|
|||||||||
|
|
|
|
|
|
|
|
|
|
|
NEOPLASIA
From the Division of Medical Genetics, Geneva
University Medical School, Switzerland and Genetics and Bioinformatics
Division, The Walter and Eliza Hall Institute of Medical Research,
Royal Parade, Parkville, Victoria, Australia; Department of Viral
Oncology and Department of Genetics and Molecular Biology, Institute
for Virus Research, Kyoto University, Japan; Second Department of
Internal Medicine, Kumamoto University School of Medicine, Japan;
Division of Medical Genetics, University of Washington School of
Medicine, Seattle; Departments of Pediatrics and Oral Biology,
University of Manitoba, Winnipeg, Canada; and Department of Medicine,
Columbia University, New York, NY.
Familial platelet disorder with predisposition to acute myelogenous
leukemia (FPD/AML) is an autosomal dominant familial platelet disorder
characterized by thrombocytopenia and a propensity to develop AML.
Mutation analyses of RUNX1 in 3 families with FPD/AML showing linkage to chromosome 21q22.1 revealed 3 novel heterozygous point mutations (K83E, R135fsX177 (IVS4 + 3delA), and Y260X). Functional investigations of the 7 FPD/AML RUNX1 Runt domain point mutations described to date (2 frameshift, 2 nonsense, and 3 missense mutations) were performed. Consistent with the position of the mutations in the Runt domain at the RUNX1-DNA interface, DNA binding of
all mutant RUNX1 proteins was absent or significantly decreased. In
general, missense and nonsense RUNX1 proteins retained the ability to
heterodimerize with PEBP2 The most frequent mutations associated with
leukemia are recurrent somatic chromosomal translocations or
inversions, many of which involve the polyomavirus enhancer-binding
protein or core-binding factor transcriptional regulation complex
(PEBP2/CBF). Several translocations involve the Studies of families that demonstrate single-gene inheritance for
leukemia predisposition should help to identify the genes and
mechanisms involved in the first steps of leukemia development. The
autosomal dominant familial platelet disorder (FPD)/AML (acute myelogenous leukemia; Online Mendelian Inheritance in Man no. 601399) is a good model to validate this hypothesis because, in addition to developing thrombocytopenia, patients show a propensity for
progression to myelodysplasia and acute myeloid
leukemia.6-9 Affected individuals within the same family
may present with variable clinical severity at varying ages. FPD/AML
was linked to 21q22.1-22.2,3,9 including the region that
contains RUNX1, and germline RUNX1 mutations were
subsequently identified in 6 pedigrees.10
Heterozygous missense mutations and biallelic nonsense or
frameshift mutations in RUNX1 were also identified in
sporadic leukemias,11 and other point mutations have been
characterized in patients with M0-AML and various myeloid
malignancies,12 including some with acquired trisomy
and tetrasomy of chromosome 21.13 While missense mutations
abolish the DNA binding and transactivation activities of
RUNX1, they do not affect heterodimerization with PEBP2 The evolutionarily conserved 128 amino acid Runt domain, present in
most of RUNX1 isoforms, is involved in DNA binding and heterodimerization with PEBP2 It has been proposed that haploinsufficiency for RUNX1 is
responsible for FPD/AML.10 However, hyperactivating,
inhibitory, and loss-of-function RUNX1 point mutations have all been
reported in sporadic leukemia.11 This diversity in the
mechanisms of pathogenesis of sporadic AML RUNX1 mutations suggested
that FPD/AML RUNX1 mutations warranted further functional
studies. Here we describe 3 new RUNX1 mutations in FPD/AML
and potential mechanisms involved in the pathogenesis of the disease
based on in vitro studies of 7 RUNX1 FPD/AML mutations.
Sample collection and genomic DNA isolation
Pedigree 2 (previously unreported) has 8 affected individuals from 4 generations with a clinical history of prolonged bleeding. Constitutional thrombocytopenia was found in 5 members, 3 of whom had
extensive platelet evaluations documenting aspirinlike aggregation defects and dense granule abnormality demonstrable by electron microscopic morphology and quinacrine fluorescence. Three individuals developed AML, and a fourth was said to have died of "pernicious anemia."
Pedigree 3 consists of 15 affected individuals from 4 generations of a
family with a bleeding disorder characterized by impaired platelet
aggregation associated with decreased numbers and contents of both
platelet-dense and Blood samples were obtained from healthy and affected
individuals after obtaining informed consent and in accordance with institutional guidelines for human subjects. DNA was isolated from
blood leukocytes using standard protocols.
Linkage analysis
Mutation analyses
RT-PCR First-strand complementary DNA (cDNA) was synthesized from patient (individuals IV:2 and V:2, pedigree 2) and control RNA using Superscript II reverse transcriptase. PCR was performed using primers AML1RTF3 (Figure 2D, primer C, 5'-CTGCTCCGTGCTGCCTACGCACTG-3') in exon 3 of AML1b and AMLRTR1 (5'-CGCAGCTGCTCCAGTTCACTGAGC-3') in exon 6 as reverse primer on these double-stranded cDNAs, cloned into a TA vector, and sequenced. A cryptic splice site was observed in one clone. Primers covering the cryptic splice junction of exon 4 and 5 (F2AMLmF4, Figure 2D, primer A, 5'-TAATGACCTCAGGGAAAAGCTTC-3') and within the nucleotides of exon 4 excluded from the mutant transcript (F2AMLF1, Figure 2D, primer B, 5'-TGTCGGTCGAAGTGGAAGAGG-3') were designed and used with AML1RTR1 to amplify a PCR product on patients' cDNA; no mutant transcript was observed in control cDNA.
Electrophoretic mobility shift assay Amino acids 24 to 189 of RUNX1, including the Runt domain, were expressed in Escherichia coli, purified, and subjected to electrophoretic mobility shift assay (EMSA) essentially as described previously.11 The wild-type expression plasmid was previously described,11 and mutant RUNX1 plasmids were obtained by site-directed mutagenesis on wild-type pQE9-RUNX1 using Strategene's "quick change" method according to the manufacturer's instruction manual (Stratagene, Amsterdam, The Netherlands). For K90fsX101 an alternative vector, pQE13, was used instead to express it in fusion with a more bulky N-terminal appendage containing hexahistidines and dihydrofolate reductase. Mutant constructs were confirmed by sequencing.Affinity assay of RUNX1-PEBP2  /CBF (1 µg) and mixed with nickel-coated
magnetic beads. The mixture was successively washed with buffers
containing 30 mM imidazole and eluted by the buffer supplied by the
manufacturer. Proteins in each fraction were analyzed as previously
described.11
Subcellular localization For functional studies of mutated RUNX1 proteins, the wild-type RUNX1 sequence (1-453) was inserted into the pEF-BOS mammalian expression plasmid. Construction of RUNX1 mutants was carried out using the megaprimers method,30 and the resultant plasmids were transfected into NIH3T3 or REF52 (H58N, K83N, R177Q, and R177X only) cells using a nonliposomal transfection reagent FuGENE6 (Boehringer, Mannheim, Germany). Immunofluorescence labeling of RUNX1 and PEBP2CBF and microscopy were as previously
described.11 For each transfection 50 cells were visualized.
Transactivation assays The luciferase reporter plasmid (pM-CSF-R-luc) and the effector plasmid (pEF-AML1), containing the 453 amino acid isoform of RUNX1 with the mutations to be studied, were transfected at a fixed ratio (0.5 and 0.8 µg per assay, respectively) into U937, HL60, and Jurkat cells. Cell extracts were prepared 48 hours after transfection and assayed essentially as described.11 To study the inhibitory effect of mutant RUNX1 proteins on wild-type RUNX1 reporter gene activation, 0.5 µg pM-CSF-R-luc and 0.1 to 0.8 µg pEF constructs expressing either RUNX1 with or without mutations or PEBP2-MYH11
were cotransfected into U937, HL60, and Jurkat cells. The total amount
of DNA transfected was kept constant (1.3 µg) by supplementing
appropriate amounts of the backbone pEF plasmid, thereby avoiding
potential artifacts due to unbalanced DNA dosages.
Partial rescue from repression was obtained by transfecting varying
doses of PEBP2
Linkage and mutation analyses of the 3 new pedigrees Each of the 3 families demonstrated features typical of FPD/AML. A candidate region genetic linkage analysis provided maximum 2-point log odds ratio scores of 3.14, 2.15, and 2.43 ( = 0) with
markers D21S1413 in pedigree 1, IFNAR in pedigree 2, and D21S65 in
pedigree 3. Extended haplotype analysis of each family indicated an
approximate 3 megabase common nonrecombinant interval on human
chromosome 21q22.1 containing RUNX1 in affected individuals.
Pedigree 1 showed an A>G substitution in exon 3 resulting in a missense mutation, K83E (Figure 2A). The mutation segregates with the disorder in the family and was absent in 188 unrelated control chromosomes. Pedigree 2 showed a one-base deletion in the splice donor site of intron 4, IVS4 + 3delA (Figure 2C). The mutation segregates with the disorder in the family and was absent in 184 unrelated control chromosomes. Reverse transcriptase (RT)-PCR on RNA from affected individuals showed the use of a cryptic donor splice site, not used in control RNA, 23 nucleotides upstream of the normal splice site (Figure 2D-F). The novel transcript generated by the use of the cryptic splice site results in a frameshift after amino acid 135, addition of 41 unrelated residues, and termination at codon 177 (R135fsX177). Pedigree 3 showed a C>A substitution in exon 7B resulting in a nonsense mutation, Y260X (Figure 2B). The mutation segregated with the disease in all family members tested. Effects of RUNX1 FPD/AML mutations on DNA binding and heterodimerization activities of the Runt domain We examined the function of the Runt domain for 2 new (K83E and R135fsX177; SWISS-PROT Q01196) and 5 described (R139Q, R174Q, K90fsX101, R174X, R177X)10 FPD/AML point mutations. On EMSA analyses, R139Q, R174X, and R177X alone showed barely detectable DNA binding but produced a supershift band of increased intensity in the presence of PEBP2/CBF, indicating that heterodimerization of
these mutants with PEBP2/CBF still occurs (Figure
3A). K83E, R135fsX177, R174Q, and
K90fsX101 showed no detectable DNA binding in the presence or absence
of PEBP/CBF. An affinity assay of these mutants for
heterodimerization with PEBP2/CBF revealed that both missense
mutants retained the ability to heterodimerize whereas both frameshift
mutants did not (Figure 3B).
Subcellular localization of the FPD/AML RUNX1 mutants and
colocalization with PEBP2
While the K83E and R139Q mutations are not within the known NLS, both mutant proteins showed reduced nuclear localization. K83E showed cytoplasmic localization in 30% of transfected cells, in contrast to the normal nuclear localization seen in K83N (a sporadic mutation; Figure 4A).11 This observation may suggest that the K83E mutation (K, positive, hydrophilic, C6; E, negative, hydrophilic, C5) results in greater loss of function than the K83N mutation (N, polar, hydrophobic, C4). The R139Q mutation is at a site involved in DNA binding18 and was expected to affect DNA binding rather than nuclear localization. However, cytoplasmic localization was observed in 40% of cells. The observed defect in nuclear localization in K83E and R139Q is consistent with earlier studies31 and may point to the existence of an additional unidentified N-terminal domain critical for nuclear localization. In contrast to RUNX1, PEBP2 Transactivation abilities of the FPD/AML RUNX1 mutants The transactivation potential of the mutant RUNX1 proteins was measured using a reporter construct based on the M-CSF receptor promoter as a myeloid-specific RUNX1-target.33,34 Wild-type RUNX1 demonstrated more than 100-fold transactivation compared with the FPD/AML mutants, which showed no significant transactivation with or without PEBP2/CBF (Figure
5A). Despite the retention of DNA binding
activity and C-terminal activation domains, R139Q failed to
transactivate. Several mutants showed persistently lower luciferase
activities than the mock-transfected plasmid, indicating that these
mutants could hinder the function of endogenous RUNX1 in U937 cells. To
allow measurement of inhibition of the wild-type transactivational
activities caused by the mutant proteins, the reporter construct and a
fixed amount of nonsaturating wild-type RUNX1 plasmid were
cotransfected with varying amounts of FPD/AML mutant plasmids encoding
mutant proteins with DNA and/or PEBP2/CBF binding activity.
PEBP2/CBF-MYH11 and K83N repressed the M-CSF receptor promoter as
previously reported.11,35 With the exception of R177X, all
of the mutants showed varying degrees of interference with the
wild-type protein function in all cell lines tested (U937, HL60, and
Jurkat). Most notably, K83E and R174Q showed potent inhibition similar
to that of PEBP2/CBF-MYH11 (Figure 5B). The results of
cotransfection reporter assays using frameshift mutants are not
reproducible within or between cell lines.
Because all inhibitory mutants maintained the ability to heterodimerize
and all colocalized with PEBP2
Five RUNX1 point mutations and a deletion of the
entire RUNX1 gene have previously been described in FPD/AML. Here we
report 3 new RUNX1 point mutations associated with the
disease. Y260X, in pedigree 3, is at the beginning of the
transactivating domain and removes a part of the negative regulatory
region for DNA binding.20,31 It is the first described
familial RUNX1 mutation outside the Runt domain. As such,
the studies described in this paper are not appropriate to test the
functional implications of this mutation. Functional analyses of 7 mutants (2 frameshift, 2 nonsense, and 3 missense mutations) were
performed to investigate the mechanisms that contribute to FPD/AML,
particularly the propensity to develop AML. The capacity to bind DNA
was reduced or abolished in all mutant proteins as expected from the
sites of substitutions in the Runt domain. The missense and nonsense
mutant proteins (largely intact Runt domain) retained the capacity to
heterodimerize with PEBP2 Do the mechanisms of pathogenesis of sporadic leukemias provide any clues to the mechanisms of pathogenesis in FPD/AML? In many sporadic hematologic malignancies, the same chromosomal rearrangements consistently appear, supporting the hypothesis that these events are a prerequisite for tumor induction. The somatic mutations acquired in leukemia by translocations and inversions are heterozygous in affected cells; normal copies of the translocated genes are still present. Consequently, the mechanism(s) of disease pathogenesis may include (1) a gain of function by chimeric fusion proteins, (2) haploinsufficiency, (3) a dominant-negative effect of the chimeric fusion proteins, and/or (4) a combination of these mechanisms. However, it is unlikely that the mutations described here have a gain-of-function activity. Haploinsufficiency of RUNX proteins The importance of gene dosage to the normal function of CBF transcriptional regulators is seen in the autosomal dominant human bone disorder, cleidocranial dysplasia, caused by mutations in RUNX2 (also called CBF 1).36,37
Runx2+/ mice show skeletal defects similar to those in
cleidocranial dysplasia.38 RUNX2 mutations may be
heterozygous missense mutations, insertions, and deletions in either
the Runt DNA binding domain or more C-terminal domains responsible for
transactivation.39 RUNX2 point mutations are predicted to
affect the folding and stability of RUNX2,18 resulting in
haploinsufficiency of RUNX2 and defects in bone
formation.36,40-43
Complete deletion of RUNX1 results in haploinsufficiency in FPD/AML.10 The 2 FPD/AML splice site mutations resulting in frameshifts R135fsX177 and K90fsX101 cause loss of DNA binding, heterodimerization, nuclear localization, and transactivation functions and are also likely to act through simple haploinsufficiency. Dominant-negative effects of mutant RUNX proteins? Translocations involving RUNX1 produce chimeric proteins, most notably RUNX1/AML1-ETO(MTG8) in AML t(8;21) and TEL-RUNX1/AML1 in ALL t(12;21). These chimeric proteins both retain the entire Runt domain and have been shown to interfere with transactivation by the normal RUNX1 in a dominant-negative manner.44-46 RUNX1/AML1-ETO and related fusion proteins have been shown to form complexes with PEBP2/CBF and other
nuclear proteins more efficiently than wild-type
RUNX1.47-49 Moreover, naturally occurring isoforms of RUNX1 proteins that contain only the Runt domain can also suppress transactivation by full-length RUNX1.50 Deletion of a
negative regulatory domain for heterodimerization in the C-terminal
region of full-length RUNX1 (amino acids 372-451) results in more
efficient heterodimer formation.31 Thus, chimeric RUNX1
proteins resulting from translocations and the isoforms of RUNX1
without the transactivation domain50 may act in the same
manner by dimerizing with PEBP2/CBF more efficiently than the
wild type.
Cotransfection of equal amounts of wild-type and mutant plasmids
closely mimics the in vivo situation of one normal and one mutant
allele. All FPD/AML RUNX1 missense and nonsense mutants studied, except
R177X, show a decrease in their transactivation capacities and an
inhibitory effect on wild-type RUNX1 upon cotransfection and thus also
seem to act as dominant-negative inhibitors. Both the missense and
nonsense mutants still interact with PEBP2 One of the notable features of the FPD/AML RUNX1 missense mutants
studied here is the colocalization of the RUNX1 and PEBP2 It was recently shown that dimerization with PEBP2 Mice heterozygous for the RUNX1/AML1-ETO or PEBP2 Progression to leukemia in FPD/AML Haploidy of RUNX1 is clearly sufficient for normal development but possibly insufficient for tumor suppression. Runx1/
mice show a complete lack of definitive hematopoietic cells in the
fetal liver, with death occurring from hemorrhages in the central
nervous system at 12.5 days after coitus.56
Runx1+/ mice demonstrate minimal changes in
phenotype,56-58 although closer hematologic analysis has
revealed a trend for bone marrow progenitor cells to have increased
sensitivity to G-CSF, possibly reflecting a propensity to develop
myelogenous leukemia.59
The presence of biallelic RUNX1 mutations in sporadic
leukemias may indicate that RUNX1 functions as a classical
tumor suppressor gene.11,13 However, in other cases of
sporadic leukemia only monoallelic RUNX1 mutations were
described.11,13 Also, in FPD/AML leukemic patients Two tumor suppressor genes (p53 and p27Kip1) have been shown to be haploinsufficient for tumor suppression in hemizygous mice where the potential complication of dominant-negative mutations can be excluded.60,61 A revision of the Knudson model has been proposed to accommodate haploinsufficiency. The normal level of biologic activity of a tumorigenic gene is tightly controlled, and an increase or decrease in this activity leads to increased tumor susceptibility. In a multiprotein complex or molecular cascade, hemizygous loss of each of 2 (or more) partners or molecules in the same pathway may be almost as tumorigenic as homozygous loss of any one partner.62 While mutations in early-acting genes such as RUNX1 predispose to
development of hematologic malignancies, the affected lineage and
consequent type of malignancy may depend upon which genes subsequently
sustain downstream "hits" from additional somatic mutation. For
example, combined positivity for antigens CD34, C-KIT, and HLA-DR
characterizes the CBF leukemias AML-M2 (RUNX1/AML1-ETO t(8;21)) and
AML-M4Eo (PEBP2 It is difficult to make a robust correlation between the type of
mutation and the proportion of patients who develop leukemia in FPD/AML
families, because of the limited number of individuals identified and a
lack of detailed clinical information. It would appear, however, that
families with mutations acting simply via haploinsufficiency (deletions
and frameshift mutations) show a smaller proportion of affected
individuals who develop leukemia than do families transmitting
mutations that may act in a dominant-negative fashion (Figure
6). This is consistent with the
hypothesis that a second mutation has to occur in RUNX1 or
other genes to cause leukemia among individuals harboring an inherited
RUNX1 mutation, and these mutations are more likely to occur
in the individuals with lower biologic activity of RUNX1. For example,
in 2 of the largest FPD/AML pedigrees, a markedly higher rate of
leukemia is seen in the family with strong predicted dominant-negative K83E mutation (57%) compared with the pedigree with a complete deletion of RUNX1 (24%).
Thus, we propose that the less functional PEBP/CBF transcriptional regulation complex present in a hematopoietic cell due to variable inhibitory effects of heterozygous mutations, or mutations of both alleles, the higher the propensity to develop leukemia. This mechanism is valid for FPD/AML or sporadic leukemia patients. Clearly, biallelic mutations would be more prone to leukemia development although additional genetic changes may still be required in other genes. Analyses of additional FPD/AML and sporadic leukemia cases with additional clinical and molecular data, including mutation analyses of genes coding for partners of PEBP/CBF or molecules in the same tumorigenic pathway, will help provide evidence for this hypothesis.
We thank the patients and their families for their participation, Loretta Dougherty and Pauline Crewther for reading the manuscript, Wendy Cook for discussions on haploinsufficiency, Frédéric Schütz and Melanie Bahlo for aid with statistics, and Dong-Er Zhang for providing the M-CSF-R-luc reporter plasmid.
Submitted April 26, 2001; accepted October 10, 2001.
Supported by grants from the Ligue Genevoise Contre le Cancer, the Fondation Pour la Lutte Contre le Cancer, the Fondation Dr Henri Dubois-Ferrière Dinu Lipatti, and the Nossal Leadership Fellowship from the Walter and Eliza Hall Institute of Medical Research to H.S.S.; the International Postgraduate Research (Australian government) and Melbourne International Research scholarships to J.M.; the Swiss FNRS (31-57149.99) to S.E.A.; Grants-in-Aid for Scientific Research from the Japanese Ministry of Education, Science, Sports and Culture, Japan, to Y.I.; grants from the NIH (DK55820 and DK58161), Doris Duke Charitable Foundation (T98006), ALS (6443-00), and the American Cancer Society (RPG-99-319-01-LBC) to M.H.
J.M. and F.W. contributed equally to this article.
The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked "advertisement" in accordance with 18 U.S.C. section 1734.
Reprints: Hamish S. Scott, Genetics and Bioinformatics Div, The Walter and Eliza Hall Institute of Medical Research, Royal Parade, Parkville, PO Royal Melbourne Hospital, Victoria 3050, Australia; e-mail: hscott{at}wehi.edu.au.
1. Lutterbach B, Hiebert SW. Role of the transcription factor AML-1 in acute leukemia and hematopoietic differentiation. Gene. 2000;245:223-235[CrossRef][Medline] [Order article via Infotrieve]. 2. Horwitz M. The genetics of familial leukemia. Leukemia. 1997;11:1347-1359[CrossRef][Medline] [Order article via Infotrieve].
3.
Ho CY, Otterud B, Legare RD, et al.
Linkage of a familial platelet disorder with a propensity to develop myeloid malignancies to human chromosome 21q22.1-22.2.
Blood.
1996;87:5218-5224 4. Horwitz M, Benson KF, Li FQ, et al. Genetic heterogeneity in familial acute myelogenous leukemia: evidence for a second locus at chromosome 16q21-23.2. Am J Hum Genet. 1997;61:873-881[Medline] [Order article via Infotrieve]. 5. Gao Q, Horwitz M, Roulston D, et al. Susceptibility gene for familial acute myeloid leukemia associated with loss of 5q and/or 7q is not localized on the commonly deleted portion of 5q. Genes Chromosomes Cancer. 2000;28:164-172[CrossRef][Medline] [Order article via Infotrieve].
6.
Dowton SB, Beardsley D, Jamison D, Blattner S, Li FP.
Studies of a familial platelet disorder.
Blood.
1985;65:557-563 7. Gerrard JM, Israels ED, Bishop AJ, et al. Inherited platelet-storage pool deficiency associated with a high incidence of acute myeloid leukaemia. Br J Haematol. 1991;79:246-255[Medline] [Order article via Infotrieve]. 8. Luddy RE, Champion LA, Schwartz AD. A fatal myeloproliferative syndrome in a family with thrombocytopenia and platelet dysfunction. Cancer. 1978;41:1959-1963[CrossRef][Medline] [Order article via Infotrieve].
9.
Arepally G, Rebbeck TR, Song W, Gilliland G, Maris JM, Poncz M.
Evidence for genetic homogeneity in a familial platelet disorder with predisposition to acute myelogenous leukemia (FPD/AML) [letter].
Blood.
1998;92:2600-2602 10. Song WJ, Sullivan MG, Legare RD, et al. Haploinsufficiency of CBFA2 causes familial thrombocytopenia with propensity to develop acute myelogenous leukaemia [see comments]. Nat Genet. 1999;23:166-175[CrossRef][Medline] [Order article via Infotrieve].
11.
Osato M, Asou N, Abdalla E, et al.
Biallelic and heterozygous point mutations in the runt domain of the AML1/PEBP2 |
Top
Abstract
Introduction
) of PEBP2
one
each from the families with the complete RUNX1 deletion and
R177X (annotated as R203X by Song et al