Biased Distribution of Chromosomal Breakpoints Involving the MLL Gene in Infants Versus Children and Adults with T(4;11) ALL

Biased distribution of chromosomal breakpoints involving the MLL gene in infants versus children and adults with t(4;11) ALL

Martin Reichel1, Esther Gillert1, Sieglinde AngermuÈ ller1, Jan Patrick Hensel1, Florian Heidel2, Martin Lode2, Thomas Leis2, Andrea Biondi3, Oskar A Haas4, Sabine Strehl4, E Renate Panzer-GruÈ mayer4, Frank Griesinger5,JoÈ rn D Beck2, Johann Greil2, Georg H Fey1, Fatih M Uckun6 and Rolf Marschalek*,1,7

1Chair of Genetics, University of Erlangen-NuÈrnberg, Staudtstr. 5, D-91058 Erlangen, Germany; 2Department of Pediatrics, University of Erlangen-NuÈrnberg, Loschgestr. 15, D-91054 Erlangen, Germany; 3Centro di Ricerca M. Tettamanti, Ospedale S. Gerardo, Via Donizetti 106, I-20052 Monza (MI), Italy; 4CCRI St. Anna Kinderspital, Kinderspitalgasse 6, A-1090 Wien IX, Austria; 5University Hospital GoÈttingen, Division of Haematology/Oncology, Robert-Kochstr. 40, D-37075 GoÈttingen, Germany; 6Parker Hughes Cancer Center and Children's Cancer Group ALL Biology Reference Laboratory, Parker Hughes Institute, 2665 Long Lake Rd STE 200, Saint Paul, Minnesota, MN 55113, USA; 7Institute of Pharmaceutical Biology, University of Frankfurt, Biocenter, Marie-Curie Str. 9, D-60439 Frankfurt/Main, Germany

Derivative chromosomes of 40 patients diagnosed with 1991) on the long arm of human chromosome 11 (band t(4;11) acute lymphoblastic leukemia (ALL) were q23) have been found in patients with hematologic analysed on the genomic DNA level. Chromosomal malignancies (for reviews see Bernard and Berger, breakpoints were identi®ed in most cases within the 1995; Marschalek et al., 1997; Cimino et al., 1998). To known breakpoint cluster regions of the involved MLL illuminate the promiscuous recombination behavior of and AF4 genes. Due to our current knowledge of the the MLL gene and the illegitimate recombination primary DNA sequences of both breakpoint cluster mechanism(s), the chromosomal breakpoints of regions, speci®c features were identi®ed at the chromo- t(4;11) ALL patients were investigated here. Dierent somal fusion sites, including deletions, inversions and PCR-based approaches have emerged (Felix et al., duplications of parental DNA sequences. After separa- 1997; Leis et al., 1998; Reichel et al., 1999) that tion of all t(4;11) leukemia patients into two age classes allowed a systematic analysis of chromosomal break- (below and above 1 year of age), the analysis of points. The availability of the primary DNA sequence chromosomal fusion sites revealed signi®cant dierences of the MLL (Gu et al., 1994; Marschalek et al., 1995) in the distribution of chromosomal breakpoints and led to and AF4 (Reichel et al., 1999) breakpoint cluster the de®nition of two hotspot areas within the MLL regions was a prerequisite for the rapid detection of breakpoint cluster region. This may point to the chromosomal breakpoints in primary biopsy material. possibility of dierent age-linked mechanisms that were Analysis of fusion sites between the MLL and AF4 leading to t(4;11) chromosomal translocations. Oncogene genes revealed signs indicative of non-homologous (2001) 20, 2900 ± 2907. DNA end joining (NHEJ), including duplications, inversions and deletions in the range from a few Keywords: chromosomal translocation t(4;11); ALL-1/ basepairs (bp) to several kilobases (kb). In more than MLL/HRX gene; AF4 gene half of the cases, `®ller DNA' and `mini-direct repeats' were found at the inter-chromosomal junctions. These results prompted the hypothesis that illegitimate Introduction recombinations leading to these translocations were likely initiated by DNA damage involving both Recurrent chromosomal translocations are frequently chromosomes that was subsequently followed by a associated with malignant transformation of hemato- misguided DNA repair process (Super et al., 1997; poietic cells and involved in the pathogenesis of Reichel et al., 1998, 1999; Gillert et al., 1999). To test myeloid and lymphoid neoplasias (Rabbitts, 1994). this hypothesis, chromosomal breakpoints in the More than 40 dierent chromosomal translocations genomic DNA were investigated by using a large involving the MLL gene (Ziemin van der Poel et al., collection of primary biopsy specimens from patients diagnosed with t(4;11) ALL. The data set from 40 novel patients with t(4;11) ALL provides evidence that for t(4;11) ALL the `DNA damage repair model of *Correspondence: R Marschalek, Institute of Pharmaceutical chromosomal translocations t(4;11)' (Reichel et al., Biology, University of Frankfurt, Biocenter, N230, 303, Marie- 1998; Gillert et al., 1999) is valid and that DNA repair Curie Str. 9, D-60439 Frankfurt/Main, Germany Received 6 November 2000; revised 13 February 2001; accepted 19 via homologous recombination was not involved. February 2001 Moreover, analysis of chromosomal breakpoints con- Age-dependent breakpoint distribution in t(4;11) leukemia cells M Reichel et al 2901 ®rmed that signi®cant dierences in the breakpoint Inverted DNA sequences from the MLL or AF4 genes distribution within the MLL gene exist between infant were found in 2/14 cases (UPN010 and 024). and non-infant patients diagnosed with t(4;11) ALL In two patients, reciprocal breakpoints were found in (Cimino et al., 1997). repetitive Alu elements present in both genes. There- fore, the possibility was explored that homologous recombination may have led to the translocated chromosomes. In UPN035, the der(4) chromosome Results was created by a recombination event between the 5' portion of an Alu element in intron 3 of AF4 and an Characteristic features for NHEJ-repair including Alu element located in intron 9 of MLL. Both Alu duplications, deletions and inversions are present at elements were oriented in opposite directions and the chromosomal breakpoints in t(4;11) cells direct comparison of neighboring DNA sequences \Speci®c genomic alterations, including duplications, ¯anking the chromosomal breakpoints revealed no inversions or deletions of parental DNA sequences at signi®cant homology between the rearranged Alu the genomic break sites, have been identi®ed at sequences (14 identical nucleotides in 58 ¯anking chromosomal breakpoint junctions of leukemic cells nucleotides). The situation was similar for sample derived from t(4;11) ALL patients (Super et al., 1997; UPN040, where two repetitive Alu elements in opposite Reichel et al., 1998, 1999; Gillert et al., 1999). In the orientations were recombined to create a der(4) present study, the previously published data for 20 chromosome. Therefore, for this case too we concluded t(4;11) patients and cell lines (Gillert et al., 1999) were that homologous recombination did not play a major extended and an update on 60 cases with t(4;11) leukemia role for the recombination process since only nine is shown in Tables 1 and 2. The main results for the 40 identical nucleotides were found in 58 ¯anking novel cases (a total of 66 alleles) are the following. nucleotides. In 26/40 cases, the breakpoints were identi®ed and In summary, the breakpoint sequences from 39/40 sequenced on both reciprocal chromosomes. In these cases of t(4;11) leukemias support the conclusion that cases, it was possible to align the breakpoint sequences broken genomic DNA of the MLL and AF4 genes was with areas from the corresponding intact chromosomes repaired with the help of NHEJ processes. The only and thus to detect all deviations from the parental exception from this rule was sample UPN041, which sequences. The comparison of the recombined alleles showed a perfect crossover event between the two with the parental DNA sequences revealed the presence participating chromosomes. However, we consider it of duplications (range: 1 ± 463 bp), inversions (range: highly unlikely that homologous recombination was 16 ± 267 bp), deletions (range: 2 ± 4.413 bp) or combi- involved in this case. Thus, in special cases, it may be nations thereof. One exception was UPN041, with possible to create fused chromosomes by illegitimate evidence of a perfect cross-over event between the recombination processes without loss or addition of MLL and AF4 genes, although the overall identity of sequences at the fusion site. DNA ¯anking the recombination point was only 36% (21 identical nucleotides in 58 ¯anking nucleotides). In Breakpoints on derivative chromosomes 4 and 11 in addition, a trinucleotide mini-direct repeat (MDR) was t(4;11) ALL identi®ed on both derivative chromosomes. Mini-direct repeats are short stretches of identical nucleotides that In the present study the majority of chromosomal are found at the break sites of both parental breakpoints were located in the MLL breakpoint chromosomes and that are presumably used to form cluster region between exons 9 and 12, and in the a stretch of mini-homology for a certain type of DNA AF4 breakpoint cluster region between exons 3 to 6. In repair. In 27/66 derivative alleles, mini-direct repeats all these cases, transcription and subsequent splice (1 ± 7 bp) were present at the chromosomal fusion sites. reactions were predicted from the sequences to give rise In 20/66 derivative alleles analysed here, inserts of ®ller to functionally fused open reading frames of both DNA sequences (1 ± 21 bp) were found at the chromo- reciprocal fusion mRNA species. In one case somal junctions. Both mini-direct repeat sequences and (UPN009), the der(11) breakpoint was located outside ®ller DNA sequences are typical for NHEJ repair of the major breakpoint cluster regions, and involved processes and were found in a total of 47/66 alleles exon 9 of MLL that was fused to intron 6 of AF4.In (71%). two other cases, breakpoints on der(4) were found to In 14/40 cases, only a der(11) allele was identi®ed involve AF4 exon 3 and MLL intron 9 (UPN007 and and any attempt to identify the breakpoint on the 056). In one case, AF4 intron 3 was fused to MLL der(4) allele failed (Table 1). In these cases, the analysis intron 14 (UPN015). Breakpoints located outside the was hampered by the fact that only three DNA major breakpoint cluster regions can give rise to fused sequences were available for alignment (two wildtype open reading frames, if exons with the proper splicing and one derivative sequence) and therefore deletions or phase are used as acceptors. However, no data on duplications at the breakpoints could not be identi®ed spliced mRNA species were available for these cases. unambiguously (Table 2). However, insertions of To achieve statistical signi®cance in the following inverted DNA, ®ller DNA or mini-direct repeats were analyses, we used the combined data of all 60 still detected in the der(11) alleles of these patients. individual t(4;11) patients that have been analysed

Oncogene Age-dependent breakpoint distribution in t(4;11) leukemia cells M Reichel et al 2902 Table 1 Chromosomal breakpoint analysis in t(4;11) patients. The following parameters were listed for all t(4;11) patients: age, localization of the der(11) breakpoint in MLL, localization of the der(4) breakpoint in MLL, localization of the der(11) breakpoint in AF4, localization of the der(4) breakpoint in AF4, and the contributors (see Materials and methods) Patient Age MLL der(11) MLL der(4) AF4 der(11) AF4 der(4) Sender

UPN001 3m 1.093 1.117 12.893 12.828 StP UPN002a 6m 2.729 ± 31.231 ± Gi UPN003 9m 5.235 ± 4.395 ± GoÈ UPN004 2m 5.293 5.331 13.357 13.302 StP UPN005a 5m 6.500 6.601 33.369 32.767 Vi UPN006 4m 3.886 3.800 5.822 5.002 StP UPN007 3m 1.060 ± 3.328 ± Mo UPN008 inf 4.416 4.278 3.878 3.883 GoÈ UPN009 4m 634 ± 48.860 ± StP UPN010 1m 6.001 ± 4.090 ± Mo UPN011 1a 6.033 ± 35.609 ± Gi UPN012a 3m 5.047 5.530 16.179 16.189 Vi UPN013 5m 1.566 ± 36.523 ± GoÈ UPN014 11m 6.399 6.360 41.123 41.152 StP UPN015 inf 6.032 8.780 36.395 36.397 Lo UPN016a 6m 560 ± ± ± ** UPN017 1m 2.641 2.657 23.399 23.518 Vi UPN018a 10w 4.629 4.554 6.871 6.990 Er UPN019 inf 4.503 ± 23.063 ± GoÈ UPN020 3m 2.891 ± 11.962 ± GoÈ UPN021 inf 6.552 6.549 47.521 47.470 Lo UPN022a 9m 6.767 6.763 12.386 12.479 Gi UPN023 4m 6.554 6.432 33.608 33.314 GoÈ UPN024 inf 2.358 ± 44.369 ± GoÈ

UPN025 1,2a 4.236 4.250 24.535 24.501 StP UPN026 10a 2.871 2.879 46.124 46.111 Ha UPN027 2a 6.540 ± 9.473 ± GoÈ UPN028 42a 2.344 ± 19.698 ± Mu UPN029 18a 915 946 27.551 27.549 StP UPN030 15a 2.845 ± 19.792 ± StP UPN031a 15a 5.462 5.985 47.594 47.868 GoÈ UPN032a 11a 2.544 2.549 9.655 9.832 Gi UPN033a 13a 6.608 6.643 29.650 29.572 Gi UPN034 43a 2.005 2.003 12.638 12.640 Er UPN035 ad 1.442 1.448 11.069 10.965 GoÈ UPN036 3.786 ± 32.399 ± GoÈ UPN037a 5a 6.329 6.480 45.213 45.087 Gi UPN038a 48a 2.511 3.524 47.446 47.584 GoÈ UPN039 5,5a 1.781 ± 33.846 ± StP UPN040 15a 1.646 1.630 7.471 7.447 Vi UPN041 2,5a 3.092 3.092 45.614 45.614 StP UPN042a 2.294 ± 36.741 ± Gi UPN043a 2a 2.997 2.673 6.622 6.687 Gi UPN044 39m 1.197 889 38.747 34.334 GoÈ UPN045a 10a 1.129 1.129 43.792 44.041 ** UPN046 ad 1.430 1.454 29.901 28.891 GoÈ UPN047a 4a 6.365 6.398 14.945 14.979 Gi UPN048a 2 3.064 3.010 9.311 9.281 Gi UPN049a 10a 2.802 3.446 31.443 31.098 Gi UPN050a 32a 2.523 2.556 36.570 36.562 ** UPN051 10a 2.722 2.300 38.374 38.375 Vi UPN052a 5a 2.386 2.769 32.909 32.595 ** UPN053 8a 2.027 2.279 6.434 4.247 Vi UPN054 15a 894 893 3.275 3.266 Er UPN055 57a 2.745 2.751 14.594 14.577 Mo UPN056 30a 3.034 3.182 29.822 29.108 Mo UPN057 13a 959 2.422 5.061 4.987 StP UPN058a 4a 762 727 43.655 43.840 Gi UPN059 15a 1.677 1.834 18.206 18.669 StP UPN060a 1,2a 3.748 3.790 30.601 30.580 Gi

aPreviously published in Oncogene, 18, 4663 ± 4671, and Cancer Res., 59, 3357 ± 3362; ± , no information available; **, established t(4;11) cell lines. Upper part: infants below 1 year of age at diagnosis. Lower part: patients above 1 year of age at diagnosis and non-classi®ed patients

in the past years in our laboratory. Patient data Chromosomal breakpoints were not evenly distrib- published previously were marked in Tables 1 and 2. uted in the breakpoint cluster regions of the MLL and

Oncogene Age-dependent breakpoint distribution in t(4;11) leukemia cells M Reichel et al 2903

Table 2 Detailed analysis of derivative alleles from t(4;11) cells. The following parameters were listed for all t(4;11) patients: duplicated areas from the MLL and AF4 genes. Inversions identi®ed in the der(11) and der(4) chromosomes, duplicated areas from MLL and AF4 genes, the presence of ®ller DNA in the der(11) and der(4) chromosomes, and the presence of mini-direct repeats (MDR) in the der(11) and der(4) chromosomes Duplications Inversions Deletions Filler DNA MDR Patient MLL AF4 der(11) der(4) MLL AF4 der(11) der(4) der(11) der(4)

UPN001 ± ± ± 16 24 65 ± ± ± UPN002a n.p. n.p. ± ± n.p. n.p. ± ± ± ± UPN003 n.p. n.p. ± ± n.p. n.p. ± ± ± ± UPN004 ± ± ± ± 38 55 ± ± 2 4 UPN005a ± ± ± ± 101 602 ± ± 4 1 UPN006 86 ± ± ± ± 820 ± ± 1 1 UPN007 n.p. n.p. ± ± n.p. n.p. ± ± ± ± UPN008 138 6 ± ± ± ± ± 2 ± ± UPN009 n.p. n.p. ± ± n.p. n.p. ± ± 1 ± UPN010 n.p. n.p. 117 ± n.p. n.p. ± ± ± ± UPN011 n.p. n.p. ± ± n.p. n.p. ± ± 2 ± UPN012a ± 9 ± ± 483 ± ± 2 1 ± UPN013 n.p. n.p. ± ± n.p. n.p. 1 ± ± ± UPN014 36 29 ± ± ± ± ± ± 3 ± UPN015 ± 2 ± ± 2750 ± ± ± ± 2 UPN016a n.p. n.p. 65 ± n.p. n.p. ± ± ± ± UPN017 ± 219 ± ± 17 ± 4 ± ± 1 UPN018a 75119±±±±±±±1 UPN019 n.p. n.p. ± ± n.p. n.p. ± ± 1 ± UPN020 n.p. n.p. ± ± n.p. n.p. ± ± 2 ± UPN021 3 ± ± ± ± 51 2 ± ± 2 UPN022a 493± ± ± ± ± ± 3 1 UPN023 122 ± ± ± ± 294 ± ± 3 2 UPN024 n.p. n.p. 267 n.p. n.p. ± ± 2 ±

UPN025 ± 4 32 35 14 34 ± ± ± ± UPN026 ± ± ± ± 8 13 ± ± ± 2 UPN027 n.p. n.p. ± ± n.p. n.p. 1 ± ± ± UPN028 n.p. n.p. ± ± n.p. n.p. ± ± ± ± UPN029 ± ± ± ± 31 2 3 2 ± ± UPN030 n.p. n.p. ± ± n.p. n.p. ± ± 2 ± UPN031a ± 275 ± ± 523 ± ± ± 7 ± UPN032a ±177±±5±±1±± UPN033a ±±±±35783±±± UPN034 2 2 ± ± ± ± ± 1 ± ± UPN035 ± ± ± ± 6 104 21 ± ± ± UPN036 n.p. n.p. ± ± n.p. n.p. ± ± 2 ± UPN037a ± ± ± ± 151 126 ± ± ± ± UPN038a ± 138 ± ± 1013 ± 1 ± ± 1 UPN039 n.p. n.p. ± ± n.p. n.p. 2 ± ± ± UPN040 17 ± ± ± ± 24 ± ± ± ± UPN041 ± ± ± ± ± ± ± ± 3 3 UPN042a n.p. n.p. 44 ± n.p. n.p. ± ± 2 ± UPN043a 32465±±±±±±3± UPN044 308 ± ± ± ± 4413 ± ± 1 1 UPN045 ± 249 ± ± ± ± ± ± ± ± UPN046 ± ± ± ± 24 10 1 4 ± ± UPN047a ±34±±33±±61± UPN048a 54±±±±303 1 ± ± UPN049a ± ± ± ± 644 345 ± ± 3 3 UPN050a ±±±43338±±±± UPN051 422 1 ± ± ± ± 2 1 ± ± UPN052a ± ± ± ± 383 314 ± 1 ± ± UPN053 ± ± ± ± 252 2187 ± ± ± 2 UPN054 ± 9 ± ± 1 ± ± 6 2 ± UPN055 ± ± ± ± 6 17 3 ± ± 2 UPN056 ± ± ± ± 148 714 ± ± ± 7 UPN057 ± ± ± ± 1463 74 ± 7 2 ± UPN058a 35185±±±±±23± UPN059 ± 463 ± ± 157 ± 2 4 ± ± UPN060a ±±±±4221±±23 aPreviously published in Oncogene, 18, 4663 ± 4671, and Cancer Res., 59, 3357 ± 3362; ± , not identi®ed; n.p., not possible. Upper part: infants below 1 year of age at diagnosis. Lower part: patients above 1 year of age and non-classi®ed patients

Oncogene Age-dependent breakpoint distribution in t(4;11) leukemia cells M Reichel et al 2904 AF4 genes (Figure 1a,b). To describe the distribution are listed. Patients above 1 year of age (child and of breakpoints, we de®ned a `local breakpoint density' adult) and non-age-classi®ed patients (n=36) were (LBD: number of chromosomal breakpoints on the listed in the lower part of Table 1. The mean positions der(11) chromosome per 100 bp in the corresponding of chromosomal breakpoints in the MLL gene were area). In the MLL gene, the local breakpoint density signi®cantly dierent for both groups: in infants was highest in intron 10 (part of cluster I in Figure 1b; (n=24), the mean position (+standard deviation) of 12 breakpoints spread over 547 bp; LBD=2.2) and chromosomal breakpoints in the der(11) chromosomes exon 11 (three breakpoints spread over 114 bp; was at nucleotides 4.141+2.118. In children and adults LBD=2.6). For the introns 9 and 11 the LBD's were (n=34; two patients were excluded due to missing 1.1 and 0.6, respectively. Within intron 9 the ®rst and ages), the mean position was at nucleotides the third Alu-elements were the main targets for 2.813+1.681. The dierence between both groups reciprocal translocation (®ve and ®ve breakpoints; was highly signi®cant (P value=0.0102; Student's t- LBD's were 1.6 and 1.7). Within intron 11, one region test, 2-tailed). No signi®cant dierence in the break- displayed a higher local breakpoint density. This region point distribution in the AF4 gene was observed for the is located between the 3'-end of Alu element #7 and two patient groups: in infants, the mean breakpoint exon 12 (cluster II in Figure 1b; seven breakpoints position in the der(11) chromosomes was at nucleotides spread over 414 bp; LBD=1.7). The long range 23.049+15.597. In the group of children and adults, average breakpoint density in the breakpoint cluster this value was 25.607+14.569 (P value=0.5320; region of the MLL gene was 0.7 (60 chromosomal Student's t-test, 2-tailed). breakpoints over 8.321 bp). Thus, the centromeric and For the purpose of statistical analysis, the MLL telomeric clusters have a 3.1 ± 3.7- and 2.4-fold higher breakpoint cluster region was arbitrarily subdivided at local density than the average overall density. There- the single XbaI site in intron 11 (Strissel-Broeker et fore, it was concluded that particular regions within the al., 1996) into a centromeric (CEN; 4.452 nt) and a MLL breakpoint cluster region were more prone to telomeric portion (TEL; 3.869 nt), with respect to the chromosomal translocations and thus, that recombina- orientation of the gene on the chromosome (see tion `hotspots' may exist within the breakpoint cluster Figure 2). Altogether, 42 chromosomal breakpoints region of the MLL gene. were identi®ed in the centromeric portion (LBD=0.9) Age at diagnosis was used to sub-categorize the and 18 in the telomeric area (LBD=0.5) of the MLL t(4;11) leukemia patients into two subgroups. The breakpoint cluster region. Chromosomal breakpoints infant subgroup (n=24) is summarized in the upper in infants (Figure 2: white circles) showed a preferred part of Table 1, where all patients below 1 year of age localization in the telomeric area (13/24; 54%),

Figure 1 Unequal distribution of chromosomal break sites in the breakpoint cluster region of the MLL gene. (a) The breakpoint cluster region of the MLL gene was subdivided into eight segments of equal length (1 ± 1000, 1001 ± 2000, etc.). The frequency of chromosomal breakpoints in each of these segments was determined and summarized for both patient groups. (b) Map of the breakpoint cluster region of the MLL gene. Chromosomal breakpoints are indicated for both patient groups (5 and 41 year of age at diagnosis). Breakpoint clusters I and II are indicated below the map. Exon nomenclature according to Nilson et al. (1996)

Oncogene Age-dependent breakpoint distribution in t(4;11) leukemia cells M Reichel et al 2905

Figure 2 Two-dimensional scatter analysis of chromosomal breakpoints from t(4;11) leukemia cells in the MLL and AF4 genes. Horizontal axis: breakpoint cluster region of the human AF4 gene (452 kb). Vertical axis: breakpoint cluster region of the human MLL gene (48 kb). Exons of both genes (AF4: exons 2 ± 7; MLL: exons 8 ± 14) are indicated by black boxes (nomenclature according to Nilson et al., 1996, 1997). Alu repetitive elements and their orientation are indicated by grey arrows. XbaI: restriction site that was used to arbitrarily divide the MLL breakpoint cluster region into a centromeric and telomeric portion. The vertical line arbitrarily divides the AF4 breakpoint cluster region into a centromeric and telomeric portion. Black dots: chromosomal breakpoints of patients older than 1 year of age. Open circles: chromosomal breakpoints of patients below 1 year of age. Shaded circles: chromosomal breakpoints of t(4;11) patients for which no age-information was available. CEN/TEL: centromeric/telomeric portion of the corresponding breakpoint cluster regions whereas breakpoints in non-infants (Figure 2: black Discussion dots) were preferentially located in the centromeric part (29/34; 85%) of the MLL breakpoint cluster The present study of genomic breakpoints of derivative region. chromosomes in t(4;11) ALL cells extended our The breakpoint cluster region of the AF4 gene was knowledge of the basic mechanisms leading to also subdivided into a smaller centromeric (nucleotides translocated chromosomes. The major conclusions 1 ± 17.000) and a larger telomeric portion (nucleotides drawn from this study are: (a) the `DNA damage 17.001 ± 52.909). Non-infants had 12 breakpoints in the repair model for chromosomal translocations' was centromeric and 22 in the telomeric portion, corre- valid for 40 additional cases with t(4;11) ALL. The sponding to nearly the same local breakpoint for both data indicate that these translocations were the result subregions density (LBD=0.07 and 0.06, respectively). of `misguided DNA repair processes' that involved Therefore, for non-infants, the breakpoints were evenly non-homologous recombination between the partici- distributed over the entire breakpoint cluster region of pating chromosomes and presumably components of the AF4 gene. By contrast, infants had 12 breakpoints the NHEJ pathway; (b) no signs for homologous in the centromeric and 12 in the telomeric subregion, recombination between AF4 and MLL were identi®ed corresponding to local breakpoint densities of 0.07 and in any of the 40 t(4;11) leukemia samples (a total of 66 0.03, respectively. Thus, unexpectedly, for infants an derivative alleles were analysed). Recombinase- uneven distribution was evident with a weak bias mediated translocation events can be excluded (Gillert towards the centromeric portion. et al., 1999), and hence it appears that t(4;11) From the data presented here, we concluded that translocations are the result of illegitimate recombina- age-classi®ed t(4;11) ALL patients have statistically tion events involving the NHEJ DNA repair pathway; signi®cant dierences in their distribution pattern of (c) the distribution of recombination events in the chromosomal breakpoints in the MLL gene (P breakpoint cluster region of the human MLL gene value=0.0036; Chi-square analysis), whereas the dis- showed signi®cant dierences between infants and non- tribution in the AF4 gene showed no signi®cant age- infant patients. This may suggest that certain chroma- dependent variation (P value=0.3961; Chi-square tin areas of the MLL breakpoint cluster region have a analysis). Infants, however, showed a weak preference preferential sensitivity to DNA damage at dierent for chromosomal breakpoints in the centromeric stages of development. Possibly, the same is true for portion of the AF4 breakpoint cluster region. the breakpoint cluster region of the human AF4 gene.

Oncogene Age-dependent breakpoint distribution in t(4;11) leukemia cells M Reichel et al 2906 However, more data are required to substantiate this carried breakpoints predominantly located in the initial observation. centromeric part of the MLL breakpoint cluster region. Chromosomal breakpoint analysis in leukemic cells This biased distribution supports earlier observations derived from t(4;11) ALL patients revealed that most based on Southern blot and RT ± PCR experiments `balanced' t(4;11)-translocations are unbalanced at the (Strissel-Broeker et al., 1996; Cimino et al., 1997). The ®ne structure level. T(4;11) translocations are asso- reason for the unequal distribution in both groups of ciated with additional rearrangements that include patients is currently unknown. The recently identi®ed duplications, inversions and deletions of parental hotspot area in the vicinity of exon 12 of the MLL DNA sequences with a median range of several gene (Stanulla et al., 1997; Strissel et al., 1998) ± which hundred basepairs. Fourteen of the 40 cases displayed is approximately identical to cluster II (Figure 1) ± may only a der(11) chromosome in the PCR experiments. In be one possible explanation. However, this explanation 5/14 cases, the cytogenetic analysis had shown the can apply only to a subset of t(4;11) ALL cases, presence of a der(4) chromosome, and in 1/14 cases, a because less than 12% of all der(11) breakpoints der(4) chromosome was clearly absent in the karyotype mapped to this particular region. Moreover, it is analysis. For the other eight cases, no cytogenetic obviously not restricted to reciprocal t(4;11) transloca- information on the karyotype was available. Thus, we tions in infant patients (4/24; 17%), because several cannot exclude the possibility, that some of these eight breakpoints in non-infant patients mapped to the same patients may still have a functional AF4/MLL fusion region (3/34; 9%). gene. However, we were unable to detect it by our In summary, analyses of chromosomal breakpoints approach using speci®c oligonucleotides that bind to in 40 novel t(4;11) patients support the `DNA damage the breakpoint cluster regions of MLL and AF4 repair model for chromosomal translocations t(4;11)' (Reichel et al., 1999). that was proposed after the initial analysis of 20 When both derivative chromosomes were success- patients (Gillert et al., 1999; Reichel et al., 1999). The fully analysed (26 cases=52 alleles), duplications (18 combined results on the distribution of 60 chromoso- alleles), inversions (three alleles) and deletions (32 mal breakpoints in the MLL gene are consistent with alleles) were observed at the chromosomal fusion sites the concept that dierent `DNA break mechanisms' (see Table 2). For UPN041, a perfectly balanced may account for the observed dierences in the translocation between AF4 and MLL was identi®ed. distribution of chromosomal breakpoints in the two This is the ®rst case in our studies, where no patient groups, because the subsequent DNA repair duplications, inversions or deletions were identi®ed at mechanism is mediated for both hotspots by the NHEJ the chromosomal fusion sites of a patient with t(4;11) pathway. This situation might be explained (1) by ALL. dierent chromatin structures at the MLL locus during The presence of ®ller DNA inserts and mini-direct embryogenesis and possibly during hematopoietic repeats at chromosomal fusion sites has been identi®ed development, or (2) by the fact that breakpoints in in a signi®cant portion (71%) of the 66 breakpoint the telomeric portion of MLL may result in an earlier alleles. Thus, it can been concluded that most of these onset of the disease. Further studies on chromatin t(4;11) translocations likely resulted from DNA structures within the MLL and AF4 breakpoint cluster damage that was repaired via the NHEJ repair regions, and studies on disease onsets correlated with pathway. breakpoint distributions will help to unravel this The precise mapping of chromosomal breakpoints phenomenon. performed here con®rmed most of the earlier knowledge about the breakpoint cluster region of the MLL gene. First, the majority of breakpoints were identi®ed within the area ¯anked by the exons 9 and 12. Materials and methods However, the apparent restriction to this area is not absolute, because chromosomal breakpoints (der(11) Patient material and der(4)) were mapped outside of this region in two Leukemic cells used in this study were collected from children cases. Thus, it is unlikely that the MLL exons 9, 10 or enrolled in treatment protocols of the German and Austrian 11 play a crucial role for the encoded fusion proteins BFM- or CoALL-study groups and the USA/Canada derived from the chimeric mRNA species. Children's Cancer Group which had been approved by An important result of this study emerged from the institutional review boards. All patients or their parents/legal breakpoint distribution analysis within the breakpoint guardians gave informed consent for treatment. Genomic cluster regions of the MLL and AF4 genes, where DNA from bone marrow biopsy material was prepared results of this and earlier studies (Gillert et al., 1999; locally and forwarded to Erlangen (contributors were Reichel et al., 1999) were combined to achieve investigators from University Hospitals in Erlangen (Er), Gieûen (Gi), GoÈ ttingen (GoÈ ), Hamburg (Ha), and Munich signi®cant results. We found a signi®cant dierence (Mu) in Germany; foreign contributors were from the between the distribution of chromosomal breakpoints University Hospitals in Vienna, Austria (Vi), Monza, Italy of t(4;11) patients below and above 1 year of age. (Mo), London, UK (Lo), and Parker Hughes Institute St. Thirteen of 24 patients below 1 year of age carried Paul, MN, USA (StP). Patients were given unique patient breakpoints in the telomeric part of the MLL break- numbers (UPN). The age of each patient at the time of point cluster region, while 29/34 patients above 1 year diagnosis (w=weeks; m=months; a=years; inf=patient

Oncogene Age-dependent breakpoint distribution in t(4;11) leukemia cells M Reichel et al 2907 below 1 year of age; ch=patient above 1 year of age; position in the published breakpoint cluster DNA sequences ad=adult) and the origins of the DNA samples are given in of both parental genes. As a result of this comparison, Table 1. duplicated, inverted or deleted areas of the MLL and/or AF4 genes were identi®ed. Mini-direct repeats were assigned when identical DNA sequences were identi®ed at the chromosomal Oligonucleotides fusion site of a derivative chromosome and within the aligned Ampli®cation of translocated genomic DNA fragments was DNA sequences of both wildtype genes. Short patches of performed by using MLL- and AF4-speci®c oligonucleotides ®ller DNA were assigned when unknown DNA sequences as published (Reichel et al., 1999). Some cases investigated by were identi®ed at chromosomal fusion sites that were not other methods have previously been published (Reichel et al., derived from the parental MLL and AF4 genes. Statistical 1998; Gillert et al., 1999). analyses for dierences in breakpoint distributions of der(11) chromosomes were performed by using Student's t-test (2- tailed) and Chi-square analysis (http://faculty.vassar.edu/ Analysis of genomic DNA fragments spanning translocation *lowry/newcs.html). Two t(4;11) patients (UPN036 and breakpoints in t(4;11) DNA 042) were not classi®ed by their age and excluded from all All DNA samples were diagnosed as t(4;11) translocations by statistical analyses. the contributors by using standard cytogenetic and RT ± PCR techniques. As a rule, the presence of an MLL/AF4 fusion Genbank accession codes gene (der(11)) was con®rmed by long range PCR (LR ± PCR) using MLL- and AF4-speci®c oligonucleotides (Reichel et al., Breakpoint data ®les of all 40 patients are available through 1999). For those cases scoring positive in LR ± PCR, DNA the Genbank accession codes AJ408891 to AJ408956. samples were analysed for the presence of an AF4/MLL fusion gene (der(4)). If no der(11) product was detected, then no attempt was made to analyse for the presence of a der(4) product. Negative results for der(11) or der(4) may be due to Acknowledgments the limitations of our method, which allows the identi®cation We are grateful to all physicians who provided DNA from of chromosomal breakpoints between the MLL and AF4 biopsy material of t(4;11) leukemia patients, especially Drs genes only when they were located inside of the known F Cotter (London, UK), B Emmerich (Munich, Germany), breakpoint cluster regions of both genes. M Gramatzki (Erlangen, Germany), and U zur Stadt For all positive cases, the genomic DNA amplimers (Hamburg; Germany). This study was supported by spanning chromosomal breakpoints were sequenced. The research grants SFB 466 from the Deutsche Forschungsge- sequence of each derivative breakpoint was compared with meinschaft (DFG) to J Greil and R Marschalek, research the known wildtype sequences of the breakpoint cluster grant 96.047.3 from the Wilhelm Sander Foundation to regions for MLL (Genbank accession code: X83604) and AF4 GH Fey, J Greil and R Marschalek, and NCI grant CA- (Genbank accession code: AJ238093). Each of the analysed 13539 to FM Uckun. FM Uckun is a Stohlman Scholar of breakpoints was given speci®c coordinates according to the the Leukemia Society of America.

References

Bernard OA and Berger R. (1995). Genes Chrom. Cancer, 13, Nilson I, LoÈ chner K, Siegler G, Greil J, Beck JD, Fey GH 75 ± 85. and Marschalek R. (1996). Br. J. Haematol., 93, 966 ± 972. Cimino G, Rapanotti MC, Biondi A, Elia L, Lo Coco F, Nilson I, Reichel M, Ennas MG, Greim R, KnoÈ rr C, Siegler Price C, Rossi V, Rivolta A, Canaani E, Croce CM, G, Greil J, Fey GH and Marschalek R. (1997). Br. J. Mandelli F and Greaves M. (1997). Cancer Res., 57, Haematol., 98, 157 ± 169. 2879 ± 2883. Rabbitts TH. (1994). Nature, 372, 143 ± 149. Cimino G, Rapanotti MC, Sprovieri T and Elia L. (1998). Reichel M, Gillert E, Nilson I, Siegler G, Greil J, Fey GH Haematologica, 83, 350 ± 357. and Marschalek R. (1998). Oncogene, 17, 3035 ± 3044. Felix CA, Kim CS, Megonigal MD, Slater DJ, Jones DH, Reichel M, Gillert E, Breitenlohner I, Repp R, Greil J, Beck Spinner NB, Stump T, Hosler MR, Nowell PC, Lange BJ JD, Fey GH and Marschalek R. (1999). Cancer Res., 59, and Rappaport EF. (1997). Blood, 90, 4679 ± 4686. 3357 ± 3362. Gillert E, Leis T, Repp R, Reichel M, HoÈ sch A, Stanulla M, Wang J, Chervinsky DS, Thandla S and Aplan Breitenlohner I, AngermuÈ ller S, Borkhardt A, Harbott J, PD. (1997). Mol. Cell. Biol., 17, 4070 ± 4079. Lampert F, Griesinger F, Greil J, Fey GH and Marschalek Strissel-Broeker PL, Super HG, Thirman MJ, Pomykala H, R. (1999). Oncogene, 18, 4663 ± 4671. Yonebayashi Y, Tanabe S, Zeleznik-Le NJ and Rowley Gu Y, Alder H, Nakamura T, Schichman SA, Prasad R, JD. (1996). Blood, 87, 1912 ± 1922. Canaani O, Saito H, Croce CM and Canaani E. (1994). Strissel PL, Strick R, Rowley JD and Zeleznik-Le NJ. (1998). Cancer Res., 54, 2327 ± 2330. Blood, 92, 3793 ± 3803. Leis T, Repp R, Borkhardt A, Metzler M, SchlaÈ ger F, Super HG, Strissel PL, Sobulo OM, Burian D, Reshmi SC, Harbott J and Lampert F. (1998). Leukemia, 12, 758 ± 763. Roe B, Zeleznik-Le NJ, Diaz MO and Rowley JD. (1997). Marschalek R, Greil J, LoÈ chner K, Nilson I, Siegler G, Genes Chrom. Cancer, 20, 185 ± 195. Zweckbronner I, Beck JD and Fey GH. (1995). Br. J. Ziemin van der Poel S, McCabe NR, Gill HJ, Espinosa III R, Haematol., 90, 308 ± 320. Patel Y, Harden A, Rubinelli P, Smith SD, LeBeau MM, Marschalek R, Nilson I, LoÈ chner K, Greim R, Siegler G, Rowley JD and Diaz MO. (1991). Proc. Natl. Acad. Sci. Greil J, Beck JD and Fey GH. (1997). Leuk. Lymphoma, USA, 88, 10735 ± 10739. 27, 417 ± 428.

Oncogene