The majority of transcription factors that are altered by chromosomal translocations in the leukemias and lymphomas (Table 19-1) can be classified into four major types on the basis of recurring structural elements within their DNA- and protein-binding domains: basic region/helix-loop-helix (bHLH), basic region/leucine zipper (bZIP), zinc finger, and homeodomain.6,8,15 Other less common but still functionally significant motifs include A-T hook, Ets-like, Runt homology, and cysteine-rich (LIM). In some cases, a transcription factor gene is rearranged to a site adjacent to a T-cell receptor (TCR) or immunoglobulin (Ig) locus, resulting in dysregulated expression of the proto-oncogenic sequences. A second, perhaps more common mechanism involves chromosomal rearrangements that fuse transcriptional control genes into functional chimeras. Such fusions are important because they give rise to novel proteins capable of interacting with DNA and other regulatory elements in ways that usurp normal cellular control mechanisms.6
Table 19-1:Transcriptional Control Genes Dysregulated by Chromosomal Translocations that Contribute to Human Leukemias and Lymphomas |Favorite Table|Download (.pdf) Table 19-1:Transcriptional Control Genes Dysregulated by Chromosomal Translocations that Contribute to Human Leukemias and Lymphomas
| Disease || Chromosomal Abnormality || Activated || Mechanism of Activation || Predominate Structural Feature * || Invertebrate Homologue † || References |
| Lymphoid Leukemia/Lymphoma |
|B-cell ALL/Burkitt ||t(8;14)(q24;q32) || MYC ||Relocation to IgH locus ||bHLHzip || ||16-18 |
|Lymphoma ||t(2;8)(p12;q24) || MYC ||Relocation to IgL locus ||bHLHzip || ||19, 20, 22, 24 |
| ||t(8;22)(q24;q11) || MYC ||Relocation to IgL locus ||bHLHzip || ||21, 23 |
|Pre-B-cell All ||t(1;19)(q23;p13) || E2A-PBX1 ||Gene fusion ||Homeodomain (PBX1) || exd (D), ceh-20 (C) ||130, 131 |
|Pro-B-cellALL ||t(17;19)(q22;p13) || E2A-HLF ||Gene fusion || bZIP(HLF) || giant (D), ces-2 (C) ||178, 179 |
|Pro-B-cellALL ||t(12;21)(p13;q22) || TEL-AML1 ||Gene fusion ||Runt homology (AML1) || runt (D) ||197-201 |
|T-cellALL ||t(8;14)(q24;q11) || MYC ||Relocation to TCRα/δ locus ||bHLHzip || ||51-53 |
| ||t(7;19)(q35;p13) || LYL1 ||Relocation to TCRβlocus ||HLH || ||48 |
| ||t(1;14)(p32;q11) || TAL1 ||Relocation to TCRα/δ locus || bHLH || ||45-47 |
| ||t(7;9)(q35;q34) || TAL2 ||Relocation to TCRβ locus || bHLH || ||47 |
| ||t(11;14)(p15;q11) || LMO1 (RBTN1) ||Relocation to TCRα/δ locus ||Cysteine-rich || ||77, 78 |
| ||t(11;14)(p13;q11) || LMO2 (RBTN2) ||Relocation to TCRα/β locus ||Cysteine-rich || ||79, 80 |
| ||t(7;11)(q35;p13) || LMO2 (RBTN2) ||Relocation to TCRβ locus ||Cysteine-rich || || |
| ||t(10;14)(q24;q11) || HOX11 ||Relocation to TCRα/δ locus ||Homeodomain || ||97-100 |
| ||t(7;10)(q35;q24) || HOX11 ||Relocation to TCRβ locus ||Homeodomain || || |
|Diffuse B-cell lymphoma ||t(3;14)(q27;q32) || BCL6 ||Relocation to IgH locus ||Zinc finger || tramtrack (D) ||212-215 |
|(large cell) ||t(3;4)(q27;p11) || BCL6 ||Relocation to TTF locus ||Zinc finger || tramtrack (D) ||213, 537 |
|B-CLL ||t(14;19)(q32;q13) || BCL3 ||Relocation to IgH locus ||IκB homology || ||538-540 |
|B-cell lymphoma ||t(10;14)(q24;q32) || LYT10 ||Relocation to IgH locus ||Rel homology || dorsal (D) ||541 |
|Lymphoplasmacytoid B-cell ||t(9;14)(p13;q32) || PAX5 ||Relocation to IgH locus ||Paired homeobox || Paired (D) ||542 |
|lymphoma || || || || || || |
| Myeloid Leukemia || || || || || || |
| AML(granulocytic) ||t(8;21)(q22;q22) || AML1-ETO ||Gene fusion ||Runt homology (AML1) || runt (D) ||226-228, 543 |
|Myelodysplasia ||t(3;21)(q26;q22) || AML1-EAP ||Gene fusion ||Runt homology (AML1) || runt (D) ||231 |
| CML(blast crisis) ||t(3;21)(q26;q22) || AML1-EVI1 ||Gene fusion ||Runt homology (AML1) || runt (D) ||230 |
| AML(undifferentiated) ||t(3;v)(q26;v) || EVI1 ||Aberrant expression ||Zinc finger || evil (D) ||322, 323 |
| AML(myelomonocytic) ||inv(16)(p13;q22) || CBFβ-MYH11 ||Gene fusion ||Complex withAML1 (CBFβ) || ||251 |
| AML(promyelocytic) ||t(15;17)(q21;q21) || PML-RARα ||Gene fusion ||Zinc finger (RARα) || ||268-272 |
| AML(promyelocytic) ||t(11;17)(q23;q21) || PLZF-RARα ||Gene fusion ||Zinc finger (RARα) || ||292 |
| AML(promyelocytic) ||t(5;17)(q32;q12) || NPM-RARα ||Gene fusion ||Zinc finger (RARα) || ||290 |
| AML(promyelocytic) ||t(11;17)(q13;q21) || NνMA-RARα ||Gene fusion ||Zinc finger (RARα) || ||291 |
| AML ||t(16;21)(p11;q22) || FUS-ERG ||Gene fusion ||Ets-like (ERG) || ||544 |
| AML ||t(12;22)(p13;q11) || TEL-MN1 ||Gene fusion ||Ets-like (TEL) || ||545 |
|Myelodysplasia ||t(3;12)(q26;p13) || TEL-EVI1 ||Gene fusion ||Zinc finger (EVI1) || evi1 (D) ||546 |
| || || || ||Ets-like (TEL) || || |
| AML(myelomonocytic) ||t(8;16)(p11;p13) || MOZ-CBP ||Gene fusion ||Zinc finger (MOZ) || Pointed (D) ||547 |
| AML(myelomonocytic) ||inv(8)(p11;q13) || MOZ-TIFZ ||Gene fusion ||CREB-binding protein (CBP) || || |
| || || || ||Zinc finger (MOZ) || ||548 |
| || || || ||Nuclear receptor coactivator || || |
| || || || ||(TIFZ) || || |
| Mixed-Lineage Leukemias ‡ || || || || || || |
|Pro-B-cellALL ||t(4;11)(q21;q23) || MLL-AF4 ||Gene fusion ||A-T hook (MLL) || trithorax (D) ||342, 344, 345 |
| AML(monocytic) ||t(9;11)(q21;q23) || MLL-AF9 ||Gene fusion ||A-T hook (MLL) || trithorax (D) ||363 |
| ALL/AML ||t(11;19)(q23;p13.3) || MLL-ENL ||Gene fusion ||A-T hook (MLL) || trithorax (D) ||341, 355, 363 |
| AML ||t(11;19)(q23;p13.1) || MLL-ELL ||Gene fusion ||A-T hook (MLL) || trithorax (D) ||355, 356 |
| AML ||t(1;11)(q21;q23) || MLL-AF1Q ||Gene fusion ||A-T hook (MLL) || trithorax (D) ||366 |
| AML ||t(1;11)(1p32;q23) || MLL-AF1P ||Gene fusion ||A-T hook (MLL) || trithorax (D) ||549 |
| AML ||t(6;11)(q27;q23) || MLL-AF6 ||Gene fusion ||A-T hook (MLL) || trithorax (D) ||389 |
| AML ||t(6;11)(q12;q23) || MLL-AF6QZ1 ||Gene fusion ||A-T hook (MLL) || trithorax (D) ||550 |
| AML ||t(10;11)(p12;q23) || MLL-AF10 ||Gene fusion ||A-T hook (MLL) || trithorax (D) ||551 |
| AML ||t(11;17)(q23;q21) || MLL-AF17 ||Gene fusion ||A-T hook (MLL) || trithorax (D) ||552 |
| AML ||t(X;11)(q13;q23) || MLL-AFX1 ||Gene fusion ||A-T hook (MLL) || trithorax (D) ||553 |
|AMML/CMML ||t(11;16)(q23;p13) || MLL-CBP ||Gene fusion ||A-T hook (MLL) || trithorax (D) ||554, 555 |
The diversity of transcription factor proto-oncogenes implicated in the human leukemias and lymphoma is striking, although increasingly their essential functions can be traced to a fundamental step in cell growth, development, or survival.8 Currently, more than 10 transcriptional control genes have been shown to play critical roles in normal hematopoiesis (Fig. 19-1). Some of these factors are lineage-specific, whereas others operate early in hematopoietic development, before lineage commitment. Still others are widely expressed but perform unique functions in a limited number of blood cell types, ostensibly by interacting with lineage-restricted proteins.4 Of major pathobiologic importance, many transcription factors that control blood cell differentiation are targets for productive rearrangement by translocations in the leukemias and lymphomas, reinforcing their roles as master regulators of hematopoietic cell development. In the following sections I summarize how chromosomal translocations modify transcription factors to generate malignant cells within the hematopoietic system.
Schematic diagram showing the relative stages at which transcription factors exert their influence on hematopoietic development. Only proteins whose activities have been demonstrated in knockout mice are shown. Factors serving as targets of chromosomal translocations in the leukemias and lymphomas are indicated in boldface type. Note that transcription factor targets can be lineage specific (E2A) or uncommitted to a particular differentiation pathway (AML1). HSC, hematopoietic stem cell; M/E, myeloid/erythroid progenitor; Ly, lymphoid progenitor; G/M, granulocyte/macrophage progenitor. (Adapted from Shivdasani and Orkin.4 Used with permission).
Acute Lymphoblastic Leukemias and Non-Hodgkin Lymphomas
The frequency distributions of the various molecular abnormalities mediated by chromosomal translocations are shown diagrammatically in Figs. 19-2 and 19-3, with key associations given in Tables 19-1 and 19-2.
Distribution of translocation-generated fusion genes among the commonly recognized immunologic subtypes of ALL in children and young adults. Key domains for DNA binding and protein-protein interaction of transcription factors are shown in boldface type; an exception is the tyrosine kinase domain indicated for BCR-ABL. The section labeled random refers to sporadic rearrangements that have so far been observed only in leukemic cells from single cases. (Adapted from Look.211 Used with permission).
Distribution of histologic subtypes of non-Hodgkin lymphoma in children and adults. Chromosomal translocations and affected genes that occur in a significant fraction (but not all) of the cases within each subtype are shown. (Adapted from Sandlund, Downing, and Crist.574 Used with permission).
Table 19-2: Tyrosine Kinase and Other Genes Dysregulated by Chromosomal Translocations in Human Leukemias and Lymphomas |Favorite Table|Download (.pdf) Table 19-2: Tyrosine Kinase and Other Genes Dysregulated by Chromosomal Translocations in Human Leukemias and Lymphomas
|Disease ||Chromosomal Abnormality ||Activated Gene ||Mechanism of Activation ||Predominate Structural Feature* ||Invertebrate Homologue+ ||References |
| Tyrosine Kinases || || || || || || |
|CMML ||t(5;12)(q33;p13) || TEL-PDGFRβ ||Gene fusion ||Tyrosine kinase (PDGFRB) || ||443 |
|Pre-B-ALL ||t(9;12)(p24;p13) || TEL-JAK2 ||Gene fusion ||Tyrosine kinase (JAK2) || Hopscotch (D) ||556, 557 |
|T-cell ALL ||t(9;12)(p24;p13) || TEL-AK2 ||Gene fusion ||Tyrosine kinase (JAK2) || Hopscotch (D) || |
|CML (atypical) ||t(9;12;14) || TEL-JAK2 ||Gene fusion ||Tyrosine kinase (JAK2) || Hopscotch (D) || |
|CMML ||t(5;7)(q33;q11,2) || HIPI-PDGFβR ||Gene fusion ||Tyrosine kinase (PDGFβR) Huntington interactin protein (HIP1) || ||558 |
|AML ||t(5;14)(q33;q32) || CEV14-PDGFβR ||Gene fusion ||Tyrosine kinase (PDGFβR) || ||559 |
|AML ||t(9;12;14)(q34;p13;q22) || TEL-ABL ||Gene fusion ||Tyrosine kinase (ABL) || abl (D) ||203, 442 |
|Anaplastic large-cell lymphoma ||t(2;5)(p23;q35) || NPM-ALK ||Gene fusion ||Tyrosine kinase (ALK) || ||457 |
|CML ||t(9;22)(q34;q11) || BCR-ABL ||Gene fusion ||Tyrosine kinase (ABL) || abl (D) ||407, 413, 416, 417 |
|ALL ||t(9;22)(q34;q11) || BCR-ABL ||Gene fusion ||Tyrosine kinase (ABL) || abl (D) ||422–424 |
|T-cell ALL ||t(1;7)(p34;q34) || LCK ||Reloaction to TCRβ locus ||Tyrosine kinase || ||560–562 |
| Other Genes || || || || || || |
|Centrocytic B-cell lymphoma ||t(11;14)(q13;q32) || Cyclin D1 ||Relocation to lgH locus ||G1 cyclin || ||493, 494, 499–502 |
|Follicular B cell lymphoma ||t(14;18)(q32;q21) || BCL2 ||Relocation to lgH locus ||Antiapoptotic domain || ced-9 (C) ||517–520 |
|AML ||t(6;9)(p23;q34) || DEK-CAN ||Gene fusion ||Nucleoporin (CAN) || ||563 |
|AML ||t(9;9)(q34;q34) || SET-CAN ||Gene fusion ||Nucleoporin (CAN) || ||528 |
|AML ||t(7;11)(p15;p15) || NUP98-HOXA9 ||Gene fusion ||Nucleoporin (NUP98) || ||529, 530 |
|AML ||t(3;5)(q35;q35) || NPM-MLF1 ||Gene fusion ||Nucleolar shuttle protein (NPM) || ||535 |
|AML ||t(10;11)(p13;q14) || CALM-AF10 ||Gene fusion ||Clathrin assembly (CLM) || Cezt (C) ||564 |
|T-cell PLL ||t(x;14)(q28;q11) || C6.1B ||Relocation to TCRαδ locus ||Unknown || ||565 |
|T-cell ALL ||t(7;9)(q34;q34) || TAN1 ||Relocation to TCRβ locus ||EGF cysteine repeats || Notch (D), lin-12 (C) ||566 |
|Pre-B-cell ALL ||t(5;14)(q31;q32) || IL-3 ||Relocation to lgH locus ||Growth factor || ||567, 568 |
|T-cell lymphoma ||t(4;16)(q26;p13) || IL2-BCM ||Gene fusion ||Growth factor || ||569 |
|T-cell PLL ||t(14;14)(q11;q32) || TCL1 ||Relocation to TCRαδ locus ||Unknown || ||570 |
| ||inv(14)(q11;q32) || TCL1 ||Relocation to TCRαδ locus ||Unknown || || |
| ||t(7;14)(q35;q32) || TCL1 ||Relocation to TCRβ locus ||Unknown || || |
|T-cell PLL ||t(X;14)(q28;q11) || MTCP1 ||Relocation to TCRαΔ locus ||Unknown || ||571, 572 |
|B-cell lymphoma ||t(11;14)(q23;q32) || RCK ||Relocation to lgH locus ||Helicase/translation initiation factor || ||573 |
MYC Activation in Burkitt Lymphoma and B-Cell Leukemia.
In Burkitt lymphoma and B-cell leukemia, arising in surface Ig-positive “virgin” B lymphoblasts with moderately abundant, vacuolated cytoplasm, the principal genetic change is a juxtapositioning of the MYC proto-oncogene next to the Ig heavy-chain gene as a result of the t(8;14)(q24;q32).16–18 MYC is a prototypic bHLH/leucine zipper transcription factor whose rearrangement from chromosome 8 to a site near strong Ig enhancer elements on chromosome 14 leads to dysregulated expression of the MYC oncoprotein. In most instances, the t(8;14) is responsible for inappropriate activation of MYC; however, two variants of this rearrangement can produce the same effect, except that they move Igκ and Igλ light chain genes from chromosome 2 and 22, respectively, to the MYC locus on chromosome 8.19–24 The MYC gene often acquires point mutations in its coding or regulatory regions, probably as a result of somatic mutation that occurs after translocation, 25–28 which in some cases encodes proteins that are unable to interact with the Rb-related gene p107. 29
A leading question since the discovery of MYC activation in Burkitt lymphoma/B-cell leukemia has been: How does the MYC oncoprotein transform B lymphocytes? The answer seems to lie in the effects of MYC dysregulation on a transcriptional network comprising at least three other factors, each also harboring bHLH/leucine zipper domains. In this cascade, MYC is able to dimerize with the MAX protein, 30,31 which can bind to DNA, to itself (MAX/MAX homodimers), and to the MAD and MXI-1 family of transcription factors.32,33 Since MYC/MAX heterodimers activate gene expression, 30,31 whereas MAD/MAX heterodimers act as trans-repressors through an association with a protein called SIN3, 34,35 and since MYC and MAD have equal affinities for MAX, 36,37 increased expression of MYC in B lymphocytes is thought to disrupt the equilibrium of MAX heterodimers, leading to untimely activation of responder genes and ultimately to malignant transformation.38 Experimental support for this hypothesis comes from the induction of B-cell neoplasms in transgenic mice carrying the MYC oncogene driven by an Ig gene enhancer.39,40 An activated MYC gene also induces tumorigenic conversion when it is introduced in vitro into B lymphoblasts infected with human Epstein-Barr virus.41 More recent observations implicate the ornithine decarboxylase gene, 42 the CDC25 cell-cycle phosphatase gene, 43 and the ARF tumor suppressor gene44 as relevant transcriptional targets of MYC/MAX heterodimers.
bHLH, LIM, And HOX11 Genes.
The role of transcription factors as the preferred targets of chromosomal translocations extends to the T-cell lymphomas and acute leukemias, in which the chromosomal breakpoints consistently appear near enhancers included in the TCR β locus on chromosome 7, band q34, or the α/δ locus on chromosome 14, band q11. Highly active in committed T-cell progenitors, these enhancers stimulate the expression of strategically translocated transcription factors that regulate early hematopoietic cell development or the development of other lineages but are not normally expressed in T lymphoid cells (see Table 19-1). Notable examples include the bHLH genes, TAL1/SCL, 45–47 TAL2/SCL2, 47 and LYL1, 48 one of which is essential for the development of all blood cell lineages (TAL1/SCL).4,49,50 The more distantly related MYC bHLH/ZIP protein is dysregulated in T-cell51–53 as well as B-cell lymphomas and leukemias.
When rearranged near enhancers within the TCR β locus on chromosome 7, band q34, or the α/δ chain locus on chromosome 14, band q11, these regulatory genes become active, and their protein products are thought to bind inappropriately to the promoter/enhancer elements of upstream target genes. The TAL1 gene, for example, is activated by the t(1;14) or by an intragenic deletion on the 5′ side of the gene that places it under the regulation of the promoter of a gene called SIL 54–58 ; these rearrangements affect up to one-fourth of all cases of childhood T-cell leukemias and lymphomas.59,60 Since the TAL1 protein can dimerize with the E2A protein through its bHLH domain to form DNA-binding complexes, 61 and with the LMO2 protein (see below), 61–65 its ectopic expression in T cells bearing the t(1;14) or activating deletions may aberrantly activate specific sets of target genes that are normally quiescent in T-lineage progenitors. It is also possible that TAL1 acts by repressing E2A activity during T-cell development, because E2A/TAL1 heterodimers are inactive as transcriptional trans-activators.66,67
Interestingly, TAL1 has emerged as an essential regulator of very early stages of hematopoietic development.68 Within the hematopoietic system, TAL1 expression is restricted to myeloid and erythroid progenitor cells, megakaryocytes, and mast cells, 69–71 and as noted previously, it is not expressed by normal T lymphocytes or their progenitors.72 Gene targeting experiments initially showed that mouse embryos lacking a functional Tal1 gene were devoid of embryonic red blood cells and died at embryonic days 9 to 10.5 of anemia.49,50 Additional studies of hematopoietic precursors generated by in vitro differentiation of Tal −/− embryonic stem cells, and by assessing the contribution of these cells in vivo to the hematopoietic systems of chimeric mice, have shown that Tal1 is required for the generation of all hematopoietic cell lineages, including T lymphocytes, suggesting that it plays an essential role in early hematopoietic development, either at the level of mesoderm induction or in maintaining the viability of multipotential hematopoietic progenitors.68 It would not be surprising if Tal1 were involved in a network of regulatory factors responsible for induction of the hematopoietic lineage, in view of the similar roles of related bHLH proteins as master regulators of mesodermal cell fate, such as those of the MyoD family (MyoD, Myf-5, MRF4, and myogenin), 73,74 which, like the Tal1 protein, form heterocomplexes with the E12/E47 products of the E2A gene. An interesting zebrafish mutant cloche affects both blood and endothelial differentiation, 75 and recent microinjection studies suggest that SCL acts downstream of cloche to specify hematopoietic and vascular differentiation.76
Other types of regulatory genes can be rearranged near TCR loci, including those encoding the LMO1 and LMO2 (for cysteine-rich LIM-domain only) proteins (also known as RBTN1/TTG1 and RBTN2/TTG2).77–80 Although T-cells normally lack expression of either protein, LMO1 is expressed in a segmental and developmentally regulated pattern in the central nervous system, 78 and LMO2 is coexpressed with Tal1 in several lineages, notably in erythroid and other hematopoietic progenitors.81 Both LMO1 and LMO2 possess zinc finger-like structures in their LIM domains82 but lack the homeobox DNA-binding domains common to other transcription factors in this family, suggesting that the LIM domain functions in protein-protein rather than protein-DNA interactions. In fact, LMO2 is coexpressed with TAL1 in several cell lineages, including erythroid progenitors, 81 and these two proteins interact to form a transcriptional complex, 64,83 both in erythroid cells and in human and murine T-cell leukemias induced by these gene products.64,65,83 The functional relevance of this complex in normal development is exemplified by the fact that gene targeting experiments in mice, in which null mutations were introduced into Lmo2, yielded the same phenotype as those described earlier for Tal1, indicating that functional complexes are required for normal primitive erythropoiesis and likely the formation of all hematopoietic lineages.68,81,84,85 Additional studies have expanded this complex to include GATA1, 86 a zinc-finger transcription factor that is also required for erythroid cell development, 87 E2A bHLH proteins, and the newly identified LIM-binding protein Ldb1/NLL, suggesting that oligomeric DNA-binding complexes containing LMO2 play important roles in hematopoiesis.88 Moreover, both LMO1 and LMO2 induce thymic lymphomas in transgenic mice whose thymocytes express these genes under the control of T-cell-specific or ubiquitously expressed promoters.89–93 Although it is controversial whether TAL1 is able to induce T-cell lymphomas in mice on its own, 94,95 it has been shown to shorten the time to development of T-cell lymphomas induced by LMO2 in a double transgenic system, apparently recapitulating the cooperativity that the these two proteins exhibit as components of multimeric transcriptional regulatory complexes in human T-cell tumors.64,65,96
HOX11 is an example of a different type of developmental gene that is inappropriately placed under the control of TCR loci. Located on chromosome 10, band q24, 97–100 this gene encodes a homeodomain transcription factor that can bind DNA and trans-activate specific target genes.101 It is most closely related to Hlx, a recently described murine homeobox gene expressed in specific hematopoietic cell lineages and during mouse embryogenesis, 102 and is distantly related to the Antennapedia homeobox genes of Drosophila, which regulate segment-specific gene expression along the anteroposterior axis of the fly embryo.103 A very specific homeotic role of Hox11 in mammalian development was demonstrated by homozygous disruption of this gene, which blocked the formation of the spleen in otherwise normal mice.104 In the mouse, Hox11 is normally expressed in specific regions of the branchial arches and ectoderm of the pharyngeal pouches of the developing hindbrain, as well as from a single site corresponding to the splanchnic mesoderm beginning at embryonic day 11.5.104 Because the nervous system develops normally in these mice, the roles of Hox11 proteins in branchial arch and hindbrain structures appear to be compensated for by other transcription factors expressed by these cells; however, the role of Hox11 in cellular organization at the site of splenic development is absolutely essential for the genesis of this organ. Further studies have shown that the splenic anlage actually develops normally in Hox11−/− mice but that the developing spleen cells undergo apoptosis, suggesting that Hox11 normally acts to promote the survival of splenic precursors during organogenesis.105 In contrast to Lmo2 and Tal1, which have important roles in hematopoietic cell development, Hox11 proteins are not normally expressed in lymphoid and other types of hematopoietic cells, and hematopoietic cells are not affected by loss-of-function mutations in this gene, except in circulating erythrocytes with asplenia-related Howell-Jolly bodies.
Activation of HOX11 by chromosomal translocations, either the t(10;14)(q24;q11) or the t(7;10)(q35;q24), in developing T cells is thought to interfere with normal regulatory cascades, thereby promoting malignant transformation. The primary oncogenic importance of aberrant expression of Hox11 in the developing thymus has been demonstrated in transgenic mice, in which this protein was redirected to the thymus, where it was associated with the development of T-cell lymphoma/leukemia at high frequencies.106 HOX11 has been shown to act as an activator of gene expression, and this activity has been shown to depend on the N-terminal 50 amino acids of the protein.107 In addition, HOX11 interacts directly with phosphatases that normally regulate a G2-phase cell cycle checkpoint, suggesting that overexpression of this protein in T-cell progenitors may cause accelerated entry into mitosis.108
The E2A gene was cloned by virtue of the fact that it encodes a protein (E12) that binds to the ΚE2 regulatory site of the Ig Κ light chain gene promoter.109 It was subsequently shown to encode three differentially spliced products, E12, E47, and E2-5, each of which belongs to the bHLH family of transcriptional regulatory proteins.109–113 The bHLH domain is comprised of a basic region responsible for sequence-specific DNA binding followed by a structural domain consisting of two amphipathic helices separated by a loop region of variable length (thus helix-loop-helix), that is responsible for homo- and heterodimerization.109,110 The bHLH family of proteins includes the daughterless Drosophila gene114,115 and members of the MyoD family of myogenic proteins.73,74 DNA-binding by E2A is mediated by either homodimers or heterodimers with other bHLH proteins, with the precise binding specificity to variations of the so-called E-box sequence motif determined by the dimerization partners of each complex.116 Recent structural analysis has supported experimental observations regarding the conformations of homo- and heterodimers formed by the E2A bHLH domains.117 In addition, the N-terminal sequences of E2A that are included in leukemogenic fusion proteins (Fig. 19-4) have been shown to contain two discrete transcriptional activation domains, called AD1 and AD2, 112,118,119 the latter of which is also referred to as a loop-helix (LH) activation domain.
Comparison of the structural features of two major E2A fusion proteins. The E2A portions of the chimeras are identical, retaining both the AD1 and AD2 transcriptional activation domains. The PBX1 fusion partner retains its DNA- and protein-binding domain (homeodomain), as does HLF (bZIP), providing a mechanism for recognition and activation of downstream target genes. Despite the normally wide distribution of E2A and the lack of normal expression of the HLF and PBX1 transcription factors in hematopoietic cells, the two chimeras act specifically on B-cell precursors. HD, homeodomain; bZIP, basic leucine zipper; bHLH, basic helix-loop-helix; Ch, chromosome.
In most tissues, E2A heterodimerizes with tissue-specific bHLH family members to coordinate gene expression during development.110,120 These binding partners include TAL/SCL and LYL1 family members, which heterodimerize with E2A62,121,122 and are themselves dysregulated in T-cell lymphomas/leukemias and aberrantly expressed as a result of translocations involving the TCR gene loci.45–47 In B cells, however, E2A is able to bind E-box sequences as a homodimeric complex, 120,123 apparently due to stabilization of the complex through an intermolecular disulfide bond, which is disrupted in non-B cells.124 The importance of E2A proteins in B-cell development is indicated by the fact that homozygous mutant mice lacking functional E2A proteins have arrested B-cell development at an early stage.125,126 Mice deficient in E2A not only lack pro-B cells but also show defects in T-cell development and acquire T-cell malignancies, suggesting that loss of function of E2A may contribute to leukemogenesis in T cells, in addition to its role as a component of heterodimeric complexes with other bHLH proteins.127,128
The E2A gene participates in two fusion events with major biologic and clinical implications in acute lymphoblastic leukemia (ALL). The first results from the t(1;19)(q23;p13) chromosomal translocation, which rearranges and joins the E2A gene within chromosome band 19p13.3 to the PBX1 gene from chromosome 1, creating an E2A-PBX1 chimera on the derivative chromosome 19129–131 (see Fig. 19-2). Because the breakpoints in the E2A gene consistently interrupt the ∼3.5-kb intron between exons 13 and 14, the encoded E2A fusion partner invariably consists of the N-terminal two-thirds of the molecule, which includes two transcriptional activation domains (AD1 and AD2), but not the bHLH DNA-binding/protein-interaction domain.130–132 The PBX1 segment makes up for this deficit by providing a homeodomain motif of ∼60 amino acids that enables the E2A-PBX1 chimera to function as a transcription factor, driven by the potent E2A trans-activating domains.133–137
An understanding of the likely oncogenic contribution of the PBX1 fusion partner requires insight into the normal function of PBX proteins. These transcription factors are the mammalian homologues of the Drosophila protein extradenticle.138,139 Mutations in the exd gene cause homeotic transformations, changes in which one body segment of the fly is transformed to resemble another segment.140,141 Thus the extradenticle protein may function as an obligatory cofactor in selector gene activity by forming complexes with major homeotic fly proteins of the Antennapedia and Bithorax clusters, termed Hom, which then bind to DNA.142–145 In view of the close sequence homology shared by extradenticle and PBX proteins, it is perhaps not surprising that the latter interact with specific human homologues of the Hom family, called HOX proteins, to determine the target genes recognized by PBX1.142,144,146,147,147–155
Given that E2A-PBX1 carries the transcriptional activation domains of E2A and the homeodomain of PBX1, how does the chimera induce malignant transformation? When Kamps and Baltimore156 infected bone marrow progenitors with retroviruses encoding E2A-PBX1, they reproducibly induced acute myeloid leukemia (AML) in mice repopulated with these progenitors. These myeloid leukemia cells could proliferate for extended periods without maturation so long as they received granulocyte-macrophage colony-stimulating factor (GM-CSF).157 In the absence of growth factor, the cells died rapidly. These observations are consistent with the block of differentiation characteristic of lymphoid cells carrying the t(1;19) in cases of ALL 158,159 and with arrested T-cell development in lymphomas of E2A-PBX1 transgenic mice.160 Thus a major effect of the chimera may be to arrest hematopoietic and lymphoid progenitors at particular stages of development.
Additional studies with cell transformation assays have established the specific E2A-PBX1 domains required for malignant conversion.161 When either of the two transactivation domains of E2A are abolished, there is a loss or reduction of transforming activity. The shortest PBX1 sequence required for oncogenesis includes the homeodomain and its immediately C-terminal 25 amino acids, which also are needed for interaction with specific HOX proteins.146,162 Unexpectedly, mutant proteins with deletion of the homeodomain and retention of the adjacent flanking region transformed NIH-3T3 cells and induced lymphomas in transgenic mice as efficiently as the full-length chimera.161,162 Other investigators have confirmed the dispensibility of sequence-specific DNA binding for transformation of fibroblasts while showing that the PBX1 homeodomain is essential for efficient arrest of myeloid cell differentiation.163 These observations suggest that interaction with members of the HOX family of proteins is sufficient to target the E2A-PBX1 fusion protein to downstream target genes with critical functions in cell transformation but not those which interfere with normal differentiation programs.161,163 Recent studies have implicated members of related homeodomain families, including Meis1 and pKnox1, as important binding partners and potentially important functional modulators of PBX and HOX proteins.164–166 Interestingly, these proteins interact through a region of PBX1 that is disrupted in the E2A-PBX1 chimera, suggesting its transforming potential may be augmented by loss of Meis-Knox interactions, which normally may influence target gene recognition, transcriptional properties, or nuclear import of PBX1.164,165 E2A-PBX1 is one of the most common fusion genes in children with ALL, occurring in 20 to 25 percent of cases with a pre-B immunophenotype (defined by cytoplasmic but not surface expression of Ig genes).158,167,168 It is also detected in adults with ALL, as well as occasional cases of pro-B-cell ALL, AML, T-cell ALL, and lymphoma.14,167–176 Patients with pre-B ALL and the t(1;19) tend to have elevated leukocytes at diagnosis and central nervous system leukemia.14,167,168 Aside from the adverse impact of these features, the E2A-PBX1 fusion gene was shown to be independently associated with a poor prognosis, 167 although in recent years intensive chemotherapy has improved clinical outcome significantly in these patients.168 A prudent clinical management strategy for patients with pre-B ALL is to consider the E2A-PBX1 fusion gene a high-risk biologic feature that warrants an aggressive approach to therapy. Otherwise, these patients may be undertreated with consequent rapid development of drug-resistant disease.
A second E2A fusion gene is created by the t(17;19)(q21-q22;13) rearrangement, 177 which joins E2A to the HLF gene within chromosome band 17q21-22178,179 (see Fig. 19-2). The breakpoint of this translocation consistently leaves the same portion of HLF in the chimeric gene but affects either intron 12 or 13 within E2A. The resulting hybrid protein is therefore termed type I (intron 13 breakpoint) or type II (intron 12 breakpoint), 180 although these structural distinctions do not appear to affect the DNA-binding and transcriptional regulatory properties of E2A-HLF.181
The HLF (hepatic leukemia factor) component of the chimera is a novel bZIP transcription factor within the PAR subfamily of proteins (defined by a proline- and acidic amino acid—rich domain).182–185 HLF recently has been shown to encode two proteins from alternatively spliced transcripts that are regulated by different promoters.186 One isoform is abundant in brain, liver, and kidney, whereas the other is restricted to hepatocytes; these proteins accumulate with different circadian patterns in the liver and have distinct promoter preferences in trans-activation experiments. Very little is known about the normal function of the PAR proteins, including HLF, but their structural similarity with the CES-2 bZIP protein that orchestrates the death of sertoninergic nerve cells in the developing worm Caenorhabditis elegans suggests a regulatory role in cell survival, 187–189 as indicated by the mechanism of E2A-HLF oncogenic activity, described below.
The E2A-HLF fusion product retains the entire DNA-binding/protein-protein interaction domain of HLF, as well as the two N-terminal transactivation domains of E2A.179,180 In leukemic lymphoblasts, the chimeric protein appears to bind DNA as a homodimer, 190,191 as one might predict given the absence of detectable levels of the known normal PAR proteins in hematopoietic precursors. Like E2A-PBX1, the E2A-HLF oncoprotein can transform NIH-3T3 cells, depending on the integrity of the HLF leucine zipper and the E2A transcriptional activation domains.192 It also induces lymphoid tumors in transgenic mice.193 However, E2A-PBX1 induces apoptosis in hematopoietic cells through a p53-independent mechanism that requires the DNA-binding homeodomain of PBX1, 194 which is the direct opposite of the effect of the conditional expression of E2A-HLF.
Analysis of the effects of E2A-HLF on cell survival has provided important insight into how E2A-HLF might take control of immature lymphoid cells. When introduced into leukemic cells carrying the t(17;19), a dominant-negative form of E2A-HLF blocked the usual action of the intact chimera, and as a result, themalignant cells underwent apoptosis.187 By contrast, the dominant-negative mutant had no effect on apoptotic events in leukemic cells without the t(17;19), suggesting that E2A-HLF may increase the number of developing lymphocytes by preventing their suicide.187 The homology between HLF and the CES-2 protein of C. elegans, 188 which functions early in a genetically controlled cell death pathway, suggests that a comparable pathway operates in human B lymphoblasts and is usurped by E2A-HLF to give rise to ALL.187 In this model (Fig. 19-5), E2A-HLF activates a downstream target gene that is normally repressed by a CES-2-like protein so that cell survival rather than cell death signals ensue. Thus the leukemogenic activity of E2A-HLF may operate through an evolutionarily conserved pathway that determines the sensitivity of specific lymphoid cells to apoptotic stimuli.
A proposed model for the anti-apoptotic role of E2A-HLF in leukemogenesis. Leukemic cells with the t(17;19) undergo programmed cell death when E2A-HLF is inhibited through a dominant negative mechanism, suggesting that the primary effect of the hybrid oncoprotein is to prolong cell survival rather than to accelerate cell growth.187 The close homology of the HLF bZIP domain to that of the CES-2 cell death-specification protein of the nematode C. elegans 188 suggests that E2A-HLF may contribute to leukemogenesis by binding to the promoters of target genes normally regulated by a mammalian ortholog of the CES-2 protein, which causes defective pro-B cells to undergo apoptosis. According to this model, E2A-HLF may activate target gene expression in contrast to the proposed repressor effects of CES-2, leading to aberrant survival through an evolutionarily conserved pathway that regulates programmed cell death during B-lymphoid cell development.
The t(17;19) defines a subset (0.5 to 1 percent) of ALL patients with a pro-B immunophenotype.177 In several reports this re-arrangement was linked to disseminated intravascular coagulation (DIC) and hypercalcemia at initial diagnosis.177,179,180,195,196 Although the rarity of t(17;19)-positive ALL has hampered efforts to assess its prognostic significance, each of seven patients withmolecularly identified E2A-HLF fusion died of leukemia despite their enrollment on contemporary treatment protocols.180,190,195,196 Drug resistance in this type of leukemia may be augmented by the role of E2A-HLF in preventing accelerated apoptosis from therapy-induced DNA damage as well as growth factor deprivation.187
Generally considered a target of chromosomal translocations in myeloid cells, the AML1 gene is joined to a second transcriptional control gene, called TEL, as a result of the t(12;21) in cases of B-lineage ALL.197–201 Although rarely detected by routine karyotyping (because the telomeric segments of 12p and 21q appear similar in banded metaphase preparations), the t(12;21) rearrangement is apparent by fluorescence in situ hybridization in approximately one-fourth of children with ALL, making TEL-AML1 the most common genetic abnormality in the lymphoid leukemias.199 The TEL-AML1 fusion product consists of the bHLH domain of TEL linked to virtually the entire coding region of AML1, including the DNA- and protein-binding domain, which bears close amino acid identity to the Runt protein of Drosophila. The exact role of the TEL-AML1 oncoprotein in cell transformation remains unclear, but emerging data suggest that the primary effect relates to a compromise of AML1 transcriptional activity, 202 which is required for normal hematopoiesis (see the section in this chapter on the involvement of the AML1-CBFβ complex in the acute myeloid leukemias).
The TEL gene is also involved in multiple other fusion genes associated with chronic myelomonocytic leukemia (CMML) (TEL-PDGFRβ), AML (TEL-MN1, TEL-ABL, TEL-EVI1), and ALL (TEL-JAK2) (see Tables 19-1 and 19-2). TEL harbors a 65-amino acid helix-loop-helix dimerization motif that is conserved in a subset of the ETS family of proteins, and this region appears to an essential requirement for constitutive tyrosine kinase and transforming activity of the activity of the TEL-PDGFRβ and TEL-ABL fusion proteins.203–205 TEL-AML1 also appears to dimerize with itself and with normal TEL proteins in the cell, and there is often associated loss of the normal TEL allele in leukemias with TEL-AML1 fusion genes, suggesting that loss of function may contribute to oncogenicity.204,206 Tel has been homozygously disrupted in mice through gene targeting, and interestingly, the Tel-deficient mice die at approximately embryonic day 11 with defective yolk sac angiogenesis and intraembryonic apoptosis of mesenchymal and neural cells.207
The TEL-AML1 fusion gene is associated with a superior treatment outcome in patients with B-lineage ALL, and relapse-free survival has approached 90 percent on several different therapeutic regimens.204,206,208–210 For example, in a recent trial, children with TEL gene rearrangements (primarily TEL-AML1) had a 5-year event-free survival probability of 91 ± 5 percent (SE) compared with 64 ± 5 percent for those with TEL in a germ-line configuration.208 The prognostic strength of TEL rearrangement (usually as a TEL-AML1 fusion gene) was independent of recognized good-risk features in ALL with B-lineage markers, such as the presenting leukocyte count and hyperdiploidy. Indeed, molecular detection of the TEL-AML1 fusion gene is the first genetic assay to allow a good-risk subset of patients to be dissected from the otherwise high-risk “pseudodiploid” subset of ALL patients.211 Thus TEL-AML1 has been added to the list of genetic abnormalities requiring recognition early in the disease course (Table 19-3).
Table 19-3: Clinical Risk Assignment in the Childhood Leukemias by Genetic Classification of the Malignant Cells |Favorite Table|Download (.pdf) Table 19-3: Clinical Risk Assignment in the Childhood Leukemias by Genetic Classification of the Malignant Cells
|Abnormality (Risk) ||Method of Detection ||Treatment |
|Hyperdiploidy, ≥53 chromosomes (good risk) ||DNA flow cytometry ||Antimetabolite therapy emphasizing high-dose methotrexate |
| TEL-AML 1 fusion human gene due to t(12;21) (good risk) ||RNA PCR to detect TEL-AML 1 fusion transcripts ||Antimetabolite therapy emphasizing high-dose methotrexate |
| E2A-PBX 1 fusion human gene due to t(1:19) (intermediate risk) ||RNA PCR to detect E2A-PBX 1 fusion transcripts ||Intensified chemotherapy with alkylating agents and topoisomerase inhibitors |
| MLL fusion gene due to 11q23 rearrangements and E2A-HLF due to t(17:19) (high risk) ||RNA PCR to detect MLL and E2A-HLF fusion transcripts ||Experimental forms of intensified chemotherapy or bone marrow transplantation |
| BCR-ABL fusion gene due to t(9:22), with high leukocyte count (ultra-high risk) ||RNA PCR to detect BCR-ABL fusion transcripts ||Bone marrow transplantation in first remission |
BCL6 Activation in Diffuse Large-Cell Lymphoma.
The t(3;14)(q27;q32) and related translocations affect the long arm of chromosome 3 in diffuse large cell lymphomas of the B-cell lineage, leading to the discovery of the BCL6 proto-oncogene, whose expression is altered in at least 30 percent of these malignancies, the vast majority of which occur in adults.212–215 BCL6 encodes a transcription factor containing six zinc-finger DNA-binding motifs near the C-terminus and a POZ regulatory domain near the N-terminus. It is related to the PLZF protein that is fused to RARα as a result of the t(11;17) translocation of acute promyelocytic leukemia. (The postulated developmental roles of highly conserved zinc-finger proteins with POZ domains are discussed in the PLZF section of this chapter.) Like the AML1-CBFβ complex in the myeloid cell lineage, but unlike most genes whose expression is altered by translocation to the vicinity of the Ig or TCR genes, BCL6 is normally expressed and developmentally regulated in cells of the same lineage in which it is linked to transformation, the B lymphocytes.216,217 The BCL6 protein is detected in cells of the lymph node germinal center, a region in which antigen-primed B cells normally undergo transformation into either memory B cells or immunoblasts destined to become plasma cells or die as a result of apoptosis.218 Because BCL6 is normally down-regulated before B cells exit from the germinal center, a reasonable hypothesis is that activated B lymphoblasts constitutively expressing BCL6 are unable to develop normally and instead replicate clonally with the considerable proliferative capacity of a large-cell lymphoma of activated B-lymphocyte origin.219 This interpretation is supported by the fact that most BCL6 rearrangements occur within the 5′-noncoding first exon or the first intron of the gene and result in dysregulation of expression of a structurally intact BCL6 protein.220
In addition to gene rearrangement mediated by chromosomal translocation, somatic point mutations of the 5′ regulatory regions of the BCL6 gene have been identified at high frequency in both the diffuse large cell and the follicular lymphomas of B-cell origin, suggesting that dysregulated expression of BCL6 may be linked casually to malignant transformation in high percentages of lymphoid tumors of these pathologic subtypes.221 Rearrangements of the BCL6 gene have been shown to have distinct clinicopathologic correlates within the adult diffuse large-cell lymphomas, occurring primarily in extranodal tumors that have not spread to the bone marrow. Importantly, they independently identify a subset of patients with a favorable prognosis.222
The distribution of gene rearrangements due to chromosomal translocations in AMLs of children and adolescents is shown in Fig. 19-6.
Distribution of translocation-generated fusion genes among the various morphologic subtypes of AML in children and young adults. The section labeled random refers to sporadic rearrangements that have so far been observed only in the leukemic cells from single cases. Key domains for DNA binding and protein-protein interaction are given for transcription factors or the type of gene affected for nontranscription factors.
Gene Rearrangements Affecting the AML1-CBFβ Complex.
The AML1/CBFβ transcription factor complex (Fig. 19-7) is the most frequent target of chromosomal translocations in the human leukemias, in that one of these linked proteins is expressed as an oncogenic chimera in as many as one-third of both ALL and AML patients. This regulatory complex, termed CBF because of its identity as a core binding factor (also known as PEBP2), 223 consists of a DNA-binding subunit, AML1 (also called CBFα2 or PEBP2αB), and CBFβ (also called PEBP2β), a subunit that does not bind DNA independently but rather heterodimerizes with AML1 or one of its closely related family members.224,225 Chromosomal translocations that modify the AML1/CBFβ complex in myeloid cells include the t(8;21), which generates AML1-ETO, 226–229 and the t(3;21), which gives rise to AML-EVI1, AML-EAP, or AML1-MDS1. 230,231
Molecular consequences of chromosomal rearrangements that modify the AML1/CBFβ transcription factor complex, the most frequent target of reciprocal translocations in the human leukemias. In the majority of cases, the structural alteration disrupts the AML1 DNA-binding partner of this complex but not CBFβ, whereas in cases with the inv(16), only the latter protein is affected. The lack of lineage specificity for genetic lesions involving AML1 can be appreciated from the very early site of action of this gene in normal hematopoiesis (see Fig. 19-1), but the molecular basis for the phenotype specificity of each fusion gene in the transformation of myeloid or lymphoid progenitors remains unknown. CML, chronic myeloid leukemia; MDS, myelodysplastic syndrome; AML, acute myeloid leukemia; ALL, acute lymphoblastic leukemia. (Adapted from Shurtleff et al.199 Used with permission).
The sequence-specific DNA-binding and protein-protein interaction properties of CBF fusion proteins are provided by a large domain within AML1, 232 showing approximately 70 percent homology with the Drosophila Runt and Lozenge proteins. The Drosophila AML1 homologues participate in several developmental processes, including sex determination, segmentation and neurogenesis (Runt), and determination of photoreceptor identity during eye development (Lozenge). The sequence element recognized by AML1 is TGTGGT, 232 an enhancer core motif that serves as a regulatory element in several viral enhancers, as well as genes whose products are involved in the regulation of hematopoiesis, such as IL-3, GM-CSF, CSF-1, myeloperoxidase, and the TCR receptors.224,233–240 The binding affinity of AML1 is markedly increased through its heterodimerization with CBFβ, an interaction also mediated through the runt homology domain.232
The Aml1 gene was inactivated recently in the germ line of mice by homologous recombination and shown to be essential for definitive hematopoiesis of all lineages.241,242 Homozygous null animals display normal morphogenesis and yolk sac-derived erythropoiesis but die between embryonic days 11.5 and 12.5 because of CNS hemorrhage, postulated to be caused by a lack of platelets and possibly potentiated by abnormalities of CNS capillary endothelium.242 Inactivation of the Cbfβ gene in the mouse germ line produced similar effects in homozygous null mice, indicating that CBFβ is required for AML1 function in vivo.243 From these observations it appears that the AML1/CBFβ complex is an essential regulator of genes required for normal hematopoietic cell development. Hence chromosomal rearrangements that target this complex may interfere with its function in ways that produce arrested differentiation and eventually fully transformed leukemias of specific cell lineages.
The t(8;21), resulting in expression of the AML1-ETO oncoprotein, is the most frequent chromosomal abnormality in the myeloid leukemias of both children and adults; it is found most often in myeloblasts with evidence of granulocytic differentiation (M2 designation by the French-American-British classification system). The fusion protein, which retains the runt homology domain of AML1 and its ability to interact with CBFβ and the core enhancer DNA sequence element, appears to interfere with AML1-mediated transcriptional activation.232,244 In fact, the C-terminal portion of ETO that is fused in frame with AML1 sequences has been shown to dominantly repress the expression of promoters normally activated by AML1.240,245,246 The role of ETO in transcriptional repression appears to be directly linked to the ability of sequences included in the fusion protein to recruit the nuclear corepressors N-CoR and mSIN3.247,248 These proteins assemble in a complex with histone deacetylase, which results in nucleosome assembly and the silencing of gene expression.249 Thus a biochemical mechanism has been identified that sheds light on the ability of the oncogenic AML1-ETO fusion protein to dominantly oppose the activity of AML1 in the regulation of genes essential for normal myeloid cell development.
The combinatorial versatility of the AML1 locus is demonstrated by its fusion with sequences from either the EVI1 gene in t(3;21)-positive chronic myeloid leukemia in blast crisis230 or to either the EAP (Epstein-Barr virus RNA-associated protein) or MDS1 genes in myelodysplastic syndrome.231 EAP and MDS1 are located in a region adjacent to EVI1 on the long arm of chromosome 3 and are often included with EVI1 in transcripts resulting from these rearrangements.250 Inclusion of both the Runt-homologous DNA-binding/dimerization domain of AML1 and the zinc-finger DNA-binding domains of EVI1 in the AML1-EVI1 chimeric protein affords ample opportunity for aberrant regulation of target genes.
The CBFβ subunit is involved in another major chromosomal rearrangement in AML, the inversion 16, which affects 15 to 18 percent of AML patients, principally those with myelomonocytic differentiation and increased bone marrow eosinophils (M4-Eo designation in the French-American-British system). This rearrangement joins most of the CBFβ gene to the C-terminus of the heavy chain gene of smooth muscle myosin (MYHII, also known as SMMHC), resulting in formation of a CBFβ-MYH11 protein.251 Significantly, the fusion protein retains the domain of CBFβ that mediates heterodimerization with AML1.252,253
Murine models to study the effects of CBFβ-MYHII on hematopoietic cell development have been produced by inserting the human MYHII cDNA in-frame into the mouse Cbfb gene through homologous recombination to “knock in” the fused gene.254 Similar experiments generated mice in which the Aml1-ETO fusion gene has been reconstructed by introducing the appropriate segment of the ETO gene into the mouse Aml1 genomic locus.255,256 Mouse embryos harboring one allele of Cbfb-MYHII or Aml1-ETO in the germ line developed CNS hemorrhages at embryologic days 12.5 to 13.5, similar to mice with homozygous loss of Aml1 or Cbfb, indicating that these chimeric proteins dominantly interfere with essential functions of the Aml1/Cbfb complex. Mice expressing Cbfb-MYHII had impairment of primitive as well as definitive hematopoiesis, however, suggesting an additional activity of this fusion protein during hematopoietic cell development.254 In addition, cells from the fetal livers of embryos expressing the Aml1-ETO fusion contained dysplastic multilineage hematopoietic progenitors that had an abnormally high self-renewal capacity and could be established as cell lines in vitro.255,256 Since both AML1 and CBFβ normally are required for definitive hematopoiesis, the oncogenicity of their respective fusion proteins may stem from disruption of a transcriptional regulatory complex producing arrested myeloid cell differentiation; however, the basis for the phenotypic specificity of leukemias resulting from various types of AML1 and CBFb fusion proteins is unknown (see Fig. 19-7). The available data imply unique activities for each chimeric protein, possibly including gain-of-function as well as loss-of-function effects, as well as global interference with the role of the heterodimeric complex.
Although AML therapy has improved during the past decade, this disease remains extremely difficult to treat with chemotherapy alone, and much of the improvement in survival has arisen from advances in hematopoietic stem cell transplantation. Clinical studies have now demonstrated, however, that the presence of either the AML1-ETO or CBFβ-MYH11 fusion genes in leukemic blast cells at diagnosis will identify patients with a relatively favorable prognosis, especially when treatment consists of intensive chemotherapy including high-dose cytarabine.12,257–263 The clinical impact of this favorable association with therapeutic outcome is enhanced by the high relative frequency of fusion genes involving the CBF complex in the overall patient population with AML, which approaches approximately one-third of newly diagnosed children and adults with this disease. Molecular analysis of samples taken after the initiation of therapy also has led to the rather surprising finding that both the AML1-ETO and the CBFβ-MYH11 fusion mRNAs can persist in the bone marrow and peripheral blood of AML patients in long-term remission following chemotherapy or bone marrow transplantation.264–267 These observations illustrate the impact that a more comprehensive understanding of the genetic basis of acute leukemia is having an on clinical management and highlight the need for further work to explain the mechanisms that underlie the intriguing correlations that are emerging between molecular findings and therapeutic response.
Retinoic Acid Receptor Rearrangements
Dysregulated chimeric transcription factors, which induce differentiation arrest at specific stages of development in the myeloid leukemias, offer a new class of intracellular targets for therapeutic attempts to promote differentiation of these leukemias in vivo so that they lose their proliferative capacity. A major example is the fusion product generated by the t(15;17)(q21;q11-22) in acute promyelocytic leukemia (APML), which links critical ligand- and DNA-binding sequences of the retinoic acid receptor-α gene (RARα) on chromosome 17 to sequences of the PML gene on chromosome 15.268–273 In its unaltered form, the RARα protein binds to the retinoic acid ligand through a defined ligand-binding domain and to DNA through a separate zinc-finger region as a heterodimer with retinoid X receptor protein.274 PML proteins, which also possess zinc-finger motifs, are normally located in novel macromolecular nuclear organelles, called PML oncogenic domains (PODs), that include at least three other proteins.275–277 These nuclear bodies are preferential targets of proteins expressed by DNA tumor viruses278–280 and are up-regulated in activated inflammatory mononuclear cells and by interferon.281–283 The PML-RARα fusion proteins disrupt these subnuclear structures, causing normal PML, RXR, and other nuclear proteins to disperse in an abnormal microparticulate pattern.275–277 The fusion proteins interfere with normal myeloid cell development, possibly through adverse effects on assembly of the PODs that contain PML, and dominant inhibitory effects as a homodimeric complex with normal retinoid receptor and peroxisome-proliferator pathways, 284–287 leading to arrested differentiation in the promyelocyte stage. PML-RARα also has been shown to have antiapoptotic activity and to result in cell survival under conditions of growth factor deprivation, which may contribute to its leukemogenic activity.285,288,289
PLZF-RARα, NPM-RARα, and NuMA-RARα.
Three variant translocations have been identified in AML that unequivocally implicate RARα in leukemias arrested at the promyelocyte stage of differentiation, since both fusion proteins involve the retinoid and DNA-binding domains of this nuclear receptor. Very little is known about the NPM-RARα or NuMA-RARα fusion proteins, which have been identified only in rare patients.290,291 NPM-RARα links RARα in-frame to N-terminal sequences of nucleophosmin (NPM), a nucleolar shuttle protein that is also involved in NPM-ALK fusion proteins in large-cell lymphoma and NPM-MLF1 in AML, while the NuMA-RARα fusion protein represents a similar in-frame fusion with a protein involved in the nuclear mitotic apparatus.
PLZF-RARα was first recognized several years ago, 292 with subsequent structural and functional studies of PLZF providing intriguing insights into the potential mechanism of action. PLZF is a transcription factor containing nine C-terminal zinc-finger motifs related to those of the Krüppel Drosophila segmentation protein and containing an N-terminal POZ (poxvirus and zinc-finger) protein-protein interaction domain.293 This domain inhibits the binding of transcription factors, including RARα, to DNA when linked in cis, suggesting that PLZF-RARα may act by sequestering RXR or other retinoid receptors within inactive multimeric complexes.293–296 PLZF is expressed in multiple tissues during development, including elevated expression at Rhombomeric boundaries in the vertebrate hindbrain.297 It is also expressed by early hematopoietic progenitors with a punctate nuclear distribution and is down-regulated during myeloid cell differentiation.298 These findings, combined with studies showing heterodimerization of PLZF-RARα with normal PLZF through the POZ domain, 295 suggest that normal PLZF also may play a role in normal hematopoietic cell differentiation, one that is inhibited by the fusion protein. Five additional cases of APML with t(11;17)-mediated expression of PLZF-RARα fusion proteins have been reported, 299 indicating that these patients share a proclivity with PML-RARα patients to develop life-threatening DIC.
The Histone Deacetylase Complex and its Role in APML.
Recent studies have provided an attractive model to explain the mechanism through which the PML-RARα and PLZF-RARα fusion proteins contribute to dsyregulated gene expression in APML. In the absence of ligand, RARs have been shown to repress the expression of target genes. The mechanism involves the recruitment of the NCOR and SMRT corepressors, which in turn mediate the assembly of an histone deacetylase complex that has the ability to silence gene expression.300–305 The PML-RARα fusion protein retains its ability to interact with the RAR corepressors and block transcription in the absence of ligand. However, unlike normal RAR proteins, which release the repressor complex and function as activators of gene expression in response to physiological concentrations of retinoids, the fusion protein remains in a repressor complex and aberrantly blocks target gene expression.306–310 Studies in the presence of higher levels of retinoids also have helped explain the responsiveness of APMLs harboring PML-RARα fusion genes to ATRA. Pharmacologic dosages of ATRA overcome the association of PML-RARα with the histone deacetylase complex and allow the recruitment of coactivators, resulting in the activation of expression of critical target genes and the induction of growth arrest with granulocytic differentiation within the malignant clone.311–314 These findings provide a mechanistic rationale for use of ATRA acid in patients with APML, which had been shown already to be effective in empirically initiated trials.311,313,315–317 In response to pharmacologic doses of this compound, PML and its associated proteins are reorganized into normal-appearing nuclear PODs, with subsequent maturation of the leukemic cells into differentiated myeloid cells with limited life-spans in the circulation. Resistance to ATRA as a single agent generally develops within 3 to 4 months, but its role in the remission induction phase of APML therapy has been established, and clinical trials combining ATRA with cytotoxic chemotherapy have led to improved survival of patients whose promyeloblasts express the PML-RARα fusion protein.316,317
Biochemical analysis of the association of histone deacetylase complexes formed with PLZF-RARα fusion proteins also has provided an explanation for the clinical observation that all-trans retinoic acid differentiation therapy is not effective in inducing remissions in patients with this variant RARα fusion protein. The mechanism of transformation appears to be quite similar, in that PLZF-RARα proteins heterodimerize with RXR and form repressor complexes that block target gene expression in a fashion unresponsive to physiological retinoid levels.300–305 However, the POZ domain of the PLZF portion of the fusion protein independently recruits the SMRT and NCOR nuclear corepressors, in complexes that are not disrupted in the presence of high levels of ATRA. Thus the presence of a second ATRA-unresponsive histone deacetylase complex formed by the PLZF fusion partner provides an explanation for the lack of sensitivity of leukemias harboring this fusion protein to treatment with a ligand that specifically overcomes repression mediated through the RARα moiety of the fusion protein.
The oncogenic properties of PML-RARα have been studied in transgenic mouse models in which expression of the fusion protein is driven by CD11b or cathepsin G regulatory sequences.318–320 Mice expressing PML-RARα driven by the cathepsin G promoter develop a myeloproliferative disorder, and 25 to 30 percent of the mice develop AML after a relatively long latency period of 6 to 15 months of age.307,319,320 By contrast, PLZF-RARα was much more active when its expression was driven by the same promoter, inducing leukemia in all the mice from two lines followed for the same time period, and leukemias in the mice were refractory to pharmacologic dosages of retinoic acid, recapitulating the resistance of human PLZF-RARα leukemias to ATRA therapy.307 A transformation model, based on retroviral transduction of the PML-RARα gene into hematopoietic progenitor cells of chickens also has been used to demonstrate leukemogenicity.321
In some cases of AML with high platelet counts, the inv(3)(q21;q26.2) or the t(3;3)(q21;q26.2) moves promoter/enhancer sequences from one site on chromosome 3 into the EVI1 locus on the same chromosome, 322,323 leading to increased gene expression. The same effect is produced in murine myeloid leukemias by insertional mutagenesis.324 The EVI1 protein binds to promoter/enhancer sequences containing the GATA sequence motif and may act by interfering with regulatory signals normally mediated by the GATA family of hematopoietic transcriptional regulators.325–328 The normal function of EVI1 is unknown, although its tissue distribution (oocytes and kidney cells) and its dominant interfering effect on normal myelopoiesis would suggest an important developmental role in regulatory pathways that interface between proliferation and differentiation.
Acute Mixed-Lineage Leukemias: MLL Fusion Genes
An extraordinarily diverse group of chromosomal translocations, deletions, and inversions affect chromosome band 11q23. In contrast to the lineage specificity of many other nonrandom rearrangements, these abnormalities occur in both lymphoid and myeloid leukemias (7 to 10 percent of ALL patients, 5 to 6 percent of AML) and in a high percentage of the so-called mixed-lineage leukemias, defined by expression of markers of more than one hematopoietic cell lineage.329–331 Leukemias with 11q23 translocations also account for a high percentage of acute leukemias in infants less than 1 year of age (80 and 45 percent of infants with ALL and AML, respectively).332–337 Perhaps the most striking association is the presence of 11q23 translocations in as many as 85 percent of secondary leukemias in patients treated with topoisomerase II inhibitors.338,339 Taken together, these examples of phenotypic diversity suggest that 11q23 genetic abnormalities mediate the transformation of multipotential hematopoietic stem cells, which give rise to leukemias in which the myeloid or lymphoid progenitors are blocked at various stages of development.
Molecular Biology of MLL.
Cloning of the gene most often affected by 11q23 abnormalities fulfilled expectations based on phenotypic, cytogenetic, and clinical studies. Many of the breakpoints that occur within the 11q23 locus interrupt the mixed-lineage leukemia gene (MLL, also called HRX, ALL-1, and HTRX1), which encodes a large protein of 3968 amino acids with a predicted molecular mass of 431 kDa.340–345 Most intriguing with regard to function are three regions of homology with the Drosophila trithorax (trx) gene, two associated with central zinc-finger domains and the third with a 210-amino acid C-terminal region of 61 percent identity called the SET (Suvar3-9, Enhancer of zeste, Trithoroax) domain.341,342,346–348 Trithorax is a master homeotic gene regulator that positively regulates the actions of a wide spectrum of homeotic (Hom) genes in the Antennapedia and Bithorax complexes of the fly and is required throughout embryogenesis for normal development of the head, thorax, and abdomen.341–344,349
The N-terminal region of the MLL protein contains three A-T hook domains, first identified in the so-called HMG (high mobility group) proteins, which are thought to help establish chromatin structure350 and to bind in the minor groove to DNA segments rich in A and T residues. The intervening region between the A-T hooks and the zinc-finger domains includes a 47-amino acid region of homology with the noncatalytic domains of mammalian DNA methyltransferase (MT), an enzyme that acts on the hemimethylated substrate produced after DNA replication to maintain the methylation pattern of cytosine residues in the genomes of somatic cells.351 Two additional domains have been defined based on their ability to affect transcriptional control, a trans-repression domain overlapping the MT-homology region and a trans-activation domain in the region C-terminus to the zinc fingers.352 The SET domain has been shown to mediate interactions with Sbf1, a protein related to dual specificity phosphatases but which lacks a functional catalytic domain.353 Interestingly, enforced expression of Sbf1 mediates transformation of NIH-3T3 fibroblasts and primary cultures of B-cell progenitors, suggesting that it functions as a SET domain-dependent positive regulator of growth-inducing kinase signaling pathways.354
Translocation breakpoints within the MLL gene occur exclusively in an 8.5-kb region located between exons 5 and 11 and join MLL sequences with genes from numerous other chromosomes to form a large fusion gene (Fig. 19-8). The resulting chimeric proteins, encoded on the derivative 11 chromosome, include the N-terminal half of MLL, with its A-T hook minor groove DNA-binding motifs, the MT-homology domain, and all or part of the associated transcriptional-repression domain.341,342 Another consistent feature of MLL fusion proteins is the absence of the two zinc-finger regions and the Trithorax homology regions normally located in the C-terminal half of the protein.
The MLL gene and some of its fusion partners. The first three genes shown on the left of the ideogram (A) are rich in serine and proline (SP) and contain nuclear localization signals (NLSs), whereas the next two are notable for a cysteine-rich zinc-finger domain and a leucine zipper motif. The AF6 protein contains a novel glycine-leucine-glycine-phenylalanine (GLGF) domain, and AF1P is distinguished by three acidic (A) regions, together with an amino acid repeat motif (aspartic acid-proline-phenylalanine (DPF). AFIQ contributes only a minimal part of its 9-kDa total mass to its fusion with MLL, suggesting that truncation of MLL may itself contribute to leukemogenesis. Regions on the right of the breakpoints (arrows) are retained in the fusion product. Of the four major structural elements of MLL (B), only the A-T hook and mammalian DNA methyltransferase domains are retained in the chimeric proteins. AML cases have been identified recently with “self fusion” rearrangements (C), which fuse the same N-terminal MLL sequences with duplicated regions of the MLL gene. (Adapted from Downing and Look.575 Used with permission).
In contrast to the similar regions of MLL affected by 11q23 rearrangements, an array of structurally diverse protein partners contributes amino acid segments to the MLL fusion proteins found in ALL, AML, and the mixed-lineage leukemias. Twelve of the genes fused to MLL by 11q23 translocations have now been cloned and sequenced (see Table 19-1), making it possible to examine their products for functional motifs that might provide clues to the mechanisms leading to the formation of active transforming proteins. At the time of its cloning, each of these fusion partners was a previously unidentified gene with unknown function. The ELL protein (also known as MEN) was cloned originally as an MLL fusion partner in translocations involving chromosome subband 19p13.1.355,356 In an exciting development, the same protein was independently purified as an elongation factor that increases the catalytic rate of RNA polymerase II transcription.357 This association seems unlikely to be circumstantial, in that ELL has a close functional analogue called elongin (SIII), the transcription elongation factor that is also regulated by the von Hippel-Lindau (VHL) tumor suppressor.358–361 Many questions remain to be answered before the functional significance of the MLL-ELL fusion is known; for example, is the MLL-ELL fusion protein (which contains almost all the ELL coding sequences, fused in-frame to the usual N-terminal segment of MLL) still active as an elongation accelerator? Is ELL subject to regulation by proteins analogous to VHL? If so, does fusion with MLL block this interaction and remove ELL from positive or negative physiological control? And perhaps most important, which genes are controlled in their expression by ELL, and how do they affect cell physiology? Although the full significance of the functional identity of this MLL fusion partner is still unknown, recognition of ELL emphasizes the potentially important roles of such proteins in chimeric constructs, particularly the potential of the chimeras to interfere with the normal roles and regulation of the fusion partner proteins, in addition to their possible inhibitory effects on the normal function of MLL itself.
Another intriguing development in research on MLL fusion genes has been the realization that the identity of the fusion partner may determine the phenotypic specificity of the chimera in hematopoietic stem cell transformation. For example, ENL, the partner gene on subband 19p13.3, is frequently involved in translocations affecting both ALL (especially in infants and children) and AML, 362 whereas ELL is restricted to de novo and therapy-related cases of AML355 and is rare in children.362 ENL is one of three related fusion partners, which include AF-4 on chromosome band 4q21, involved in the frequent 4;11 translocations found in ALL, and AF-9 on band 9p22, the gene affected by the 9;11 translocation important in both primary and secondary acute monocytic leukemias of children and adults. Each of these proteins appears to contribute domains with similar structural attributes to chimeric proteins, 341,342,345,363–365 in that they contain nuclear localization signals and regions rich in serine and proline that may function as transcriptional trans-activation domains.363,365 In support of this possibility, ENL was shown to trans-activate reporter gene expression in mammalian cells and yeast, 365 through the C-terminal serine- and proline-rich region of homology between ENL and AF-9, which is included in the chimeric proteins.
Some MLL partner genes appear to contribute functional domains to the chimeric proteins, whereas others truncate MLL in ways that may interfere with its normal function. AF-10 and AF-17, potential examples of the first mechanism, are involved in the t(10;11)(p12;q23) and t(11;17)(q23;q21) and contain leucine zipper motifs near their C-termini and cysteine-rich zinc-finger motifs near their N-termini. These structural elements are retained in the oncogenic fusion proteins and may provide dimerization motifs with functional significance. The alternative model is best represented by the AF1q gene involved in the t(1;11)(q21;q23). This 9-kDa protein lacks homology to any known protein sequence, 366 and the minimal contribution of its sequences to the uniformly involved MLL N-terminal region implies that loss of MLL function through haploinsufficiency or dominant-negative interference may contribute to leukemogenesis. This interpretation is reinforced by several patients with AML who have lacked 11q23 translocations but have contained partial internal duplications of MLL linking the intact gene to a duplication of its N-terminal region.367,368 In these rearrangements, a region beyond the A-T hook and methyltransferase domains is internally duplicated in-frame with the remainder of the coding sequences (see Fig. 19-4). Thus the partially duplicated MLL gene product contains the N-terminal A-T hooks and methyltransferase domains separated from the zinc-finger motifs, indicating that dissociation of N-terminal domains of MLL from regulatory C-terminal regions is a general structural feature of oncogenic MLL fusion proteins.
MLL, MLL-AF9, and MLL-ENL in Animal Models.
In Drosophila, the MLL homologue trx is a member of a large family of trithorax group proteins that have a positive role in the maintenance of cell-type specific patterns of HOM-C gene expression, apparently acting through epigenetic mechanisms that establish and sustain a receptive chromatin configuration.369 Inactivation of the murine Mll gene in the germ line by homologous recombination has suggested that it has a similar function during normal mammalian development.347,370 Complete loss of Mll was lethal during embryogenesis, with the embryos lacking detectable expression of the major Hox genes tested, consistent with the requirement for trx in maintenance of HOM-C gene expression in Drosophila. Interestingly, mice lacking function of one Mll allele showed a phenotype resulting from haploinsufficiency, with hematopoietic abnormalities that included anemia, thrombopenia and reduced numbers of B cells, bidirectional homeotic transformations of the axial skeleton, and sternal abnormalities. Skeletal abnormalities appeared to be due to shifts in the normal pattern of major Hox gene expression, due to inadequate Mll gene dosage, so that the hematopoietic cell phenotype may have resulted from a similar mechanism, based on studies that implicate mammalian Hox genes in blood cell development.7,371–376 More recent studies have documented decreased numbers of yolk sac-derived CFU-GEMM, CFU-M, and BFU-E colonies in Mll-null embryos.377 Overall, the results in Mll-deficient mice suggest a dual role for 11q23 translocations in human leukemogenesis, including both a gain-of-function effect mediated by the fusion oncogene and simultaneous effects on hematopoietic cell development from haploinsufficiency due to the loss of one normal MLL allele.347
The leukemogenicity of the MLL-AF9 fusion gene was demonstrated recently in an animal model by generating this fusion oncogene in embryonic stem (ES) cells and using them to generate chimeric mice.378 Although ES cells containing the Mll-AF9 fusion gene gave rise to cells of all lineages in chimeric animals and the cells of numerous tissues expressed the fusion gene, the only tumors to develop in the mice were AMLs, reinforcing the association of this fusion gene with human AML. It also appeared that the fusion gene contributed an early growth advantage to progenitors within the myeloid lineage, because circulating myeloid cells derived from the targeted ES cells were a prominent component of this cell compartment in most of the Mll-AF9 chimeras from the time of birth. This effect was not observed in mice generated from an ES cell line modified to express a truncated and epitope-tagged Mll allele, implying that the disordered growth advantage imparted to myeloid cells did not arise from Mll haploinsufficiency but rather from expression of the Mll-Af9 fusion protein. Although leukemia induction was highly efficient in this model, the latency period ranged from 4 to 12 or more months, implying that additional mutations affecting other oncogenes or tumor suppressors must occur before a fully transformed leukemic clone can emerge.378
A retroviral gene transfer assay has been used successfully to document the oncogenic capacity of MLL-ENL (also known as HRX-ENL) by showing its activity in the immortalization and leukemic transformation of myelomonocytic progenitors in mice.379 Detailed structure-function analysis has indicated that the DNA-binding motifs of MLL are required for the full transforming activity of the fusion protein, including the methyltransferase and A-T hook domains.380 Within the ENL sequences of the fusion protein, the C-terminal 84 amino acids of ENL, which comprise two helical structures highly conserved with AF9, are both necessary and sufficient for transformation. These structures were shown to function as transcriptional activators, suggesting that the fusion protein acts to dysregulate the expression of target genes that contribute to transformation within the myeloid lineage.
Origins of Therapy-Related AML.
MLL-associated translocations are a prominent feature of leukemias in patients treated with the epipodophyllotoxins, 338,339,381 but the basis for this association remains uncertain. An intriguing correlation has emerged from analysis of 130 breakpoints by restriction mapping and more than a dozen by genomic DNA sequencing analyses in cases of de novo or therapy-related leukemias with 11q23 translocations affecting the MLL gene.382–387 That is, the centromeric portion of the 8.5-kb genomic MLL breakpoint cluster region consistently showed the largest number of breakpoints in cases of de novo leukemias, whereas the telomeric portion contained the majority of breakpoints found in therapy-related cases.386,387 Scaffold attachment sites have been mapped in the vicinity of these breakpoint regions, as well as high-affinity topoisomerase II cleavage sites, which may influence the distribution of breakpoints.386
With regard to the molecular mechanisms involved in the origin of 11q23 translocations in de novo and secondary AML, studies to date have focused on (1) recombination within Alu repeats, (2) involvement of V-D-J recombinase enzymes in B-lymphoid progenitors, and (3) the possible role of topoisomerase II inhibitors acting to promote breaks at consensus cleavage sites for this enzyme, which would serve as substrates for recombination events leading to MLL translocations in the therapy-related leukemias. The first two possibilities have gained credible support, 368,382,383,385,388 but they do not explain the majority of cases.382,389
An intriguing mechanism of genetic recombination involves cleavage by topoisomerase II, as suggested by the frequent identification of MLL rearrangements in therapy-related cases of AML of patients treated with agents that inhibit this enzyme.338,339,381 Antineoplastic drugs with this property include both the epipodophyllotoxin and anthracycline classes of drugs. AML linked to these agents tends to appear as an acute leukemia without a myelodysplastic phase within 6 to 24 months after diagnosis of the primary malignancy, in contrast to the longer latency periods and frequent myelodysplastic prodromes of secondary AML induced by alkylating agents. AML arising after treatment with a topoisomerase inhibitor typically has monoblastic or myelomonoblastic morphology, 338,339,381,390–392 suggesting that the target cell is a myeloid progenitor cell stimulated to enter cell division by chemotherapy-induced neutropenia.
Topoisomerase II catalyzes a two-step reaction involving both double-stranded DNA cleavage and strand relaxation and religation.393 Both the epipodophyllotoxins and the anthracyclines stabilize the DNA-topoisomerase II complex after cleavage, resulting in the accumulation of double-strand DNA breaks, which are prime substrates for nonhomologous recombination.394,395 Analysis of several 11q23 translocations has identified topoisomerase II consensus binding sites adjacent to chromosomal breakpoints.385 Other cases of therapy-related AML lack topoisomerase II-binding sites adjacent to the breakpoints, so additional 11q23 translocation junctions will need to be analyzed before the frequency and importance of this mechanism can be fully assessed.382,386
Aside from factors predisposing to nonhomologous recombination, how could topoisomerase inhibitors preferentially induce AML with characteristic MLL fusion proteins? The rapid development of these secondary leukemias suggests a collaborative mechanism in which both the drug and the fusion protein act synergistically to accelerate the multistep process leading to AML. A model incorporating the known effects of the epipodophyllotoxins on G2-phase cell cycle checkpoint control and topoisomerase II activity is shown in Fig. 19-9. These compounds arrest cycling cells in G2 phase, and most committed myeloid progenitors harboring epipodophyllotoxin-induced DNA breaks are likely targeted to undergo apoptosis, based on the degree of neutropenia that accompanies a typical course of therapy with these agents (see Fig. 19-9, top panel). Normal myeloid progenitors arrested in G2 occasionally survive, however, with double-strand DNA breaks at the sites of topoisomerase II integration. As these lesions are repaired, some of the breaks are joined by nonhomologous recombination and result in chromosomal rearrangements. The myeloid progenitors with 11q23 translocations producing in-frame MLL fusion genes begin to proliferate because of a proliferative advantage conferred by the hybrid MLL protein and growth factors produced in response to epipodophyllotoxin-induced neutropenia. According to this model, MLL fusion proteins may exacerbate this process by relaxing cell cycle checkpoints normally activated by the presence of the integrated topoisomerase II:drug complex, leading to attenuated apoptosis and increased survival of cells with genetic damage at other loci (see Fig. 19-9, bottom panel). In the face of repetitive epipodophyllotoxin treatment, this could lead to the acquisition of additional genetic lesions affecting oncogenes or tumor suppressors in the expanding clone that already expresses an MLL fusion protein, with rapid progression of a multistep process culminating in overt AML.
Model accounting for the mechanisms linking epipodophyllotoxin therapy, MLL fusion proteins, cell cycle progression, and the relaxation of cell cycle checkpoints, leading to reduced levels of apoptosis in myeloid progenitor cells after genotoxic chemotherapy (hence increased survival of cells with damaged DNA). The accelerated acquisition of additional genetic lesions in clonogenic preleukemic cells eventually culminates in overt myeloid leukemia. See text for further explanation. (Adapted from Downing and Look.575 Used with permission).
The exceedingly high frequency of 11q23 translocations associated with infant leukemias suggests a further mechanism that could lead to MLL gene rearrangement and biologically active chimeric proteins. A number of pairs of infant twins have been shown to have identical MLL gene rearrangements.396,397 In some cases, the leukemias had identical Ig gene rearrangements, consistent with transformation of a common progenitor cell that had completed V-D-J recombination. In other cases the Ig rearrangements differed between twin leukemias, suggesting that transformation occurred before Ig gene recombination and that the leukemic clones had evolved independently. Nonetheless, the identification of identical MLL rearrangements at the DNA sequence level in each twin indicates that the leukemic clone arose in one infant and spread through the placenta to the other sibling. The documentation of MLL rearrangements in utero and the high frequency of 11q23 translocations in infant leukemias (approaching 80 percent of ALL patients and 50 percent of AML patients) suggest that pluripotent progenitor cells with self-renewal capacity are in a proliferative state in the developing bone marrow, rendering them uniquely susceptible to transformation by chimeric MLL oncoproteins. This susceptibility may be related to patterns of gene expression or epigenetic changes in chromatin configuration that are found in subsets of progenitors that are expanding to populate the hematopoietic system during infancy.
Myeloid progenitors similarly susceptible to the transforming effects of chimeric MLL proteins or prone to productive MLL rearrangements may be reactivated in patients undergoing therapy with epipodophyllotoxins, accounting for the rapid onset of secondary AML in children and adults treated with these agents. We have recently identified altered transcripts for the p27KIP1 cell cycle kinase inhibitor in leukemias expressing MLL-AF4398 and have shown that MLL-AF4 induces cell cycle arrest in cell lines when its expression is driven by a conditional promoter, 399 suggesting that hematopoietic stem cells may need to have the capacity to bypass negative cell cycle effects of this fusion protein before they become susceptible to malignant transformation. Moreover, the short latency period between MLL rearrangement and overt leukemia following 11q23 translocations in infants and after epipodophyllotoxin therapy suggests that MLL fusion proteins themselves predispose susceptible hematopoietic progenitors to undergo secondary mutations necessary for the development of a fully transformed leukemic clone.