein such as Gal4, but this region alone is not sufficient for immortalization by MLL-CBP, which also requires its bromodomain. Thus, the mechanism by which MLL-CBP immortalizes myeloid progenitors may involve epigenetic regulation of chromatin accessibility surrounding regulatory regions within MLL target genes, as opposed to classical transcriptional deregulation. We have also detected transactivation potential (albeit weaker than ENL or ELL) for ML-AF10 that is dependent upon regions of AF10 required for immortalization. The critical role of the AF10 LZ motif suggests that this fusion may promote immortalization through homo and/or hetero-dimeric interactions. In support of this possibility, provocative evidence has recently been presented suggesting that inducible homo-dimerization of the amino terminus of MLL is sufficient to convert MLL to a strong transcriptional activator in reporter assays (Galoian et al., 2000). Future studies will determine whether MLL dimerization promotes leukemogenesis in vivo.

Thus, conversion of MLL to a constitutive transcriptional effector may be sufficient to provide a gain of function that promotes leukemogenesis in vivo. However, distinct MLL fusion proteins may influence the rate of target gene transcription at different levels during the multi-step process of recruiting a basal transcriptional complex. These events may affect the earliest stages of transcriptional regulation through epigenetic disruption of normal nucleosome architecture within the regulatory regions of target genes, thereby facilitating continued access by other transcription factors which recruit the basal machinery. Alternatively, MLL fusion proteins may interact indirectly with components of the transcription machinery through heterologous interactions with partner proteins or directly through multimerization. Whilst it seems unlikely that the many cytoplasmic MLL partner proteins possess intrinsic transactivation properties themselves, it remains an attractive hypothesis that these partner proteins may also mediate MLL fusion protein immortalization through homodimeric interactions.

The AF4 protein family and infant pro-B ALL

MLL-AF4, which is associated with infant pro-B ALL, is the most prevalent of the numerous MLL fusion proteins (Domer et al., 1993; Morrissey et al., 1993; Nakamura et al., 1993; Johansson et al., 1998). Since the initial identification of AF4 as the major MLL partner gene associated with pro-B ALL, three other mammalian AF4 homologues have been cloned. The predicted proteins, FMR2, LAF4 and AF5q31, exhibit significant homology to multiple regions of the predicted AF4 protein (Gecz et al., 1996; Gu et al., 1996; Ma and Staudt, 1996; Nilson et al., 1997; Taki et al., 1999). Alignment of the AF4 family members suggests a multiple domain structure, as depicted in Figure 4. Functional studies of AF4 and LAF4 suggest that they act as transcriptional activators when fused to the GAL4 DNA binding domain (Prasad et al., 1995; Ma and Staudt, 1996; Morrissey et al., 1997). Interestingly, potent transcriptional activation was localized to a central conserved domain, which is consistently retained in all MLL-AF4 fusion proteins reported to date.

Both AF5q31 and LAF4 have been independently identified as MLL partner genes in individual cases of infant pro-B ALL (Taki et al., 1999; von Bergh et al., 2001). Thus, fusion of MLL to AF4 family members is consistently associated with the development of pro-B ALL with short latency. In contrast, congenital mutations in the FMR2 gene are associated with mental retardation (Gecz et al., 1996; Gu et al., 1996). These mutations involve considerable di-nucleotide expansions that result in loss of FMR2 function. It is currently unclear whether an MLL-FMR2 fusion would immortalize pro-B cell progenitors or whether FMR2 exhibits neural-specific functions.

Recent gene targeting studies have revealed that, despite its ubiquitous expression, AF4 function is required for normal hematopoiesis (Isnard et al., 2000). Approximately 20% of AF4-deficient mice die within 10 days of birth. Whilst the cause of death is currently unclear, these mutants exhibited lymphoid defects. However, the majority of AF4-deficient mice survive with normal lymphopoiesis. The low penetrant phenotype of the null AF4 allele may be due to functional compensation by other AF4 family members such as AF5q31 and/or LAF4. In the bone marrow, loss of AF4 function did not disrupt progenitor B cell development, however the transition from pre-B cell to the newly generated mature B cells was significantly reduced. Furthermore, AF4 mutant mice exhibited defective thymocyte development from immature double negative (CD4-CD8-) to the intermediate double positive (CD4+CD8+) population. These studies suggest that AF4 may positively regulate the expression of critical target genes that are required for the correct execution of early lymphopoiesis in both the B and T cell lineages. Interestingly, despite thymocyte defects associated with the loss of AF4 function, the MLL-AF4 fusion protein is rarely associated with T-ALL.

The pair-rule gene lilliputian (lilli) represents the sole orthologue of the AF4 family in the Drosophila genome (Tang et al., 2001; Wittwer et al., 2001). The predicted lilli protein retains the conserved central and carboxy-terminal domains present in all mammalian AF4 family members, with the highest degree of identity in the carboxy terminal domain (Wittwer et al., 2001). Significantly, an amino-terminal region that is conserved in the mammalian AF4 family proteins is absent. Instead, the extended amino terminus of lilli possesses multiple glutamine residues and a single copy of the DNA binding AT hook motif.

In the fly, genetic analysis suggests that lilli functions in two distinct pathways, either as a putative downstream nuclear target of the MAPK pathway or a potential upstream regulator of decapentaplegic (DPP), the orthologue of mammalian TGF-beta (Su et al., 2001). Mutant lilli alleles include a variety of truncating point mutations distributed along the length of the protein from a region upstream of the central conserved transactivation domain (Wittwer et al., 2001). All mutations delete the conserved carboxy-terminal domain. Moreover, two point mutations representing single amino-acid substitutions (Y1459F and E1461V) were recovered within the carboxy terminal domain. Both residues are conserved with mammalian AF4 family members. It is tempting to speculate that MLL-AF4 family fusion proteins in leukemias may simultaneously deregulate the MAPK pathway in addition to providing an MLL gain of function. In agreement with this possibility, ras mutations (that presumably activate the MAPK pathway) are not commonly associated with MLL leukemias (Mahgoub et al., 1998).

Mechanism by which MLL fusion proteins transform hematopoietic progenitors

All three general types of oncogenic MLL proteins retain the amino terminal AT hook motifs of MLL, suggesting that they may recognize putative MLL response elements within the regulatory regions of target genes. It is likely, therefore, that a common mechanism for transformation by these proteins occurs via an MLL-dependent pathway. However, it has been unclear whether such an MLL-dependent mechanism utilizes loss or gain of MLL function, particularly given the structural diversity of many of the MLL partner proteins and the ability to promote leukemic disease without fusion to a partner protein.

Two experimental approaches have demonstrated that MLL fusion genes act as dominant alleles. First, chimeric mice generated from ES cells carrying a single MLL-AF9 'knock-in' allele succumbed to AML, whilst mice generated from MLL heterozygote ES cells remained disease free (Corral et al., 1996). Transcription of the knock-in allele is under the control of the endogenous MLL promoter elements and therefore its expression closely recapitulates the spatial, temporal and quantitative level of regulation of the MLL-AF9 allele in human leukemias. In a second parallel approach, retroviral transduction was used to drive expression of the MLL-ENL fusion protein in primary myeloid progenitors. Expression of MLL-ENL, but neither the amino terminal region of MLL or the carboxy terminal region of ENL yielded myeloid transformation in vitro or the development of AML in vivo (Lavau et al., 1997).

Further confirmation that MLL fusion proteins act via a dominant gain-of-function has been recently reported (Dobson et al., 1999). If MLL fusion proteins act as dominant-negative inhibitors of MLL function, their activity would be predicted to block embryonic development at a stage when normal MLL function is required. The MLL-AF9 'knock-in' allele has been transmitted through the germline to generate a mouse strain genetically predisposed to AML. Since MLL function is required for mid-gestation embryogenesis (Yu et al., 1995; Yagi et al., 1998), the production of viable germline MLL-AF9 mutant mice confirms that expression of the fusion protein does not compromise MLL function in vivo (Dobson et al., 1999). This distinguishes MLL fusion protein mediated transformation from that of AML1/CBFbeta fusion proteins, which phenocopy AML1/CBFbeta loss of function in vivo due to their dominant negative activity (Castilla et al., 1996; Okuda et al., 1996; Wang et al., 1996; Yergeau et al., 1997). Interestingly, MLL-AF9 germline mutant mice exhibit an early, rapid expansion of myeloid progenitors prior to disease onset (Dobson et al., 1999). Recent data from a second knock-in model suggest that fusion of MLL to beta-galactosidase is sufficient to promote leukemogenesis without involvement of an MLL partner protein. In this model, tumor latency was twice as long as that for the MLL-AF9 knock-in strain (Dobson et al., 2000). It was proposed that oligomerization mediated by beta-galactosidase could confer leukemogenic properties to MLL.

MLL loss of function has also been suggested as a potential mechanism contributing to hematopoietic transformation. This could be partial (haplo-insufficiency) or complete (either by dominant negative inhibition or recessive mutation of both alleles). Mutation of both copies of MLL by translocation on one allele and complete deletion of the second allele has been reported in the ML-1 leukemic cell line, although such events have not been consistently observed (Strout et al., 1996). The SET domain of wild type MLL interacts with the tumor suppressor SNF5 (Rozenblatt-Rosen et al., 1998; Versteege et al., 1998) raising the possibility that loss of this interaction in MLL-deficient cells could contribute to leukemogenesis. SNF5 is present in at least two distinct complexes with either MLL or the SWI/SNF chromatin remodeling complex and it is currently not clear which complex mediates its growth inhibitory activity (Croce, 1999). Conversely, genetic analysis indicates that chimeric or germline heterozygous mice carrying a single mutant copy of MLL are not predisposed to acute leukemogenesis over an 18 month observation period (Corral et al., 1996; Ayton, unpublished observations). Moreover, MLL fusion proteins transform primary myeloid progenitors in the presence of two copies of the wild type MLL gene (Lavau et al., 1997). It therefore seems unlikely that MLL haplo-insufficiency plays a critical role in MLL-mediated leukemogenesis.

Induction of apoptosis by MLL fusion proteins

Many attempts to obtain stable expression of MLL fusion proteins in established cell lines or transgenic mice have failed. A recent report by Caslini et al. (2000) suggests a molecular explanation for these observations. Induction of high level expression of the MLL-AF9 fusion protein in U937 cells led to rapid cell death within 72 h and was associated with prior G1 arrest, up regulation of the p53 target gene p21CIP1 and induction of differentiation. Interestingly, the U937 cell line is p53 deficient, suggesting that death induced by MLL fusion proteins may be p53 independent. However, extremely low levels of MLL fusion protein expression are tolerated when expression is delivered by retrovirus to primary myeloid hematopoietic progenitors without concomitant death. How can we reconcile the ability of MLL fusion proteins to provoke either apoptosis or immortalization?

The effects of MLL fusion proteins may be dosage dependent, such that high levels of expression above a particular threshold induce an apoptotic response whilst low levels are sufficient to promote immortalization. In support of this possibility, in its normal physiological context, MLL fusion protein expression in human leukemias and the MLL-AF9 knock-in mouse model, are driven by the weak endogenous MLL promoter. Furthermore, despite the ability of the MSCV retroviral LTR to drive high level expression, all immortal primary myeloid cell lines transduced by MLL fusion genes express extremely low levels of transcript that can only be detected by sensitive methods such as RT-PCR. These results suggest that myeloid immortalization selects for retroviral integrations that facilitate low level expression of MLL fusion proteins. Many expression vectors and transgenic cassettes direct high level expression, suggesting that the previous failure to generate stable expression of MLL fusion proteins in cell lines in vitro or transgenic mice in vivo may have been due to toxic levels of expression.

Alternatively, MLL fusion protein-mediated immortalization may only occur within a developmentally restricted hematopoietic progenitor population. Inappropriate expression of MLL fusion proteins in cells other than the critical susceptible target population may result in activation of an apoptotic response. Finally, the ability of MLL fusion proteins to promote both an apoptotic response as well as immortalization may be simultaneous events. However, immortalization may only take place in a cell in which the MLL death pathway is non-functional, due to either the lack of expression or activity of a critical component of the death pathway or the prior acquisition of a mutation that abrogates the MLL apoptotic response. Many MLL fusion protein associated acute leukemias occur in infants (reviewed in Felix and Lange, 1999). The extremely short latency of MLL infant leukemias suggests that these individuals may be genetically predisposed to acute leukemogenesis. A genetic lesion that abrogates the MLL death pathway could represent one such mutation.

Lineage selective transformation by MLL fusion proteins - do MLL partner proteins specify the lineage of transformation?

Prior to the molecular cloning of numerous MLL partner genes, it had been noted that particular translocations involving the 11q23 region were specifically associated with either acute lymphoid or myeloid leukemias. Two possible mechanisms could account for lineage specific transformation by MLL fusion proteins. First, the illegitimate recombination events that generate chromosomal translocations between MLL and a particular partner gene may occur in a lineage-specific fashion. In this scenario, the resulting MLL fusion proteins would retain an ability to transform either lymphoid or myeloid progenitors, but their expression would be restricted to the particular lineage in which the translocation occurred. In such a model, lineage-specific transformation would be dependent upon factors that influence the propensity of MLL and specific partner genes to undergo illegitimate recombination.

Alternatively, lineage-specific transformation by MLL fusion proteins may be driven by partner protein function. Since particular partner proteins are preferentially associated with transformation of either lymphoid or myeloid lineages, such a model dictates that the normal function of an MLL partner protein that participates in lineage-specific transformation may also be lineage-restricted. Although expression patterns for most of the MLL partner proteins do not immediately suggest lineage-restricted functions, post-translational modification may dictate their biochemical activity. Thus, it remains plausible that a ubiquitously expressed partner protein could interact with a lineage-specific cofactor to mediate lineage-restricted function, as previously described for the widely expressed E2a proteins that are essential for normal B lymphopoiesis (Bain et al., 1994).

Most recently, the AF4 homologue LAF4 has been identified as an MLL partner protein in a single case of pro-B ALL (von Bergh et al., 2000). However, unlike the widely expressed AF4, LAF4 exhibits lymphoid-restricted expression subject to both spatial and temporal regulation during B lymphopoiesis (Ma and Staudt; 1996). High levels of LAF4 expression were detected in immature pre-B cell lines, with reduced levels associated with progressive B cell maturation and complete loss of expression in plasma cells. Outside the B cell lineage, high levels of LAF4 expression were only detected in tissues such as spleen and thymus that facilitate lymphopoiesis, and was otherwise absent from many other tissues examined. Thus, the lineage and stage restricted expression profile of LAF4 suggests a potential lineage-restricted function during the early stages of B lymphopoiesis. The ability of at least one MLL partner protein, LAF4, to participate in lineage and stage specific transformation of the same lineage and stage in which it is preferentially expressed, supports an instructive role for the MLL partner protein in directing lineage susceptibility for transformation. On this basis, we predict that other MLL partner proteins that participate in 11q23 translocations associated with AML, such as ELL, AF9 and AF10, may perform critical functions during myelopoiesis.

Human acute leukemias and mouse models both suggest that lineage selection is not absolute but preferential. For example, of 183 cases of the MLL-AF4-expressing acute leukemias, 94.5% (173 cases) were pro B ALL, whilst only 3.3% (six cases) were diagnosed as AML (Johansson et al., 1998). The majority of MLL-AF9 knock-in mice develop AML, despite the fact that the MLL promoter is active in both lymphoid and myeloid lineages. Few mice developed B lineage ALL, albeit with longer latency than AML (Hess et al., 1997; Dobson et al., 1999).

Progression of leukemic disease initiated by MLL fusion proteins

Retroviral gene transfer and knock-in mouse models have confirmed that expression of MLL fusion proteins promote acute leukemogenesis, but the long latency in vivo and monoclonal nature of the leukemias suggest that further secondary mutations are required (Lavau et al., 1997, 2000a,b; Dobson et al., 1999). What is the nature of these cooperative mutations?

Enhanced expression of MLL fusion proteins is associated with disease progression

A potential class of mutations that are acquired during the transition to overt leukemia is suggested by studies using MSCV retroviral constructs that contain an IRES EGFP cassette downstream of either MLL-CBP or MLL-ELL. The IRES element allows the production of two distinct proteins from a single bi-cistronic transcript: an upstream MLL fusion protein and the downstream EGFP (Mountford et al., 1994). With such retroviral vectors, levels of EGFP expression are directly correlated with the transcriptional activity of the 5' LTR element that drives expression of the MLL fusion protein.

Approximately 10 weeks after bone marrow transplantation of hematopoietic stem cells transduced with either MLL-CBP or MLL-ELL, analysis of EGFP expression by flow cytometry revealed an almost complete lack of myeloid cells expressing significant levels of the MLL fusion protein (Lavau et al., 2000a,b). However, these animals eventually succumbed to a fatal AML disease between 100-200 days, with all tumors expressing high levels of the MLL fusion protein, as assessed by expression of EGFP. These results suggest that the evolution of AML in vivo is associated with clonal selection for secondary mutations that facilitate enhanced MLL fusion protein expression.

In further support of this notion, Western blot analysis of a variety of cell lines derived from human 11q23 acute leukemias readily detect expression of MLL fusion proteins (Joh et al., 1996; Li et al., 1998). Primary in vitro immortalized murine myeloid progenitor cell lines are able to induce AML after transplantation into recipient mice, but only after a significant latency period (Lavau et al., 1997; DiMartino et al., 2000). In such cell lines only extremely sensitive methods such as RT-PCR are capable of detecting MLL fusion protein expression in primary murine immortalized myeloid progenitor cell lines (Lavau et al., 1997; DiMartino et al., submitted).

Are there functional differences within the leukemogenic MLL fusion protein family?

For human acute leukemias that carry 11q23 translocations expressing an MLL fusion protein, information is available regarding the age of individuals at disease presentation, but we are unable to determine when the MLL translocation event took place and therefore, when MLL-mediated immortalization was initiated. Thus, it is impossible to estimate latency periods for the development of human adult leukemias. However, the expression of a specific subset of MLL fusion proteins is preferentially associated with the development of infant acute leukemia. Since infant acute leukemias arise within the first year after birth, one can estimate that a particularly short latency takes place from an initial immortalization event to the accumulation of secondary mutations and finally acute leukemogenesis. Thus, the ability to induce short latency acute leukemogenesis in infants can be used as a biological phenotype to discriminate potential functional differences within the MLL fusion protein family.

Interestingly, recent studies using mouse in vivo models of MLL mediated leukemogenesis have indicated that, similar to human leukemias, dominant expression of specific MLL fusion proteins results in the development of AML. However, functional analysis for four distinct MLL fusion proteins suggests that the latency for onset of AML in vivo is variable and partner protein dependent (Lavau et al., 1997, 2000a,b; Dobson et al., 1999; DiMartino et al., submitted). In vivo data from mouse models have been collected from two independent protocols. Firstly, immortalized cell lines were transplanted into sub-lethally irradiated syngeneic recipients and/or immuno-compromised SCID recipients (Lavau et al., 1997; DiMartino et al., submitted). Secondly, cells enriched for hematopoietic stem cell activity and transduced with MLL fusion retroviruses were transplanted into lethally irradiated syngeneic recipients (Lavau et al., 1997, 2000a,b). The results of such studies are summarized in Figure 5. Thus, murine myeloid progenitors expressing MLL-ENL or MLL-AF10 fusion proteins have the ability to promote AML in vivo with a short latency, whereas the MLL-ELL and MLL-CBP fusion proteins require significantly longer latency periods prior to the onset of AML. In stark contrast, all four MLL fusion proteins behave in an identical manner in their ability to immortalize myeloid progenitors in vitro.

Do MLL fusion proteins segregate into two distinct complementation groups?

How can we explain the ability of distinct MLL fusion proteins to induce rapid or long latency tumors in vivo? An interesting correlation can be observed between examples of MLL mutants that promote tumors with long latency in both humans and mice. In human leukemias, some MLL fusion proteins including MLL-ELL and MLL-CBP, as well as the MLL partial tandem duplication mutants predominantly occur in elderly patients (although the sample size for MLL-CBP is extremely small due to its rare occurrence). Similarly, in mice, the MLL-beta-galactosidase knock-in allele also promotes leukemogenesis with long latency, without involvement of a partner protein. We propose that MLL fusion proteins that promote short versus long latency leukemias may represent distinct complementation groups. What links the individual members within such groups? Do they suggest functional interactions between members of each group?

A number of explanations could account for the variable latency of leukemia onset induced by MLL mutants. First, all leukemogenic MLL mutants may regulate an identical array of MLL dependent target genes. However, distinct MLL fusion proteins may induce quantitative differences in the rate and/or efficiency of target gene transcription, as depicted in Figure 6. In this model, variable tumor latency is determined by MLL target gene dosage, such that MLL-ENL and MLL-AF10 fusion proteins induce significantly higher rates of target gene transcription relative to MLL-ELL or MLL-CBP.

Alternatively, distinct MLL fusion proteins may disrupt a variable range of subordinate pathways (see Figure 6). It seems likely that all leukemogenic MLL mutants disrupt at least some common target pathways, as assessed by their ability to immortalize primary myeloid progenitors in vitro. However, MLL fusion proteins that promote AML with long latency may require the acquisition of at least one additional rate limiting mutation before disease progression can occur. In contrast, short latency leukemias associated with fusion of MLL with specific partner proteins such as ENL, AF9 and AF10 suggest that in these cases, a simultaneous double hit may occur, reducing the requirement for subsequent mutations in pathways that negatively regulate the in vivo expansion of myeloid progenitors immortalized by MLL-ELL or MLL-CBP. This 'second hit' may represent an MLL independent mutation in addition to a gain of MLL function. Whilst ample genetic and biochemical data suggest that many MLL partner proteins retain the potential to contribute to MLL mediated leukemogenesis, it is currently unclear whether expression of select MLL fusion proteins results in a second hit that disrupts partner protein function.

Is disruption of partner protein function a contributory factor in MLL fusion protein mediated leukemogenesis?

A recent study for acute promyelocytic leukemia (APL) has suggested that both reciprocal products from the t(11; 17) translocation are required for full APL development in mice (He et al., 2000). Whilst transgenic expression of PLZF-RARA resulted in development of a chronic myeloid leukemia (CML), co-expression with the reciprocal RARA-PLZF fusion protein in compound transgenic animals induced classical APL. This is the first example of a leukemia-associated translocation that generates two distinct proteins that cooperate to induce a specific tumor phenotype. Interestingly, the protein product from the reciprocal chromosomal translocation, RARA-PLZF, acts as a dominant negative molecule to inhibit PLZF function (He et al., 2000). Future studies will determine whether both products of other balanced translocations in leukemias act to simultaneously disrupt the function of both genes that participate in chromosomal translocations.

The critical role of the partner protein in converting MLL to an oncogenic fusion protein via a dominant gain of MLL function is now well documented (Corral et al., 1996; Lavau et al., 1997; Slany et al., 1998; Dobson et al., 1999; DiMartino et al., 2000). However, as mentioned earlier, it is not clear whether the sole transforming effect of the MLL fusion protein is due to a gain of MLL function. We have previously speculated that MLL fusion proteins may result in a double genetic hit whose cooperative effects promote leukemogenesis (Ayton and Cleary, 2001). In contrast to APL, expression of the der(11) encoded MLL fusion protein alone is frequently observed in human leukemias and is sufficient to induce AML in mouse models. Thus, unlike the recently described APL paradigm, any potential disruption to both MLL dependent and independent pathways would be due to the dominant action of a single fusion protein.

Is there any evidence that MLL fusion proteins disrupt an MLL independent pathway via alteration of partner protein function to promote leukemogenesis? Many of the proteins fused to MLL are novel and of unknown function, but can be categorized into two broad classes consisting of nuclear or cytoplasmic proteins (Ayton and Cleary, 2001). Whilst the nuclear partner proteins have demonstrated transcriptional activation potential in vitro, their in vivo functions are mostly unknown (Rubnitz et al., 1994; Prasad et al., 1995; Ma and Staudt, 1996; Slany et al., 1998). Two MLL partner proteins, CBP and AF10, participate in reciprocal chromosomal translocations that result in fusions with genes other than MLL (Borrow et al., 1996; Dreyling et al., 1996, 1998). Recent additions to the MLL partner protein family, identified from rare translocations, have included examples of proteins of known function, such as CBP, its close homologue p300, and Abi-1, a negative regulator of c-abl (Ida et al., 1997; Sobulo et al., 1997; Taki et al., 1997, 1998). Genetic and biochemical data have implicated the cytoplasmic partner protein AF6 and the forkhead transcription factors AFX and AF6q21 as components of mitogenic and survival signaling pathways, respectively (Miyamoto et al., 1995; Kuriyama et al., 1996; Ogg et al., 1997; Paradis and Ruvkun, 1998; Brunet et al., 1999; Kops et al., 1999; Medema et al., 2000).

Mechanistically, how could expression of an MLL fusion protein subvert the function of the native partner protein? One possibility is that partner protein function is subject to haploinsufficiency after disruption of one allele by a translocation event. CBP exhibits haploinsufficiency associated either with multiple developmental defects of Rubinstein-Taybi syndrome in humans or variably penetrant embryonic lethal phenotype of heterozygous mouse mutants (Petrij et al., 1995; Yao et al., 1998). However, the ability of MLL fusion proteins, such as MLL-CBP, MLL-ELL and MLL-AF10, to immortalize myeloid progenitors in vitro and induce AML in vivo in the presence of two copies of the respective partner genes, suggests that partner protein haplo-insufficiency may not play a critical role in MLL fusion protein-mediated transformation (Lavau et al., 1997, 2000a,b; Di Martino et al., 2000; Di Martino et al., submitted).

Alternatively, MLL fusion proteins may also exert a dominant effect on the partner protein pathway as well as the MLL pathway. Whether such a dominant effect would result in a subsequent loss or gain of function may well be dependent upon the context of partner protein function. For example, AF6q21/FKHRL1 and AFX function to directly transactivate target genes such as Fas ligand and the tumor suppressor p27kip1 to promote apoptosis and/or arrest of the cell cycle in G1 (Brunet et al., 1999; Medema et al., 2000). Thus, an MLL-AF6q21 fusion protein that behaves as a dominant negative inhibitor of AF6q21 function could induce enhanced cell survival and/or deregulated G1 checkpoint control in addition to an MLL gain of function to cooperatively promote leukemogenesis. A similar argument could be made for the MLL-Abil fusion protein. Abil functions as a negative regulator of c-abl tyrosine kinase activity and v-abl transformation (Shi et al., 1995). Disruption of normal Abil function via dominant-negative inhibition could result in oncogenic activation of c-abl. Perhaps the most compelling example of an MLL partner protein with myeloid growth suppressive properties is the recently described GRAF protein, a putative GTPase activating protein (GAP) for RhoA (Borkhardt et al., 2000). In addition to its fusion to MLL, bi-allelelic mutations of the GRAF gene were identified in cases of MDS and AML. GRAF mutations included deletions of one allele associated with either point mutation or insertions into the GAP domain of the second allele. GRAF represents the first MLL partner protein directly associated with tumor suppressor properties.

Conversely, some MLL fusion proteins may act as dominant gain of function mutants of partner protein function. For example, genetic and biochemical data implicate mammalian AF6 and its fly homologue Canoe, as a novel downstream effector of Ras and Notch signaling (Miyamoto et al., 1995; Kuriyama et al., 1996; Matsuo et al., 1997). Most recently, genetic screens in Drosophila have also identified lilli, the fly orthologue of the mammalian AF4 family, as a potential downstream nuclear target and effector of the MAP kinase pathway (Su et al., 2001). Uncontrolled activation of unknown AF6 or AF4 targets along these mitogenic pathways could synergistically promote leukemogenesis in combination with MLL gain of function.

Although some MLL fusion proteins induce myeloid leukemias following short incubation periods in murine experimental models, the fact remains that a finite latency period is required prior to disease onset, suggesting that even the most potent MLL fusion proteins are not sufficient for development of AML. Furthermore, the primary leukemias that develop are monoclonal and their transplantation into secondary recipients results in a substantial reduction of tumor latency. The future use of genetic screens should elucidate the cooperative events required for the progression of MLL-associated leukemias.

Acknowledgements

We thank our colleagues in the Cleary Lab for many helpful discussions and sharing of unpublished results. Our studies are supported in part by funds from the National Cancer Institute.

References

Ayton, Cleary. (2001). Transcription Factors: Normal and Malignant Development of Blood Cells Ravid K and Licht J (eds). Wiley-Liss: New York, pp. 447-463.

Bahri SM, Chia W, Yang X. (2001). Mech. Dev. 100, 291-301. MEDLINE

Bain G, Maandag EC, Izon DJ, Amsen D, Kruisbeek AM, Weintraub BC, Krop I, Schlissel MS, Feeney AJ, van Roon, M. (1994). Cell 79, 885-892. MEDLINE

Bannister AJ, Kouzarides T. (1996). Nature 384, 641-643. MEDLINE

Borkhardt A, Bojesen S, Haas OA, Fuchs U, Bartelheimer D, Loncarevic IF, Bohle RM, Harbott J, Repp R, Jaeger U, Viehmann S, Henn T, Korth P, Scharr D, Lampert F. (2000). Proc. Natl. Acad. Sci. USA 97, 9168-9173. MEDLINE

Borrow J, Stanton Jr VP, Andresen JM, Becher R, Behm FG, Chaganti RS, Civin CI, Disteche C, Dube I, Frischauf AM, Horsman D, Mitelman F, Volinia S, Watmore AE, Housman DE. (1996). Nat. Genet. 14, 33-41. MEDLINE

Brunet A, Bonni A, Zigmond MJ, Lin MZ, Juo P, Hu LS, Anderson MJ, Arden KC, Blenis J, Greenberg ME. (1999). Cell 96, 857-868. MEDLINE

Cairns BR, Henry NL, Kornberg RD. (1996). Mol. Cell. Biol. 16, 3308-3316. MEDLINE

Caligiuri MA, Strout MP, Schichman SA, Mrozek K, Arthur DC, Herzig GP, Baer MR, Schiffer CA, Heinonen K, Knuutila S, Nousiainen T, Ruutu T, Block AW, Schulman P, Pedersen-Bjergaard J, Croce CM, Bloomfield CD. (1996). Cancer Res. 56, 1418-1425. MEDLINE

Caslini C, Shilatifard A, Yang L, Hess JL. (2000). Proc. Natl. Acad. Sci. USA 97, 2797-2802. MEDLINE

Castilla LH, Wijmenga C, Wang Q, Stacy T, Speck NA, Eckhaus M, Marin-Padilla M, Collins FS, Wynshaw-Boris A, Liu PP. (1996). Cell 87, 687-696. MEDLINE

Chaplin T, Ayton P, Bernard OA, Saha V, Della Valle V, Hillion J, Gregorini A, Lillington D, Berger R, Young BD. (1995a). Blood 85, 1435-1441. MEDLINE

Chaplin T, Bernard O, Beverloo HB, Saha V, Hagemeijer A, Berger R, Young BD. (1995b). Blood 86, 2073-2076. MEDLINE

Cleary ML. (1991). Cell 66, 619-622. MEDLINE

Corral J, Lavenir I, Impey H, Warren AJ, Forster A, Larson TA, Bell S, McKenzie AN, King G, Rabbitts TH. (1996). Cell 85, 853-861. MEDLINE

Croce CM. (1999). Cancer Res. 59, 1778s-1783s. MEDLINE

Cross SH, Meehan RR, Nan X, Bird A. (1997). Nat. Genet. 16, 256-259. MEDLINE

Dhalluin C, Carlson JE, Zeng L, He C, Aggarwal AK, Zhou MM. (1999). Nature 399, 491-496. Article MEDLINE

DiMartino JF, Miller T, Ayton PM, Landewe T, Hess JL, Cleary ML, Shilatifard A. (2000). Blood 96, 3887-3893. MEDLINE

Dobson CL, Warren AJ, Pannell R, Forster A, Lavenir I, Corral J, Smith AJ, Rabbitts TH. (1999). EMBO J. 18, 3564-3574. MEDLINE

Dobson CL, Warren AJ, Pannell R, Forster A, Rabbitts TH. (2000). EMBO J. 19, 843-851. MEDLINE

Domer PH, Fakharzadeh SS, Chen CS, Jockel J, Johansen L, Silverman GA, Kersey JH, Korsmeyer SJ. (1993). Proc. Natl. Acad. Sci. USA 90, 7884-7888. MEDLINE

Dreyling MH, Martinez-Climent JA, Zheng M, Mao J, Rowley JD, Bohlander SK. (1996). Proc. Natl. Acad. Sci. USA 93, 4804-4809. Article MEDLINE

Dreyling MH, Schrader K, Fonatsch C, Schlegelberger B, Haase D, Schoch C, Ludwig W, Loffler H, Buchner T, Wormann B, Hiddemann W, Bohlander SK. (1998). Blood 91, 4662-4667. MEDLINE

Felix CA, Lange BJ. (1999). Oncologist 4, 225-240. MEDLINE

Galoian K, Milne T, Brock H, Shilatifard A, Slany R, Hess J. (2000). Blood 96, 457a.

Gecz J, Gedeon AK, Sutherland GR, Mulley JC. (1996). Nat. Genet. 13, 105-108. MEDLINE

Gu Y, Shen Y, Gibbs RA, Nelson DL. (1996). Nat. Genet. 13, 109-113. MEDLINE

Hayette S, Tigaud I, Vanier A, Martel S, Corbo L, Charrin C, Beillard E, Deleage G, Magaud JP, Rimokh R. (2000). Oncogene 19, 4446-4450. MEDLINE

He L, Bhaumik M, Tribioli C, Rego EM, Ivins S, Zelent A, Pandolfi PP. (2000). Mol. Cell 6, 1131-1141. MEDLINE

Hendrich B, Bird A. (1998). Mol. Cell. Biol. 18, 6538-6547. MEDLINE

Hess JL, Yu BD, Li B, Hanson R, Korsmeyer SJ. (1997). Blood 90, 1799-1806. MEDLINE

Huth JR, Bewley CA, Nissen MS, Evans JN, Reeves R, Gronenborn AM, Clore GM. (1997). Nat. Struct. Biol. 4, 657-665. MEDLINE

Ida K, Kitabayashi I, Taki T, Taniwaki M, Noro K, Yamamoto M, Ohki M, Hayashi Y. (1997). Blood 90, 4699-4704. MEDLINE

Isnard P, Core N, Naquet P, Djabali M. (2000). Blood 96, 705-710. MEDLINE

Joh T, Kagami Y, Yamamoto K, Segawa T, Takizawa J, Takahashi T, Ueda R, Seto M. (1996). Oncogene 13, 1945-1953. MEDLINE

Johansson B, Moorman AV, Haas OA, Watmore AE, Cheung KL, Swanton S, Secker-Walker LM. (1998). Leukemia 12, 779-787. MEDLINE

Kops GJ, de Ruiter ND, De Vries-Smits AM, Powell DR, Bos JL, Burgering BM. (1999). Nature 398, 630-634. Article MEDLINE

Kourlas PJ, Strout MP, Becknell B, Veronese ML, Croce CM, Theil KS, Krahe R, Ruutu T, Knuutila S, Bloomfield CD, Caligiuri MA. (2000). Proc. Natl. Acad. Sci. USA 97, 2145-2150. Article MEDLINE

Kuriyama M, Harada N, Kuroda S, Yamamoto T, Nakafuku M, Iwamatsu A, Yamamoto D, Prasad R, Croce C, Canaani E, Kaibuchi K. (1996). J. Biol. Chem. 271, 607-610. MEDLINE

Kuwada N, Kimura F, Matsumura T, Yamashita T, Nakamura Y, Wakimoto N, Ikeda T, Sato K, Motoyoshi K. (2001). Cancer Res. 61, 2665-2669. MEDLINE

Lavau C, Szilvassy SJ, Slany R, Cleary ML. (1997). EMBO J. 16, 4226-4237. MEDLINE

Lavau C, Du C, Thirman M, Zeleznik-Le N. (2000a). EMBO J. 19, 4655-4664. MEDLINE

Lavau C, Luo RT, Du C, Thirman MJ. (2000b). Proc. Natl. Acad. Sci. USA 97, 10984-10989. MEDLINE

Lavau C, Luo RT, Simeone F, Polak PE, Du C, Kawamata S, Thirman MJ. (2000c). Blood 96, 573a.

Li Q, Frestedt JL, Kersey JH. (1998). Blood 92, 3841-3847. MEDLINE

Lochner K, Siegler G, Fuhrer M, Greil J, Beck JD, Fey GH, Marschalek R. (1996). Cancer Res. 56, 2171-2177. MEDLINE

Ma C, Staudt LM. (1996). Blood 87, 734-745. MEDLINE

Mahgoub N, Parker RI, Hosler MR, Close P, Winick NJ, Masterson M, Shannon KM, Felix CA. (1998). Genes Chromo. Cancer 21, 270-275.

Martinez-Balbas MA, Bannister AJ, Martin K, Haus-Seuffert P, Meisterernst M, Kouzarides T. (1998). EMBO J. 17, 2886-2893. Article MEDLINE

Matsuo T, Takahashi K, Kondo S, Kaibuchi K, Yamamoto D. (1997). Development 124, 2671-2680. MEDLINE

Medema RH, Kops GJ, Bos JL, Burgering BM. (2000). Nature 404, 782-787. Article MEDLINE

Megonigal MD, Cheung NK, Rappaport EF, Nowell PC, Wilson RB, Jones DH, Addya K, Leonard DG, Kushner BH, Williams TM, Lange BJ, Felix CA. (2000). Proc. Natl. Acad. Sci. USA 97, 2814-2819. MEDLINE

Mitelman F, Johansson B, Mertens F. (1994). Catalogue of Chromosome Abberations in Cancer 5th Edn. Wiley-Liss: New York,

Miller T, Williams K, Johnstone RW, Shilatifard A. (2000). J. Biol. Chem. 275, 32052-32056. MEDLINE

Miyamoto H, Nihonmatsu I, Kondo S, Ueda R, Togashi S, Hirata K, Ikegami Y, Yamamoto D. (1995). Genes Dev. 9, 612-625. MEDLINE

Morrissey J, Tkachuk DC, Milatovich A, Francke U, Link M, Cleary ML. (1993). Blood 81, 1124-1131. MEDLINE

Morrissey JJ, Raney S, Cleary ML. (1997). Leuk. Res. 21, 911-917. MEDLINE

Mountford P, Zevnik B, Duwel A, Nichols J, Li M, Dani C, Robertson M, Chambers I, Smith A. (1994). Proc. Natl. Acad. Sci. USA 91, 4303-4307. MEDLINE

Nakamura T, Alder H, Gu Y, Prasad R, Canaani O, Kamada N, Gale RP, Lange B, Crist WM, Nowell PC, Croce CM, Canaani, E. (1993). Proc. Natl. Acad. Sci. USA 90, 4631-4635. MEDLINE

Nilson I, Reichel M, Ennas MG, Greim R, Knorr C, Siegler G, Greil J, Fey GH, Marschalek R. (1997). Br. J. Haematol. 98, 157-169. MEDLINE

Nissen MS, Langan TA, Reeves R. (1991). J. Biol. Chem. 266, 19945-19952. MEDLINE

Ogg S, Paradis S, Gottlieb S, Patterson GI, Lee L, Tissenbaum HA, Ruvkun G. (1997). Nature 389, 994-999. Article MEDLINE

Okuda T, van Deursen J, Hiebert SW, Grosveld G, Downing JR. (1996). Cell 84, 321-330. MEDLINE

Paradis S, Ruvkun G. (1998). Genes Dev. 12, 2488-2498. Article MEDLINE

Pegram LD, Megonigal MD, Lange BJ, Nowell PC, Rowley JD, Rappaport EF, Felix CA. (2000). Blood 96, 4360-4362. MEDLINE

Petrij F, Giles RH, Dauwerse HG, Saris JJ, Hennekam RC, Masuno M, Tommerup N, van Ommen GJ, Goodman RH, Peters DJ, Breuning MH. (1995). Nature 376, 348-351. MEDLINE

Prasad R, Yano T, Sorio C, Nakamura T, Rallapalli R, Gu Y, Leshkowitz D, Croce CM, Canaani E. (1995). Proc. Natl. Acad. Sci. USA 92, 12160-12164. MEDLINE

Rabbitts TH. (1994). Nature 372, 143-149. MEDLINE

Reeves R, Nissen MS. (1990). J. Biol. Chem. 265, 8573-8582. MEDLINE

Reeves R, Langan TA, Nissen MS. (1991). Proc. Natl. Acad. Sci. USA 88, 1671-1675. MEDLINE

Rowley JD. (1992). Genes Chrom. Cancer 5, 264-266. MEDLINE

Rozenblatt-Rosen O, Rozovskaia T, Burakov D, Sedkov Y, Tillib S, Blechman J, Nakamura T, Croce CM, Mazo A, Canaani E. (1998). Proc. Natl. Acad. Sci. USA 95, 4152-4157. Article MEDLINE

Rubnitz JE, Morrissey J, Savage PA, Cleary ML. (1994). Blood 84, 1747-1752. MEDLINE

Sano K, Hayakawa A, Piao JH, Kosaka Y, Nakamura H. (2000). Blood 95, 1066-1068. MEDLINE

Schichman SA, Caligiuri MA, Gu Y, Strout MP, Canaani E, Bloomfield CD, Croce CM. (1994). Proc. Natl. Acad. Sci. USA 91, 6236-6239. MEDLINE

Secker-Walker LM. (1998). Leukemia 12, 776-778. MEDLINE

Shi Y, Alin K, Goff SP. (1995). Genes Dev. 9, 2583-2597. MEDLINE

Shilatifard A, Lane WS, Jackson KW, Conaway RC, Conaway JW. (1996). Science 271, 1873-1876. MEDLINE

Shilatifard A, Duan DR, Haque D, Florence C, Schubach WH, Conaway JW, Conaway RC. (1997a). Proc. Natl. Acad. Sci. USA 94, 3639-3643.

Shilatifard A, Haque D, Conaway RC, Conaway JW. (1997b). J. Biol. Chem. 272, 22355-22363. MEDLINE

Slany RK, Lavau C, Cleary ML. (1998). Mol. Cell. Biol. 18, 122-129. MEDLINE

Sobulo OM, Borrow J, Tomek R, Reshmi S, Harden A, Schlegelberger B, Housman D, Doggett NA, Rowley JD, Zeleznik-Le NJ. (1997). Proc. Natl. Acad. Sci. USA 94, 8732-8737. Article MEDLINE

Strout MP, Mrozek K, Heinonen K, Sait SN, Shows TB, Aplan PD. (1996). Genes Chrom. Cancer 16, 204-210. MEDLINE

Su M, Wisotzkey R, Newfeld S. (2001). Genetics 157, 717-725. MEDLINE

Syntichaki P, Topalidou I, Thireos G. (2000). Nature 404, 414-417. Article MEDLINE

Taki T, Sako M, Tsuchida M, Hayashi Y. (1997). Blood 89, 3945-3950. MEDLINE

Taki T, Shibuya N, Taniwaki M, Hanada R, Morishita K, Bessho F, Yanagisawa M, Hayashi Y. (1998). Blood 92, 1125-1130. MEDLINE

Taki T, Kano H, Taniwaki M, Sako M, Yanagisawa M, Hayashi Y. (1999). Proc. Natl. Acad. Sci. USA 96, 14535-14540. MEDLINE

Tang AH, Neufeld TP, Rubin GM, Muller HA. (2001). Development 128, 801-813. MEDLINE

Thirman MJ, Levitan DA, Kobayashi H, Simon MC, Rowley JD. (1994). Proc. Natl. Acad. Sci. USA 91, 12110-12114. MEDLINE

Tkachuk DC, Kohler S, Cleary ML. (1992). Cell 71, 691-700. MEDLINE

Tsai FY, Keller G, Kuo FC, Weiss M, Chen J, Rosenblatt M, Alt FW, Orkin SH. (1994). Nature 371, 221-226. MEDLINE

Versteege I, Sevenet N, Lange J, Rousseau-Merck MF, Ambros P, Handgretinger R, Aurias A, Delattre O. (1998). Nature 394, 203-206. Article MEDLINE

von Bergh A, Beverloo B, Slater R, Groot A, Rombout P, Kluin P, Schuuring E. (2000). Blood 96, 693a.

Wang Q, Stacy T, Binder M, Marin-Padilla M, Sharpe AH, Speck NA. (1996). Proc. Natl. Acad. Sci. USA 93, 3444-3449. Article MEDLINE

Wittwer F, van der Straten A, Keleman K, Dickson BJ, Hafen E. (2001). Development 128, 791-800. MEDLINE

Yagi H, Deguchi K, Aono A, Tani Y, Kishimoto T, Komori T. (1998). Blood 92, 108-117. MEDLINE

Yano T, Nakamura T, Blechman J, Sorio C, Dang CV, Geiger B, Canaani E. (1997). Proc. Natl. Acad. Sci. USA 94, 7286-7291. MEDLINE

Yao TP, Oh SP, Fuchs M, Zhou ND, Ch'ng LE, Newsome D, Bronson RT, Li E, Livingston DM, Eckner R. (1998). Cell 93, 361-372. MEDLINE

Yergeau DA, Hetherington CJ, Wang Q, Zhang P, Sharpe AH, Binder M, Marin-Padilla M, Tenen DG, Speck NA, Zhang DE. (1997). Natl. Genet. 15, 303-306.

Yu BD, Hess JL, Horning SE, Brown GA, Korsmeyer SJ. (1995). Nature 378, 505-508. MEDLINE

Yu BD, Hanson RD, Hess JL, Horning SE, Korsmeyer SJ. (1998). Proc. Natl. Acad. Sci. USA 95, 10632-10636. MEDLINE

Zeleznik-Le NJ, Harden AM, Rowley JD. (1994). Proc. Natl. Acad. Sci. USA 91, 10610-10614. MEDLINE

Figures

Figure 1 Schematic depictions of wild type and mutant MLL proteins. Various putative functional domains are shown as colored boxes and labeled as follows: AT1-3, AT hook DNA binding motifs 1, 2 and 3; SNL 1 and 2, speckled nuclear localization signals 1 and 2; CxxC, cysteine rich motif homologous to DNA methyltransferase and MBD1; PHD 1-3, PHD fingers 1, 2 and 3; TA, transactivation domain; SET, SET domain. Regions that are structurally conserved with Drosophila trx are indicated by the (trx) underlabel

Figure 2 The transduction/transplantation assay for immortalization of primary myeloid progenitor populations by MLL fusion proteins. Bone marrow (BM) cells are harvested from mice 5 days after 5-fluorouracil (5-FU) treatment. Progenitors are enriched by immuno-selection and transduced with retroviral constructs encoding MLL fusion genes. Transduced progenitors are either transplanted into syngeneic recipients or plated in methylcellulose medium containing a cocktail of cytokines. Immortalized progenitors continue to form colonies upon serial replating and can be adapted to grow in liquid medium as cell lines, which are tested for leukemogenicity by transplantation into SCID or sublethally irradiated syngeneic mice

Figure 3 The minimal transforming domains of MLL fusion partners. MLL fusion proteins are shown schematically with gray shading of the fusion partners. Boxes with red filling delineate the minimal regions of partner proteins necessary and sufficient for in vitro myeloid immortalization. NTC and TA, conserved amino terminal and transactivation domains of ENL; Bromo + HAT, bromodomain and histone acteyltransferase domains of CBP; OM + LZ, octapeptide and leucine zipper motifs of AF10

Figure 4 The AF-4 family of MLL fusion partners. Various putative functional domains as defined by Nilson et al. (1997) are shown as colored boxes and labeled as follows: NHD, amino terminal homology domain; ALF, region of homology between AF-4, LAF-4 and FMR2, pSER, poly serine stretch, STD, strong transactivation domain; CHD, carboxy terminal homology domain; GLU 1-4, glutamine rich regions 1-4; ATH, AT hook. Numbering denotes the amino acid residues comprising each of the respective homology domains

Figure 5 Latency periods for induction of murine myeloid leukemias by MLL fusion proteins. The leukemogenic potentials of murine myeloid cells expressing MLL fusion proteins were evaluated under different experimental transplant conditions. Open bars indicate the latency for development of leukemia following transplantation of primary myeloid cells directly after transduction with recombinant retroviruses. Shaded and black bars denote latencies following transplantation of immortalized cell lines into SCID or irradiated, syngeneic mice, respectively. Regardless of methodology, some MLL fusion proteins (e.g. MLL-ENL and MLL-AF10) are consistently associated with shorter latencies for development of leukemias

Figure 6 Alternative models of MLL-mediated leukemogenesis. Hypothetical scenarios that could account for differences in latency of MLL-associated leukemias are depicted. On the left, simple gain of MLL function results in long latency leukemias due to requirements for secondary events not directly mediated by MLL or its partner protein. On the right, alternative scenarios directly associated with the primary MLL translocation are proposed to collaborate with MLL gain of function to shorten the latency for leukemia development. These may involve the MLL subordinate pathway (e.g. quantitative or qualitative differences in MLL target gene expression dependent on the specific partner fused with MLL) or pathways regulated by the MLL fusion partner (e.g. reduced dosage of the wild type partner protein or dominant effects on its function mediated by the MLL fusion protein)

10 September 2001, Volume 20, Number 40, Pages 5695-5707
Table of contents    Previous  Article  Next    [PDF]