Characterisation of dic(9;20)(p11–13;q11) in childhood B-cell precursor acute lymphoblastic leukaemia by tiling resolution array-based comparative genomic hybridisation reveals clustered breakpoints at 9p13.2 and 20q11.2
Department of Molecular Medicine and Surgery, Karolinska Institutet, Stockholm
Dr Jacqueline Schoumans, Department of Molecular Medicine and Surgery, CMM L8:02, Karolinska Institutet, SE-171 76 Stockholm, Sweden. E-mail: firstname.lastname@example.org
Although the dic(9;20)(p11–13;q11) is a recurrent chromosomal abnormality in paediatric B-cell precursor acute lymphoblastic leukaemia (BCP ALL), occurring in approximately 2% of the cases, its molecular genetic consequences have not been elucidated. In the present study, high-resolution genome-wide array-based comparative genomic hybridisation (array-CGH) and fluorescence in situ hybridisation (FISH) were used to characterise the 9p and 20q breakpoints (BPs) in seven childhood BCP ALLs with dic(9;20), which was shown to be unbalanced in all of them, resulting in loss of 9p13.2-pter. Five of the cases had loss of 20q11.2-qter, whereas two displayed gain of 20cen-pter. All BPs on 9p clustered in a 1.5 Mb segment of the sub-band 9p13.2; in three of the cases, the 20q BPs mapped to three adjacent clones covering a distance of 350 kb at 20q11.2. Thus, the aberration should be designated dic(9;20)(p13.2;q11.2). One of the ALLs, shown to have a complex dic(9;20), was further investigated by FISH, revealing a rearrangement of the haemapoietic cell kinase isoform p61 (HCK) gene at 20q11. The disruption of HCK may result in a fusion gene or in loss of function. Unfortunately, lack of material precluded further analyses of HCK. Thus, it remains to be elucidated whether dic(9;20)(p13.2;q11.2) leads to a chimaeric gene or whether the functionally important outcome is loss of 9p and 20q material.
The dicentric chromosome abnormality dic(9;20)(p11–13;q11) was first reported as a non-random aberration in B-cell precursor acute lymphoblastic leukaemia (BCP ALL) one decade ago (Rieder et al, 1995; Slater et al, 1995), and to date, almost 50 cases have been published (Mitelman et al, 2006). The reason for the rather recent detection of this recurrent abnormality is undoubtedly that it is a subtle rearrangement that may be mistaken for monosomy 20 by G-banding alone. Thus, fluorescence in situ hybridisation (FISH) analyses are necessary for accurate identification of the abnormality, with dic(9;20) ‘masquerading’ as monosomy 20 and a concomitant deletion of 9p (Rieder et al, 1995; Slater et al, 1995; Heerema et al, 1996; Clark et al, 2000).
Previous FISH analyses, using chromosome painting and centromeric probes, have revealed that the dic(9;20) contains centromeres of both chromosomes 9 and 20 that results in loss of 9p and 20q material (Rieder et al, 1995; Slater et al, 1995; Heerema et al, 1996; Clark et al, 2000). However, the dic(9;20) has, as yet, not been characterised in detail at the molecular level. It is therefore unknown whether it results in a fusion gene, akin to the PAX5/ETV6 chimaera generated by the dic(9;12)(p13;p13) in ALL (Strehl et al, 2003), or whether the pathogenetically important outcome of the dic(9;20) is loss of genetic material, e.g. deletion of tumour suppressor genes. We have carried out a detailed molecular analysis of the dic(9;20) in seven paediatric BCP ALLs, using array-based comparative genomic hybridisation (array-CGH) and locus-specific FISH.
Materials and methods
DNA was extracted from diagnostic bone marrow (BM) cells from seven children with dic(9;20)-positive BCP ALL, treated at the Karolinska University, Lund University and University Hospital of Umeå. The basic clinical and cytogenetic features are summarised in Table I. The project was approved by the Research Ethics Committee at the Karolinska Institutet.
Table I. Clinical and cytogenetic features of the seven dic(9;20)-positive paediatric B-cell precursor acute lymphoblastic leukaemias.
*Abnormalities in bold type were identified and/or revised using spectral karyotyping.
Cytogenetic, FISH and spectral karyotyping analyses
Metaphase chromosome spreads from diagnostic BM samples were prepared and G-banded according to standard procedures. In addition, the ALLs were, as part of clinical routine, screened for BCR/ABL1 [t(9;22)(q34;q11)], ETV6/RUNX1 [t(12;21)(p13;q22)] and TCF3/PBX1 [t(1;19)(q23;p13)] fusions by FISH or reverse transcription-polymerase chain reaction and for MLL rearrangements [11q23 translocations] using FISH or Southern blotting. The dic(9;20) was investigated by FISH in all seven cases using the LSI 9p21/CEP-9 dual colour and the CEP-20 probes according to the manufacturer's instructions (Vysis, Downers Grove, IL, USA). Cases 1, 5 and 6 (Table I) were also analysed using spectral karyotyping (SKY). The FISH and SKY analyses were performed as previously reported (Nordgren et al, 2002).
Further FISH mapping of the breakpoints (BPs) found in the array-CGH analysis (see below) could only be performed on metaphases from case 5 (Table I) due to lack of material in the other cases. In that case, the BP on chromosome 9 was shown to be located in a segmentally duplicated region, visible by pericentric cross-hybridisation of the FISH probes (data not shown). Therefore, seven BP-flanking bacterial artificial chromosome (BAC) clones were selected (tel to cen): RP11-422B15, RP13-198D9, RP11-402N8, RP11-15E6, RP11-73E13, RP11-475B17 and RP11-290L7. For mapping of the 20q BP, the BAC clones (cen to tel) RP1-310O13, RP11-620H13 and RP11-483M19 were applied. To narrow down further the BP on 20q, a small (25 kb) P1-derived artificial chromosome (PAC) clone RP1-180I13 partly overlapping with BAC RP11-620H13 was used. The BACs RP11-183E16 and RP11-434N22 were used for investigations of the PTPRT gene (see below). All clones were obtained from the Sanger Institute (http://www.sanger.ac.uk/cgi-bin/software/archives/new_clone_login.cgi) or CHORI BACPAC Resources (http://bacpac.chori.org). Labelling, hybridisations and FISH analyses were performed as previously described (Schoumans et al, 2005). The signals were visualised using a Zeiss Axioplan 2 fluorescence microscope equipped with a cooled CCD camera (Sensys; Photometrics Ltd, Tucson, AZ, USA), and analysed using the smartcapture software (Vysis). For the interphase studies, at least 200 nuclei per slide were analysed.
One microgram of genomic DNA was labelled using a random labelling kit according to the manufacturer's protocol (Enzo Life Sciences Inc., NY, USA). Normal female or male reference DNA, consisting of a pool of 10 healthy individuals (Promega, GmbH Mannheim, Germany), and patient DNA was differentially labelled with Cy5-dCTP and Cy3-dCTP respectively, pooled, mixed with human Cot-1 DNA, precipitated with NaCl and isopropanol and re-suspended in a formamide-based buffer. The labelling reactions were applied to the arrays, which were then placed in hybridisation chambers (Corning Inc. Life Sciences, Acton, MA, USA) and incubated for 72 h at 37°C in a water bath. The slides were washed in 50% formamide/2 × saline sodium chloride (SSC) at 50°C for 15 min, 2 × SSC/0·1% Tween-20 at 50°C for 10 min and 0·2 × SSC at room temperature for 10 min. The slides were then dipped in water for a few seconds, immediately blow dried with nitrogen, and scanned using the GenePix® Professional 4200A scanner (Axon Instruments, Foster city, CA, USA).
The analyses of microarray images were performed with the genepix pro 6.0 software (Axon Instruments), and the quantified data matrix was loaded into the BioArray Software Environment (BASE) database (Saal et al, 2002). Background correction of Cy3 and Cy5 intensities was calculated using the median-feature and median-local background intensities provided in the quantified data matrix. Within arrays, intensity ratios for individual probes were calculated as background-corrected intensity of patient sample divided by background-corrected intensity of reference sample. Spots with bad morphology or high background were excluded. Lowess normalisation was applied (Yang et al, 2002), and to identify gains and losses throughout the genome the two BASE-implemented automatic BP identification plug-ins CGH-Plotter (Autio et al, 2003) and GLAD (Hupéet al, 2004) were used. The threshold for gains and losses was set to log2 (ratio) of ± 0·2 except for case 2 (Fig 1), which most likely contained many normal cells.
Two of the ALLs displayed dic(9;20) as the sole change (cases 3 and 7; Table I), whereas the remaining five had additional abnormalities, with gains of chromosomes 20 (cases 1 and 6) and 21 (cases 1, 2, and 5) being recurrent. None of the cases harboured BCR/ABL1, ETV6/RUNX1, TCF3/PBX1 or MLL rearrangements. The CDKN2A (‘p16’) gene at 9p21 was heterozygously deleted in all cases except case 7, in which it was homozygously deleted. FISH analyses using the LSI 9p21/CEP9 dual colour and the CEP20 probes revealed the presence of both CEP probes on the dic(9;20), confirming that it was dicentric.
All unbalanced rearrangements identified by G-banding and SKY analyses were confirmed by the array-CGH (Table II). The chromosomal origin of the marker in case 2 could, however, not be ascertained with certainty, but the array-CGH analysis suggested that it contained chromosome 7 material (data not shown). In addition, cryptic losses were identified in all cases except case 6. Apart from deletions involving the IGH@ locus at 14q32.33 (cases 3, 4, 5 and 7), none of these changes were recurrent.
Table II. Chromosomal imbalances detected by array-based comparative genomic hybridisation (array-CGH) in the seven dic(9;20)-positive paediatric B-cell precursor acute lymphoblastic leukaemias.
The array-CGH analyses confirmed that the dic(9;20) was unbalanced, resulting in loss of 9p13.2-pter in all seven cases. The chromosome 20 imbalances were more heterogeneous. Two cases (cases 1 and 6, Tables I and II) had two normal chromosomes 20, leading to gain of 20cen-pter. In the remaining cases, four displayed loss of 20q11.2-qter, whereas one (case 5) had a more complex pattern, with loss of 20q11.2-q12 as well as of 20q13.1-qter (Fig 1).
The array-CGH studies revealed clustered, albeit non-identical, BPs at 9p and 20q in all seven cases (Fig 1 and Table II). In cases 1 and 6, the 9p BPs occurred in a region corresponding to RP11-422B15, whereas the chromosome 20 BPs seemed to be (near) centromeric. In cases 2, 3 and 7, the BPs were located in three adjacent clones (RP11-101J1, RP11-469D03 and RP11-777A17) on 9p and in three adjacent clones on 20q (RP11-815L24, RP11-610D23, and RP11-19D5) (Fig 2). Cases 4 and 5 also displayed similarities. Both had BPs within a region corresponding to the same clone on 9p (RP13-198D9) and in two overlapping clones (RP11-602P9 and RP11-483M19) on 20q; these latter BACs were located between the two clustered BPs in cases 1 and 6 and cases 2, 3 and 7 respectively.
FISH and array-CGH findings in case 5
The array-CGH analysis of case 5 revealed a more complex dic(9;20), in which the chromosome segment 20q12.13, containing only the PTPRT gene, was retained (Figs 3 and 4). That PTPRT was not deleted was confirmed by FISH using RP11-183E16 and RP11-434N22, which partly cover this gene, showing that it (or at least part of it) remained in the BP region.
Further FISH studies identified RP11-73E13, adjacent to RP13-198D9 (Table I), as the 9p BP spanning clone, because this clone showed a clearly weaker signal intensity on the dic(9;20). The overlapping RP11-16E6, as well as the RP11-422B15, RP13-198D9 and RP11-402N8 clones, showed no signals at the 9p BP, while RP11-475B17 and RP11-290L7 displayed normal signals. On chromosome 20, both RP11-620H13 and RP1-180I13 showed weaker signal intensities on the dic(9;20), whereas RP1-310O13 was translocated to the dic(9;20) and RP11-483M19 was deleted (Fig 3). The 25 kb BP spanning PAC clone RP1-180I13, which only covers the 5′ end of the haemopoietic cell kinase isoform p61 gene (HCK), displayed clearly reduced signal intensity, suggesting deletion of part of this gene. Unfortunately, lack of material precluded any further analyses.
The salient results of the present array-CGH and FISH analyses of seven paediatric dic(9;20)-positive BCP ALLs were the high frequency of cryptic deletions, the consistent losses of 9p and 20q material, the clustering of breakpoints both at 9p and at 20q, and the possible rearrangement of the HCK gene at 20q.
All but one of the ALLs harboured chromosomal losses not identified by conventional G-banding analyses (Tables I and II). The only recurrent change was the deletion involving the IGH@ locus at 14q32.33, found in four of the cases. This, most probably, reflects a somatic immunoglobulin rearrangement clonotypic for the leukaemic blasts. Hence, we deem it unlikely that these 14q32 deletions were causally associated with the leukaemia development. The other cryptic deletions involved 4p11-12, 10q11, 12q23-24 and 15q26 (Table II). The molecular genetic consequences and the leukaemogenic impact of these remain to be elucidated, but it may be noteworthy that deletions involving these chromosome bands have been reported in a handful of BCP ALLs previously, although not in the context of dic(9;20) (Mitelman et al, 2006). The CDKN2A gene at 9p21 was deleted in all cases; this was expected considering that the BPs at 9p were proximal to this locus. However, it was homozygously lost in case 7, which hence harboured a cryptic del(9)(p21p21). Previously, a similarly cryptic homozygous CDKN2A deletion has been reported in one dic(9;20)-positive ALL (Andreasson et al, 2000). This ALL subgroup may, to some extent, be associated with homozygous CDKN2A loss. Obviously, this needs to be investigated in a larger series, in which cryptic deletions of this gene should be actively searched for by FISH or molecular genetic means.
The present investigation clearly demonstrates that the dic(9;20) is unbalanced, leading to loss of 9p13.2-pter in all cases and to deletion of 20q11.2-qter in five of them. The two exceptions were cases 1 and 6, which harboured two normal chromosomes 20 (Table I); gain of the entire 20p was seen in these (Fig 1 and Table II). Thus, the pathogenically important outcome of dic(9;20), without additional copies of chromosome 20, may be the simultaneous loss of tumour suppressors or haploinsufficiency of genes located distal to 9p13.2 and 20q11.2, although the possibility of a fusion gene cannot be excluded. In fact, the observed clustering of breakpoints at 9p13.2 and at 20q11.2 not only means that the cytogenetic description of this abnormality should be dic(9;20)(p13.2;q11.2) but also, and more importantly, that it may rearrange genes located at the BPs. However, it should be stressed that although the BPs clustered, they were non-identical (Table II and Figs 2 and 3), being distributed in a 1·5 Mb segment in 9p13.2 and a 350 kb segment in 20q11.2. The identification of several separate breakpoints within the 9p13 region could indicate that it is particularly prone to chromosomal breakage and recombination, something that would explain why 9p is the most common chromosome arm involved in dicentric chromosomes in ALL (Raimondi et al, 2003; Mitelman et al, 2006).
To date, we know of only two dicentric aberrations that have been shown to result in fusion genes, namely the dic(8;11)(p12;q14) [ODZ4/NRG1] in breast cancer (Liu et al, 1999) and the dic(9;12)(p13;p13) [PAX5/ETV6] in BCP ALL (Strehl et al, 2003). Regarding the dic(9;20), a possible target gene in 9p13 could be the PAX5 gene, previously shown to be involved not only in the dic(9;12) but also in various B-cell lymphomas with t(9;14)(p13;q32) (Poppe et al, 2005) and to be of great importance in B-cell differentiation (Busslinger, 2004). However, the 9p BPs in the dic(9;20) cases mapped hundreds of kb distal to PAX5, making it a less likely candidate, although one cannot exclude a position effect leading to aberrant expression of this gene in ALLs with dic(9;20). With regard to possible target genes in 20q11.2, four characterised genes are known to be located in the 350 kb segment to which the BPs clustered in three of the seven cases, namely DNMT3B, CMMD7, BAK1 and MAPRE1 (Fig 2). However, FISH analysis of case 5 (Fig 4) disclosed a possible rearrangement of another gene – the HCK gene, which maps 600 kb proximal of these genes. Unfortunately, lack of material precluded any further studies of this gene, and it remains to be elucidated whether it is involved in a fusion gene or if loss of function is the functionally important outcome. It should be stressed, however, that none of the other dic(9;20)-positive cases harboured BPs close to HCK, as ascertained by array-CGH. This, together with the fact that the BPs were non-identical, although clustered, in the seven investigated cases, may suggest that loss of 9p and 20q material is the pathogenetically important consequence of the dic(9;20)(p13.2;q11.2).
This study was supported by the Swedish Cancer Society, the Swedish Children's Cancer Foundation, the Gustav Vth Jubilee Fund, the Cancer Society of Stockholm, the Swedish Society of Medicine, the Mary Béve Foundation for Paediatric Cancer Research, and the Knut and Alice Wallenberg Foundation via the SWEGENE program. E. Kuchinskaya is a recipient of a grant from the Swedish Institute.