Alu elements mediate MYB gene tandem duplication in human T-ALL

While conducting high-resolution microarray-based comparative genomic hybridization (array CGH) analysis of full-complexity DNA derived from 17 established T-ALL cell lines, as well as bone marrow samples from 8 cases of newly diagnosed T-ALL, we observed a highly localized increase in copy number within the 6q23 region, which encompasses the MYB gene and part of the nearby downstream AHI1 gene in 6 cell lines and 2 clinical samples (Fig. 1, A–C; and Figs. S1 and S2).

To determine the mechanisms of increased MYB copy number in T-ALL, we used fluorescent in situ hybridization (FISH) with a commercially available probe for MYB. 2 of the 17 T-ALL cell lines examined showed increased copies of MYB relative to the copy number of the chromosome 6 centromere (Molt4 and Molt13; Table I and Fig. S3), representing translocations of chromosome 6q regions to other chromosomes. These cell lines did not show the localized increases in MYB copy number found with array CGH (Fig. 1 B, samples 7 and 8). In fact, none of the cell lines with a discrete increase in MYB copy number had abnormalities affecting this locus by interphase FISH (Fig. S3). Thus, we postulated that the localized increase of MYB copy number observed by array CGH must affect a region adjacent to one of the normal MYB alleles, resulting in a merged signal by FISH on interphase nuclei and on metaphase chromosome spreads. We therefore used fiber-FISH to hybridize 5′ and 3′ MYB fosmid probes onto stretched DNA fibers from interphase cells of seven different T-ALL cell lines. This strategy demonstrated that MYB is tandemly duplicated on one allele in all five human T-ALL cell lines analyzed with array CGH findings of localized increases in MYB copy number (Fig. 1, D and E; and Table I). Thus, the increased MYB copy number in T-ALL that we and others (15, 16) have observed using array CGH is caused by tandem duplication.

To clone the breakpoints between the duplicated copies of MYB in T-ALL and define the mechanism underlying MYB tandem duplication, we performed long-range PCR. From the duplicated genomic region on chromosome 6, as defined by others (15, 16), and from our own array CGH data on T-ALL cell lines and patient samples (Tables I and II and Fig. 1), we designed a forward primer 54 kb 3′ of the end of the MYB coding sequence and a reverse primer at the 5′ end of the MYB gene (primers 1 and 6; Fig. 2 A). This strategy generated an amplified band of ∼10 kb from T-ALL cell lines (Fig. 2 A). Two more rounds of nested PCR with forward primers progressively 3′ of the MYB gene and reverse primers further 5′ of the MYB gene yielded a 1-kb product that was sequenced (primers 3, 4, 5, 7, and 8; Fig. 2 A). Comparison of the sequence of this product with the normal human genome sequence identified a junction between the duplicated copies of MYB. Analysis of 15 T-ALL samples and cell lines showed that this junction always occurred within a 257-bp region of sequence homology that occurs both 3.5 kb 5′ and 73.5 kb 3′ of MYB. Analysis of this sequence using RepeatMasker (http://www.repeatmasker.org) revealed that it constitutes an Alu element. Alu repeats are short, repetitive DNA sequences of <500 bp that are estimated to account for 10% of the human genome (17). The Alu repeats on either side of the MYB gene are 76% identical, implicating homologous recombination as the underlying mechanism of the tandem duplication (Fig. 2, B and C) (18). Sequence differences between the upstream and downstream Alu elements enabled us to group the junctions of T-ALL cell lines into five classes (Fig. 2 D) (1–5). In each class, the downstream Alu element had merged directly with the upstream Alu region of the duplicated gene in a manner that is most consistent with a synthesis-dependent strand annealing (SDSA) event. In brief, the sequences from the downstream Alu element from one sister chromatid invade the upstream Alu element on the other sister, accompanied by DNA synthesis that proceeds past the downstream homologous sequence. The event is completed by strand annealing of the two ends that now contain identical sequence information (Fig. 3). We demonstrate that Alu elements on either side of the MYB gene mediate its somatic tandem duplication in a significant subset of T-ALL cell lines and patients. Previous reports have identified a larger region of duplication in some samples, which encompasses more than the MYB locus (15, 16). We have also seen that the duplicated region can be larger, and, in these cases, we cannot detect the duplication using our long-range PCR method (e.g., cell line 1 in Fig. 1 B). In these cases, the duplication either occurs by a different mechanism or is mediated by another region of homology.

As shown in Table I, we detected MYB tandem duplication by long-range PCR, but not by multiplex ligation-dependent probe amplification (MLPA) or array CGH in 4 T-ALL cell lines: TALL1, Supt7, DND41, and Koptk1. We also detected MYB tandem duplication by long-range PCR in 11 of 25 (44%) primary T-ALL samples (Table II), which is a higher frequency than that found by MLPA in these samples. Thus, nested long-range PCR detects the tandem duplication in a minority of cells, whereas MLPA and array CGH are quantitative assays that detect MYB tandem duplication only when it occurs in a majority of the cells. Collectively, our results indicate that MYB tandem duplication occurs relatively often in T-ALL patients, but is clonally selected in only approximately one fifth of the cases during evolution of the disease, presumably because some malignant clones have increased MYB copies that have translocated to other chromosomes (as in Molt4 and Molt13; Table I and Fig. S3) or already express sufficient MYB protein levels by virtue of their differentiation state to optimally synergize with other oncogenic events, such as mutation of the NOTCH1 gene (Fig. S4). This conclusion is supported by MLPA analysis showing the absence of MYB tandem duplication in remission samples from two patients whose malignant clones were positive for MYB duplication at diagnosis (Table II).

To determine whether MYB duplication occurs in normal lymphocytes, we performed long-range PCR on 96 DNA samples extracted from lymphoblastoid cell lines from the HapMap panel of individuals. We detected a band representing MYB tandem duplication in 45 of 96 samples (46.9%), and sequencing of all the products revealed that the breakpoints in normal samples were of the 5 types previously identified in T-ALL cell lines (Fig. 2 D), with the identification of one additional breakpoint (no. 6). Further analysis with MLPA did not detect an increase in MYB copy number in any of the 36 HapMap lymphoblastoid cell lines that were analyzed, implying that the duplication occurs infrequently, if at all, as a germline copy number variation, but rather is a somatic event affecting a small fraction of normal lymphocytes. To determine whether RAG expression is necessary for Alu-mediated MYB duplication, we have analyzed the AML cell lines HL60 and Meg01. We have detected the MYB gene duplication in these two lines, and sequencing of the PCR products revealed that the duplication was Alu-mediated, as in the T-ALL cell lines and samples. Therefore, RAG expression is not required for Alu-mediated MYB gene duplication.

We also analyzed 10 peripheral blood and 37 buccal swab samples from healthy individuals for the MYB tandem duplication by long-range PCR and MLPA. We detected the duplication in 2 of 10 blood samples and none of 37 buccal samples by long-range PCR. However, none of the blood or buccal samples was positive for MYB tandem duplication by MLPA. These results demonstrate that MYB tandem duplication occurs spontaneously in normal blood cells and is not an artifact of generating lymphoblastoid cell lines from normal cells. They also indicate that circulating blood mononuclear cells are more prone than epithelial cells to undergo MYB tandem duplication through Alu-mediated homologous recombination (Fisher's exact test, P = 0.042). Blood cells may be more likely to undergo this recombination event because they are more prone to double-strand breaks that initiate the Alu-mediated recombination.

Recent studies in mice have demonstrated that distinct levels of MYB are required at different stages of hematopoietic development for differentiation and proliferation (19, 20). Therefore, it is not surprising that a relatively modest increase in MYB expression, produced as a result of tandem duplication or translocation of a single copy, would be clonally selected in some T-ALLs during their molecular pathogenesis. Intriguingly, in several of our primary T-ALL samples with evidence by nested long-range PCR of rare cells with MYB tandem duplication, this subset of cells did not have a growth advantage, and the MYB copy number of the majority of cells was normal by CGH and MLPA. Thus, if a particular T-ALL clone already expresses sufficiently high levels of MYB for transformation because of its differentiation state, then the increased levels provided by MYB duplication may not provide an advantage for clonal selection.

We and others (15, 16) have observed that MYB duplication occurs often in T-ALL cases with activating mutations in NOTCH1. Knockout studies in the mouse, as well as MYB small interfering RNA experiments in T-ALL cell lines, demonstrated that MYB plays an important role in T cell differentiation (15, 21–23). Furthermore, γ-secretase inhibitors were shown to synergize with MYB small interfering RNA to inhibit T-ALL cell growth (15). These results are consistent with studies in the zebrafish showing that Notch and Myb act in parallel pathways to specify hematopoietic stem cell fate in the AGM (24) and indicate that NOTCH activation and MYB duplication synergize not only in normal blood cell development, but also in malignant T cell transformation.

Our study introduces a new mechanism by which an evolving T-ALL clone can increase the dosage of a critical oncogene during leukemic transformation. Homologous recombination between Alu elements in meiotic germ cells leading to deletions and duplications are responsible for several genetic diseases, including glycogen storage disease type II, Lesch-Nyhan syndrome, and Ehler-Danlos syndrome (25). Alu elements have also been implicated in mediating somatic genetic alterations in cancer, such as MLL translocations and partial tandem duplications in acute myelogenous leukemia (26, 27), BRCA1 gene deletions in breast cancer (28), and TRE rearrangement in Ewing's sarcoma (29). Alu elements have also been found near the LMO2 and LCK chromosomal breakpoints in T-ALL (30, 31), suggesting that they may also play role in these chromosomal translocations. Because Alu elements make up a large percentage of the human genome, it is likely that the copy number of many other oncogenes might be increased, or tumor suppressors deleted, through the recombination mechanism that we describe. The emergence of technologies for global comparative genomic hybridization and single-nucleotide polymorphism analysis in human cancers should enable efficient detection of these pathogenic copy number alterations, especially when the analyses are paired with analysis of normal germline DNA from the same individual.

Cryopreserved lymphoblasts or lymphoblast lysates ALL were collected with informed consent and institutional review board approval from patients with newly diagnosed T-ALL who were treated with Dana-Farber Cancer Institute Acute Lymphoblastic Leukemia Consortium protocols 95-001 or 00-001. Genomic DNA was extracted with the PUREGENE kit according to the manufacturer's protocol (Gentra Systems).

Genomic DNA was fragmented and random-prime labeled, as previously described (32). Labeled DNAs were hybridized to microarrays containing 44K (cell lines) or 244K (primary samples) 60-mer oligonucleotides (Agilent Technologies), as previously described (33). Microarray data have been deposited in the GEO database under accession no. GSE7615.

FISH was performed with a commercially available probe mix containing a MYB (6q23) probe labeled with spectrum aqua and a centromeric probe for chromosome 6 labeled with spectrum orange (Vysis, Inc.). Slides and probe were co-denatured at 75°C, hybridized overnight at 37°C, washed, counterstained with 4′,6-diamidino-2-phenylindole, and analyzed under a fluorescence microscope (Carl Zeiss, Inc.). For each cell line, a minimum of 100 nuclei and any identifiable metaphases were scored. The orange chromosome 6 signals and the aqua LSI MYB signals were counted simultaneously in each nucleus/metaphase. MYB was considered to be amplified if the cells showed more aqua signals than orange signals.

Cells were lysed overnight using PUREGENE Cell Lysis Solution (Gentra Systems). The DNA was stretched on poly-l-lysine–coated slides, and the DNA fibers were fixed in 100% ethanol. To generate FISH probes, fosmids G248P89100B2 and G248P89828A1 were labeled with digoxigenin and biotin, respectively, using BioPrime Array CGH Labeling Module (Invitrogen). DNA fibers and labeled FISH probes were co-denatured for 3 min at 95°C and hybridized overnight in a humidified chamber at 37°C. The haptens were detected using anti–digoxigenin-fluorescein/Fab fragments (Roche) and Strepavidin/Alexa Fluor 594 conjugate (Invitrogen). Slides were analyzed under a fluorescence microscope (model BX51AI; Olympus). A normal allele showed green dots followed by red dots. A tandem duplication was seen as a direct repeat of the normal signal pattern.

MLPA probe mix containing all probes was added to 150 ng denatured genomic DNA. A universal primer pair was used to amplify all ligated probes in a multiplex PCR assay. The PCR product was analyzed by sequence-type electrophoresis and samples were compared with two control samples. A difference in relative peak area indicated a copy number change of the DNA sequence targeted by the probe. A probe was considered amplified if the ratio sample/control was >1.3.

Reactions were performed with 500 ng of genomic DNA and the LA PCR kit version 2.1 (Takara Mirus Bio) according to the manufacturer's specifications. Primer sequences are available upon request.

Figs. S1 and S2 show array CGH data on T-ALL cell lines and patient samples analyzed by CGH Analytics (Agilent Technologies). Fig. S3 shows MYB FISH on T-ALL cell lines. Lines with tandemly duplicated MYB do not show increased copies of MYB by this method; however, two T-ALL cell lines with extra copies of MYB dispersed to other chromosomes were identified. Fig. S4 shows that MYB duplication results in increased levels of MYB RNA and protein levels in T-ALL cell lines.

We would like to thank James Haber for helpful discussions and John Gilbert for editing and general comments.

This work was performed and supported in part by the Center for Applied Cancer Science of the Robert A. and Renee E. Belfer Foundation Institute for Innovative Cancer Science, a National Cancer Institute grant CA11560, and a Leukemia and Lymphoma Society Translational grant 6161-05 (C. Lee), National Institutes of Health (NIH) grants CA109901 and CA68484-11 (A.T. Look), and a German Research Foundation Fellowship Award Tc-57/1-1 (J. Tchinda). J. O'Neil is supported by a Harvard Medical School Hematology Training Grant. R.A. DePinho is an American Cancer Society Research Professor and an Ellison Medical Foundation Senior Scholar. R.S. Maser is a Damon-Runyon Postdoctoral Fellow. R. Rothstein is supported by NIH grants GM050237, GM067055, and CA115853.

The authors have no conflicting financial interests.