Rapid Isolation of Chromosomal Breakpoints from Patients with t(4;11) Acute Lymphoblastic Leukemia: Implications for Basic and Clinical Research¹

Certain chromosomal translocations are known to be the initial steps of the malignant transformation of hematopoietic cells leading to the development of myeloid and lymphoblastic leukemias and lymphomas (1). Translocations to the MLL³ gene (2, 3) are associated with an AML or ALL disease phenotype. More than 30 different chromosomal translocations to the 11q23 region have so far been identified, all of which are associated with hematological malignancies, and many translocation partner loci have been analyzed at the molecular level (for review see Refs. 4, 5, 6). Among these translocations, the chromosomal translocation t(4;11) is regularly associated with high-risk infant acute pro-B lymphoblastic leukemias and has been proposed to initiate the development of this malignancy (4, 5, 6, 7, 8). Leukemic blasts with chromosomal translocation t(4;11) are highly resistant against current treatment protocols. Although an initial remission can be achieved in about 80–90% of all cases, the relapse rate is exceedingly high. Thus, there is need for a better control of applied therapy. Monitoring of MRD allows the control of the efficacy of induction chemotherapy or subsequent allogeneic stem cell transplantation. Molecular methods for MRD-monitoring of t(4;11)-patients, so far, were based mostly on RT-PCR analysis of breakpoint mRNAs (9). However, RT-PCR experiments using one t(4;11) break-region transcript as the sole molecular marker may generate ambiguous results, as it has been demonstrated recently (10). Derivative (der) 11 breakregion transcripts were identified in normal biopsy material without any accompanying cytogenetic evidence for a chromosomal rearrangement t(4;11). In addition, the RNA may be degraded and steady-state RNA concentration or transcriptional activity may change during therapy, which may negatively affect a precise analysis.

The use of genomic DNA as template for PCR experiments is very attractive because chromosomal fusion sites are characteristic and usually stable features of t(4;11) cells (11). However, this type of analysis, so far, was hampered by the fact that the DNA sequence of the breakpoint cluster region of the human AF-4 gene was not available, and systematic analyses of t(4;11) breakpoints on the genomic DNA level could not be performed. In this study, the complete breakpoint cluster region of the AF-4 gene was subcloned and sequenced. The sequence was used to design sets of specific oligonucleotides, allowing investigators to retrieve chromosomal breakpoint information from t(4;11) patient DNA. This resulted in the assignment of optimally suited DNA primer pairs for each individual patient. This new tool can be applied in any clinical setting and may close a diagnostic gap for this group of high-risk acute leukemias.

Leukemic cells used in this study were collected from children enrolled in treatment protocols of the German BFM-95 study group, which had been approved by institutional review boards. All patients and their parents gave informed consent for treatment and for collection of material for biological studies. Bone marrow biopsy material from patients P. B. (1-year-old male) and E. T. (3-year-old female) was collected at the time of diagnosis. Patient biopsy material was kindly provided by Drs. E. R. Panzer-Grümayer and O. A. Haas from the St. Anna Childrens Hospital (Vienna, Austria). All other patient breakpoint data shown in Fig. 1 and Table 1 were obtained from earlier studies (12).

Breakpoint analysis was performed by amplification of DNA fragments using MLL- and AF-4-specific oligonucleotides (x, directed to the telomer; y, directed to the centromer): MLLx2 (5′-GGAAAAGAGTGAAGAAGGGAATGTCTCG-3′), MLLy7 (5′-CAAAACTTGTGGAAGGGCTCACAACAGACTTGG-3′), AF4x2 (5′-CAGTTTGTACAATGACGACAGAAACCTGCTTCG-3′), AF4x4 (5′-GGCACCAGTGCATCTGCTGTCAACAATGGAG-3′), AF4x6 (5′-GGATGAACATGCGTCAGTCATGCACAGCTCAG-3′), AF4x8 (5′-AGAATTCCCTGCTGCCATGGGTGGTGCCTG-3′), AF4x10 (5′-GGTAGGGTACTGTTACCATCCTGTTTTACTGAG-3′), AF4x12 (5′-CTTGGTATCATCCTGCTGCTTTTTCTCCAGGTG-3′), AF4x14 (5′-CAGTATTTCCTGAGCTAGAGGGCTGTTGTGAG-3′), AF4x16 (5′-CAAAGGTCAAGTACACGACGGCTTCTTCAG-3′), AF4x18 (5′-GTGATGAAAAGCTAATAGGCGAGGACGATAGTG-3′), AF4x20 (5′-GCTGTTGAGATTACTGCTGGAATAAGCCCAGG-3′), AF4x22 (5′-CCCAAACTGCATTTAGGCAACTACTCTTATAGG-3′), AF4x24 (5′-AGAAAAGTGTGACTTATCTAACTAGCGAGTTG-3′), AF4x26 (5′-GTCCTGGTTCATCCAACATAACAAAACTCTTGG-3′), AF4x28 (5′-ATGCCCATTTATGATTTGGAAATCATAACTTGG-3′), AF4x30 (5′-CATGGCCGCCTCCTTTGACAGCAATACATACG-3′), AF4x32 (5′-TGATTTCTTCAACACTTGATCTTATGGTACTGG-3′), AF4y2 (5′-CCTCAACTGACTGAGAAGGCATCTTCAGTTTGG-3′), AF4y4 (5′-GGGTCCATGTGCATGAGCACTCACAGAGATG-3′), AF4y6 (5′-ACTTGGGTCTTGGAAATTAAATGCAAGCATCTG-3′), AF4y8 (5′-GCCAAAGGGATTAAGAAGATTAGAATGGGCTTG-3′), AF4y10 (5′-AGGAACAGTGTTTACCTACTTCTGTCTCCAGAG-3′), AF4y12 (5′-TGGGTCTGCTCAAAGTGAGGAATAATCATACAG-3′), AF4y14 (5′-TTAACCAGTCTTTCCTCCATGATATGAGGACAG-3′), AF4y16 (5′-CACTGTCTCGTGTGCTGGCGGCTGCAATGG-3′), AF4y18 (5′-AAGACTCATCTGATACACTGATTTACTGATCTGG-3′), AF4y20 (5′-CCACATGAGGCCCCATGAAGCCCATTGTGG-3′), AF4y22 (5′-CCTGGCCCAGTGTTCACTTGCTAAACTTCACG-3′), AF4y24 (5′-TCAAGAAAATCTAGTTTGTAAGTCCTTCTCTGG-3′), AF4y26 (5′-CTCTTCAACACAATGGACTTCATTGGAGTAGG-3′), AF4y28 (5′-TGTTTACTGACTGGCCATCTGCTTTGTTCAGG-3′), AF4y30 (5′-GGTTTTGGGTTACAGAACTGACATGCTGAGAG-3′), AF4y32 (5′-CGATGACGTTCCTTGCTGAGAATTTGAGTGAG-3′).

Genomic DNA from 2 × 10⁵ to 2 × 10⁷ cells was purified by ion exchange chromatography (Qiagen Genomic tips 100/G; Qiagen Ltd., Hilden, Germany), according to the manufacturer’s recommendations. The DNA was dissolved in 100 μl of TE buffer (pH 8.0) and its concentration was determined by absorbance at 260 nm. In general, 40–60 μg of high molecular weight DNA (50–100 kb) were obtained from 10⁷ viable cells.

Long-range PCR experiments were performed as published (13). Briefly, oligonucleotide MLLx2 was used in combination with one of the oligos of the AF4y2 to AF-4y32 primer set, whereas MLLy7 was used in combinations with one of the AF4x2 to AF4x32 primers. All DNA amplimers presumably spanning chromosomal breakpoints were cloned and sequenced. All data were compared with the known wild-type sequences of the breakpoint cluster regions of the MLL and AF-4 genes (3, 11, 12, 14, 15, 16) and sequences newly derived in this study. All sequence data were deposited in public databases under the following accession codes: breakpoint cluster region (bcr) of the MLL gene: X83604; patient breakpoints: AJ000166 to AJ000180; Y16596 to Y16609; AJ235330 to AJ235380; Y18922 to Y18929; and breakpoint cluster region of the AF-4 gene: AJ238093.

Large segments of the AF-4 gene had previously been isolated, and two of the AF-4 contigs (contigs 2 and 3) were partly overlapping with the AF-4 breakpoint cluster region (bcr; Ref. 16). The gap between these contigs was closed by screening individual PAC clones using genomic probes from AF-4 contigs 2 or 3. Five PAC clones containing the missing information (RPCI 1, 3–5, 1048b13; RPCI 1, 3–5, 263e20; RPCI 1, 3–5, 666m14; RPCI 1, 3–5, 698g15; and RPCI 1, 3–5, 583j12) were obtained from the German Resource Center in Heidelberg (RZPD). Appropriate restriction fragments were subcloned and finally sequenced on both strands. Where necessary, the orientation of certain genomic fragments was verified by genomic PCR experiments using oligonucleotides specifically binding to terminal regions of subcloned restriction fragments (data not shown). The final sequence has been deposited in public databases (Genbank accession code AJ238093), and the structure of the AF-4 breakpoint cluster region is shown in Fig. 1. The exon/intron structure (AF-4 exons 2–7) and a restriction map for the enzymes BamHI (B), EcoRI (E), and HindIII (H) are indicated.

The complete DNA sequence of the AF-4 bcr (52.909 bp) was analyzed for the presence of repetitive DNA elements by using the Blast2 server from the National Center for Biotechnology Information.⁴ Repetitive DNA element sequences (Alu J: U14567, LINE-1: U93562, and Kpn-1: X03145) were used to identify homologous DNA sequences within the breakpoint cluster region of the AF-4 gene. A total of 26 Alu repetitive elements (19 of 26) or Alu remnants (<250 bp: 7 of 26), but no repetitive LINE-1 or Kpn-1 elements were identified. A search for potential topoisomerase II consensus sequences or nonamer/heptamer signal sequences was performed by using the computer program DNA Inspector II. Due to the highly degenerated character, potential topoisomerase II consensus sequences were only accepted with deviations from the known consensus sequence (5′-RNYNNCNNGYNGGTNYNY-3′; Ref. 17) of not more than one mismatch. Two perfect sites and 30 sites with one mismatch were identified during the analysis of both DNA strands (Fig. 1). Nonamer/Heptamer signal sequences were accepted with up to five mismatches from the known consensus sequences of the canonical 7-23-9, 7-12-9, 9-23-7, or 9-12-7 signal sequences (18). Ten mismatched DNA sequences were identified (three sites with four mismatches; seven sites with five mismatches), which are also shown in Fig. 1. The precise locations of all Alu elements and consensus sequences in the breakpoint cluster region of the human AF-4 gene are summarized in Table 2.

Using the sequence information of the complete AF-4 breakpoint cluster region, 64 different oligonucleotides were designed. These oligonucleotides bind specifically to the nonrepetitive DNA fraction and are suitable for the amplification of reciprocal breakpoints in the genomic DNA of t(4;11)-patients. Two oligonucleotides specifically bind to exons 8 and 14 of the human MLL gene (nomenclature according to Refs. 5 and 15). The oligonucleotide in exon 8 (MLLx2) points to the 3′-end of the MLL gene (downstream oligonucleotide), whereas the oligonucleotide binding exon 14 (MLLy7) points to the 5′-end of the MLL gene (upstream oligonucleotide), respectively. Both MLL-specific oligonucleotides were first used either in combination with 32 downstream oligonucleotides (AF4x1 to AF4x32) or 32 upstream oligonucleotides (AF4y1 to AF4y32) of the AF-4 breakpoint cluster region. In subsequent experiments, only every second primer of the upstream or downstream sets was used. Thus, in a total of only 32 PCR reactions, we were able to amplify der11 and der4 breakpoint fragments for each individual patient. In some cases, there was a redundancy of three oligonucleotide combinations that gave a positive signal in the PCR reaction, either with increasing or decreasing product length of the amplified breakpoints (Fig. 2,A). In other cases, only one or two positive signals were obtained. For each patient, the smallest amplimers from the der11 and der4 breakpoints were subjected to digestions with EcoRI, HindIII, and a combination of both restriction enzymes to narrow down the genomic fusion site in each derivative amplimer (examples shown for two patients in Fig. 2,A). Finally, each breakpoint was analyzed by cloning and sequencing the junction between chromosomes 4 and 11, or vice versa (Fig. 2 B).

Specific genomic alterations were identified at chromosomal breakpoint junctions of t(4;11) cell lines and in patient biopsy material, including duplications, inversions, or deletions of parental DNA directly at the genomic fusion sites of both derivative chromosomes (11, 12). For the two t(4;11) patients described in this study, the same phenomenon was observed (Fig. 2 C). In patient P. B., a 222-bp duplication of parental AF-4 DNA was observed in the der4 and der11 chromosome, whereas a segment of 17-bp parental MLL DNA was lost and neither found in the der4 or der11 chromosome, respectively. In addition, a 4-bp filler DNA segment was found to be inserted between the ALL-1 and AF-4 DNA in the der11 chromosome. In patient E. T., a 9-bp duplication of parental AF-4 DNA was observed in the der4 and der11 chromosomes, whereas 482-bp parental MLL DNA was lost and neither found in the der4 or der11 chromosomes, respectively. In addition, a 2-bp filler DNA fragment was identified directly at the breakpoint junction of the der4 chromosome.

Breakpoints for t(4;11) patients identified and deposited in public databases by other authors are listed below for the sake of completeness. However, these data were not included in the tables and figures, which exclusively show information generated by our own laboratory. The references to the publications first reporting these sites are included in the database entries [AF029698: der11 breakpoint (der11-bp) at AF-4 breakpoint cluster nucleotide (nt) 31.623; AF029699: der4 breakpoint (der4-bp) at AF-4 breakpoint cluster nt 10.779; AF029700: der4-bp at nt 4.500; AF024540: der11-bp at nt 19.487; AF024543: der11-bp at nt 10.853; S79285: der11-bp at nt 33.950; S79287: der4-bp at nt 33.947; AJ000170: der11-bp at nt 28.033; AJ000179: der11-bp at nt 35.611; AJ000180: der11-bp at nt 36.691, fused to the inverted region of nt 36.691 to 36.734].

Two major advances follow from the results of this study: (a) an improved method allowing investigators to diagnose and monitor patients with t(4;11)-ALL more rapidly and more reliably than it was possible so far; and (b) an improved ability to carry out the retrospective analysis of breakpoints in biopsy material and leukemia-derived cell lines. The results of this study provide investigators with an increased probability of actually finding the breakpoint in each individual sample and of correctly reconstructing the detailed recombination events that may have led to a particular translocation. Both advances are achieved by using an optimally suited pair of PCR primers and the PCR amplification of genomic DNA for the analysis of each individual breakpoint. A method for the rapid assignment of the optimally suited primer pair for individual breakpoints is presented in this study.

For routine clinical diagnosis and monitoring of t(4;11) breakpoints, the preferred method, so far, was the RT-PCR method based on the amplification of breakregion mRNAs and corresponding first-strand cDNAs by the use of mixed primer pairs with one primer each derived from exon sequences of the MLL and AF-4 translocation partner genes (9). This method has the advantage of generating rapid information about the location of the breakpoints to the nearest exon, irrespective of their precise location within the large introns in the breakpoint cluster region of the AF-4 gene (15). The disadvantages of this method are that it requires a significant amount of cells (10⁵ to 10⁷) and that these cells must be sufficiently fresh and well preserved to guarantee the integrity of the breakregion mRNA. Due to the relatively large amounts of cells required, this method was often difficult to apply for the purpose of monitoring MRD during the course of a therapy. RT-PCR methods and, in particular, nested RT-PCR methods also sometimes generated results that were difficult to interpret, pointing to the existence of breakpoint RNA species in cases in which no cytogenetic evidence for a t(4;11) translocation was observed (10). The method presented here allows the assignment of breakpoints with comparable speed as RT-PCR methods (i.e., within a couple of hours). It requires fewer cells, and the quality of the biopsy material is less critical than for RT-PCR analysis because genomic DNA is degraded less rapidly than breakregion RNA. Therefore, poorly preserved samples that may no longer be suitable for a RT-PCR analysis may still give useful results with the method presented here. Because it requires fewer cells, this method may also allow MRD monitoring in a greater number of cases than conventional RT-PCR methods. Continuous complete remission of patients with t(4;11)-ALL so far has only been observed in cases in which RT-PCR-based MRD monitoring did not detect any breakregion transcripts. Patients scoring positive by RT-PCR did not remain in complete remission (9). The method presented here can be used to obtain an independent verification of RT-PCR results with equal speed and sensitivity. In some cases this is likely to represent a valuable advantage. This may particularly apply to cases where RT-PCR methods indicated a rearrangement in the absence of cytogenetic evidence for a t(4;11) translocation (10). In such cases, the demonstration of a rearrangement on the genomic DNA level by the methods presented here may add decisive additional evidence.

The application of the method described here to a sizeable cohort of t(4;11) leukemias generated useful new information. First, the distribution of chromosomal breakpoints (Fig. 1 and Table 1) suggests a nonrandom occurrence of translocation-breaks within the AF-4 gene. Preferred sites (clusters) of breakpoints were detected within the 48-kb breakpoint cluster region of the AF-4 gene, which is located between exons 3 and 7 (Fig. 1). A total of 37 breakpoints were mapped in this region, including 33 assigned in previous studies (11, 12). Ten of these breakpoints fell into cluster 1 (∼6 kb), 13 into cluster 2 (∼9 kb), and 10 constituted cluster 3 (∼5 kb). Only the newly assigned breakpoints for patients P. B. and E. T. fell outside of these clusters (Fig. 1). Together, these three clusters covered 22 of the 48 kb of the breakpoint region. The DNA in these clusters was found to contain significantly fewer repetitive Alu elements than the remainder of the breakpoint region. Only nine Alu elements (>250 bp) were found within or overlapping these three clusters, whereas eight Alu elements plus nine Alu remnants (<250 bp) were found in the remainder of the breakpoint region (Fig. 1 and Table 2). Only the breakpoint of patient C. B. on the derivative 4 chromosome (nt 47.869 in the AF-4 breakpoint cluster region) fell directly into an Alu element (nt 47.863–48.164); all other 36 breakpoints were located in nonrepetitive DNA sequences (Tables 1 and 2). Therefore, these new data confirm and extend the conclusion previously drawn, that recombination between Alu elements is unlikely to be the mechanism responsible for a major fraction of t(4;11) translocations (11, 12). Recombination between Alu elements is important, however, in the generation of internal duplications within the MLL gene (19) and deletions in the human BRCA1 and hMLH1 genes (20, 21).

Only one breakpoint was found to coincide with a consensus recognition sequence for topoisomerase II (17). This was the breakpoint on the derivative 11 chromosome of the cell line MV4;11 (nt 43.793). None of the remaining 36 breakpoints (Fig. 1 and Table 1) coincided with topoisomerase II binding sites or signal sequences by V(D)J recombinases (Table 2). Thus, the new data obtained here confirmed and extended previously published conclusions, that the actions of topoisomerase II and V(D)J recombinases are unlikely to be a major component in the events generating t(4;11) translocations (12).

For one patient (AJ000172; Table 1), only the breakpoint on the derivative 11, but not the breakpoint on the derivative 4, chromosome was obtained by applying the new methods described above. This result was unexpected because a derivative 4 breakregion transcript was identified by RT-PCR analysis of biopsy material from the same patient.⁵ A similar result was obtained for patient L. L. M. and the cell strains derived from this patient (11). The breakpoint on the derivative 11 chromosome of this patient was found to reside outside of the presently known breakpoint cluster region of the AF-4 gene. The most plausible explanation of these findings is that in a few rare cases the translocation breakpoints cannot be definitively located by the methods presented here, presumably due to incomplete characterized partial or total duplications of sequences from the wild-type AF-4 gene.⁵

Another interesting set of observations made using the methods described above relates to the distribution of translocation breakpoints within the MLL and AF-4 genes and the correlation between the location of breakpoints and the age of the patient at the time of diagnosis. It has been suggested recently that early infant leukemias have a preferred breakpoint distribution in the 3′-terminal portion of the MLL breakpoint cluster region (22, 23). When we analyzed the current set of data, there was a difference between the distribution of breakpoints of patients below or above 1 year of age, at least for the breakpoints in the MLL gene. Three of four patients below or around 1 year of age have breakpoints in the telomeric part of the MLL breakpoint cluster region and have their breakpoints predominantly in AF-4 cluster 1 (two of three) or cluster 2 (one of three). Eight of 12 older patients have their breakpoints in the centromeric portion of MLL and their corresponding breakpoints distributed in the AF-4 clusters 1 (3 of 8), 2 (3 of 8), and 3 (2 of 8). Older patients with breakpoints in the telomeric portion of the MLL breakpoint cluster region (4 of 12) have their breakpoints in AF-4 cluster 1–2 (1 of 4), 2 (1 of 4), or 3 (2 of 4). This may reflect a specific tendency; however, more data are necessary to validate this observation as a real significant difference.

Finally, the breakpoints of both patients that were reported here for the first time (patients E. T. and P. B.) showed similar characteristic features as the previously published breakpoints (11, 12). Both were not balanced at the fine-structure level and showed deletions and duplications of sequences at the breakpoint. For patients E. T. and P. B., MLL sequences were deleted (482 bp and 17 bp, respectively), and AF-4 sequences were duplicated (9 bp and 222 bp, respectively) at the breakpoints (Figure 2, B and C). Filler DNA segments of 2 bp and 4 bp were also detected at the breakpoints on the derivative 4 and derivative 11 chromosomes, respectively, for both patients. The presence of duplications, inversions, deletions, filler DNA and mini-direct repeat sequences at the translocation breakpoints has previously been reported and was interpreted to indicate that these translocations were likely to result from DNA damage and the subsequent action of the error-prone DNA repair pathway (12). The error-prone pathway is only one of several known DNA-repair mechanisms, and it is not the most frequently used pathway, but a default option that is mainly used when the major pathways are inoperative or impaired. The finding of tell-tale signs for the action of the error-prone pathway rather than the major repair pathways at t(4;11) translocation breakpoints indicates that t(4;11) translocations may preferentially occur in cells in which the major DNA repair pathway is impaired. This is an attractive novel hypothesis, and the new methods presented here provide improved tools for testing its predictions in a large number of leukemia cases.

Pieter de Jong and P. Ioannou are gratefully acknowledged for construction of the RPCI 1, 3-5 Human PAC library at Roswell Park Cancer Institute (Buffalo, NY). We thank Drs. Panzer-Grümayer and Haas from the Childrens Hospital for providing patient biopsy material.