Leukemia (2007) 21, 2137–2144 & 2007 Nature Publishing Group All rights reserved 0887-6924/07 $30.00 www.nature.com/leu
ORIGINAL ARTICLE Combined high-resolution array-based comparative genomic hybridization and expression profiling of ETV6/RUNX1-positive acute lymphoblastic leukemias reveal a high incidence of cryptic Xq duplications and identify several putative target genes within the commonly gained region H Lilljebjo¨rn1, M Heidenblad1, B Nilsson1, C Lassen1, A Horvat1, J Heldrup2, M Behrendtz3, B Johansson1, A Andersson1 and T Fioretos1 1
Department of Clinical Genetics, Lund University Hospital, Lund, Sweden; 2Department of Pediatrics, Lund University Hospital, Lund, Sweden and 3Department of Pediatrics, Linko¨ping University Hospital, Linko¨ping, Sweden
Seventeen ETV6/RUNX1-positive pediatric acute lymphoblastic leukemias were investigated by high-resolution array-based comparative genomic hybridization (array CGH), gene expression profiling and fluorescence in situ hybridization. Comparing the array CGH and gene expression patterns revealed that genomic imbalances conferred a great impact on the expression of genes in the affected regions. The array CGH analyses identified a high frequency of cytogenetically cryptic genetic changes, for example, del(9p) and del(12p). Interestingly, a duplication of Xq material, varying between 30 and 60 Mb in size, was found in 6 of 11 males (55%), but not in females. Genes on Xq were found to have a high expression level in cases with dup(Xq); a similar overexpression was confirmed in t(12;21)-positive cases in an external gene expression data set. By studying the expression profile and the proposed function of genes in the minimally gained region, several candidate target genes (SPANXB, HMGB3, FAM50A, HTATSF1 and RAP2C) were identified. Among them, the testis-specific SPANXB gene was the only one showing a high and uniform overexpression, irrespective of gender and presence of Xq duplication, suggesting that this gene plays an important pathogenetic role in t(12;21)-positive leukemia. Leukemia (2007) 21, 2137–2144; doi:10.1038/sj.leu.2404879; published online 9 August 2007 Keywords: array CGH; expression profiling; ETV6/RUNX1; SPANXB
Introduction Acute lymphoblastic leukemia (ALL) is the most common malignancy in children, with an incidence of 3.9 cases per 100 000 per year.1 In pediatric B-cell precursor ALL, the most common structural genetic abnormality, present in 20–25% of the cases, is the t(12;21)(p13;q22), which results in the ETV6/ RUNX1 (also known as TEL/AML1) fusion gene.2–5 ETV6/ RUNX1-positive ALLs are generally associated with a favorable outcome with an event free survival approaching 90%,6 although late relapses have been noted by several groups.7,8 Several lines of evidence suggest that ETV6/RUNX1 is not sufficient for leukemic transformation, such as detection of the gene fusion in Guthrie cards from patients who later developed ETV6/RUNX1-positive ALLs,9 identical gene fusions in twins with concordant leukemia,10 identification of the chimeric transcript in normal cord blood samples11 and lack of leukemia Correspondence: H Lilljebjo¨rn, Department of Clinical Genetics, Lund University Hospital, Lund SE-221 85, Sweden. E-mail:
[email protected] Received 13 March 2007; revised 8 June 2007; accepted 28 June 2007; published online 9 August 2007
in mouse models.12 Thus, although t(12;21)-positive preleukemic clones, at least in some cases, arise already in utero, additional mutations are most likely required for overt leukemia. Genetic changes secondary to ETV6/RUNX1 are found in more than 80% of t(12;21)-positive ALLs,13 the most common being deletion of the normal ETV6 gene, seen in 70% of cases investigated by fluorescence in situ hybridization (FISH).13 Other frequent aberrations include duplication of the normal (20%)13 or derivative chromosome 21 (10%),13 deletion of 6q (18%)14 and deletion of 9p (7%).14 Apart from the additional 12p and 21q abnormalities, often found by FISH with probes specific for ETV6 and RUNX1, most of the secondary abnormalities have been detected by chromosome banding analyses alone, or in some instances by multicolor FISH.15–17 Because of the limited resolution level of these methods, many small–and all cytogenetically cryptic–genomic imbalances of possible prognostic and pathogenetic importance have so far gone undetected. In the present study, 17 pediatric ETV6/RUNX1-positive ALLs were investigated with high-resolution array-based comparative genome hybridization (array CGH), making it possible to detect genomic imbalances as small as 100 kb. The array CGH profiles were also compared with gene expression patterns, showing that the genomic imbalances had a strong impact on the expression level of the affected genomic regions. The most frequent genomic imbalances detected were gain of Xq material and duplication of the der(21); notably, the X duplications were found only in male patients. Overexpression of genes in the duplicated Xq region was observed in cases with dup(Xq), and this was validated in an externally produced independent data set. By further studies of the proposed function and expression profile of genes mapping within the minimally gained Xq region, a number of genes, among them the testis-specific gene SPANXB, that possibly are involved in the transformation of preleukemic t(12;21)-positive cells to overt leukemia were identified.
Patients, materials and methods
Patients ETV6/RUNX1-positive ALLs with available DNA were selected from a larger series of pediatric leukemias previously studied by cDNA microarray.18 Seventeen children (11 boys and 6 girls; median age 6 years, range 1–14 years) with ALLs harboring the t(12;21), as ascertained by reverse transcription polymerase chain reaction (RT-PCR) and/or FISH, were included in the study
Analyses of ETV6/RUNX1-positive ALL reveal frequent Xq duplications H Lilljebjo¨rn et al
2138 (Table 1); 14 samples were obtained at diagnosis and 3 (nos. 3, 6 and 12) at relapse. All cases were analyzed cytogenetically using standard protocols and were molecularly screened for the presence of MLL rearrangements and BCR/ABL1, TCF3/PBX1 (also known as E2A/PBX1) and ETV6/RUNX1 fusion genes at the Department of Clinical Genetics, Lund, Sweden, as a part of routine diagnostic procedures. The study was approved by the Research Ethics committee of Lund University.
Array CGH
The array CGH was performed as previously described.20,21 DNA was analyzed using 32K array slides containing 32 433 bacterial artificial chromosome (BAC) clones (BACPAC Resources, Oakland, CA, USA) covering 98% of the human genome at a 100 kb resolution (Swegene DNA Microarray Resource Center, Lund University). Detailed information is available in Supplementary Information.
cDNA microarray The cDNA microarray analyses have previously been reported.18 In short, samples from 87 B-lineage ALLs, of which 20 harbored the ETV6/RUNX1 fusion gene, were hybridized to 27K microarray slides containing 25 648 cDNA clones (Swegene DNA Microarray Resource Center) representing 13 737 Unigene clusters and 11 592 Entrez gene entries, according to Unigene build 188. RNA extraction, amplification, labeling, hybridization, scanning, post-hybridization washing and feature analysis were performed as described.18
Data analyses The images from the CGH and gene expression arrays were analyzed using the GenePix Pro 4.0 software (Axon Instruments, Foster City, CA, USA), and the obtained data matrices were uploaded to the BioArray Software Environment. Detailed information is available in Supplementary Information.
were t(12;21)-positive, was used.22 The high hyperdiploid cases (n ¼ 17) were excluded for reasons outlined in ‘Supplementary Information, Data analyses’. Since no information on gender was available in the external data set, hierarchical clustering analysis of seven top ranking genes determined by significance analysis of microarrays from a k-nearest neighbor classifier developed by us23 was used to determine the sex of each case.
RT-PCR, RQ-PCR, sequencing and bisulfite sequencing Expression of the SPANX, HMGB3, FAM50A, HTATSF1 and RAP2C genes was investigated by RT-PCR and real-time quantitative RT-PCR (RQ-PCR). DNA fragments amplified with SPANXB-specific primers were sequenced using the BigDye Terminator v1.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) with the same primers as in the PCR and analyzed on an ABI PRISM 3100-Avant genetic analyzer (Applied Biosystems). SPANXB promoter methylation was studied using bisulfite sequencing of genomic DNA. Detailed information for the RT-PCR, RQ-PCR and bisulfite sequencing is available in Supplementary Information.
FISH The array CGH findings for chromosomes X, 6 and 11 (see Results) were confirmed in three cases using metaphase FISH. Lack of material precluded FISH analyses of the three remaining cases with dup(Xq). BACs (BACPAC resources) and whole chromosome paint probes for 6, 11 and X were used (Abbott, Abbott Park, IL, USA). ETV6/RUNX1-positive metaphases were identified using the dual-color translocation probe LSI TEL/ AML1 ES (Abbott).
Results
Abnormalities revealed by array CGH External data set To validate our findings in ALLs with ETV6/RUNX1, an external data set consisting of 118 pediatric B-lineage ALLs, of which 20 Table 1
In total, 45 imbalances (median size 24 Mb; range 1–181 Mb), excluding copy number polymorphisms, were found in 14 (82%) of the 17 cases (Table 2). The imbalances comprised 17
Cytogenetic findings in the 17 ETV6/RUNX1-positive ALLs
Case no.
Sex
Age (years)
1 2 3 4 5 6a 7 8 9 10 11 12
M M M F M M M M M M F F
5 6 9 1 6 14 8 5 5 7 4 7
13 14 15 16 17
F F M F M
6 3 9 7 4
Karyotype 46,XY,t(12;21)(p13;q22)/47,idem,+16/47,idem,der(6)t(X;6)(?;q1?),+16 No mitoses 47,XY,t(12;21)(p13;q22),+der(21)t(12;21),inc/46,XY ??,X?,del(12)(p13p13),t(12;21)(p13;q22) ??,X?,t(12;21)(p13;q22),+der(21)t(12;21) 46–48,XY,+5,add(11)(p1?5),del(12)(p11),t(12;21)(p13;q22),inc/46,XY 47,XY,t(12;21)(p13;q22),+21 ??,X?,der(6)t(X;6),t(12;21)(p13;q22) 46,XY,t(12;21)(p13;q22),add(16)(q21)/46,XY 46,XY,der(12)t(12;21)(p13;q22),ider(21)(q10)t(12;21) 46,XX,del(12)(p13p13),t(12;21)(p13;q22)/46,XX 44,X,X,?der(1)add(1)(p36)t(1;21)(q?;q?),4,5,del(6)(q?21),add(11)(p11),12, del(12)(p13p13),der(21)t(12;21)(p13;q22),+der(?)t(?;12)(?;p?),+der(?)t(?;12)(?;q?),+der(?)t(?;21)(?;q?),inc/46,XX 46,XX,del(6)(q21),t(12;21)(p13;q22)/47,XX,t(12;21),+21/47,XX,t(12;21),+der(21)t(12;21) 47–48,XX,t(12;21)(p13;q22),+21,inc/46,XX ??,X?,t(12;21)(p13;q22) 46,XX,t(12;21)(p13;q22)/46,XX ??,X?,der(11)t(X;11),t(12;21)(p13;q22),+der(21)t(12;21)
Abbreviations: ALLs, acute lymphoblastic leukemias; array CGH, array-based comparative genomic hybridization; FISH, fluorescence in situ hybridization; M, male; F, female. Abnormalities identified by FISH following array CGH are indicated in bold type. a The karyotype of case 6 has previously been reported.19 Leukemia
Analyses of ETV6/RUNX1-positive ALL reveal frequent Xq duplications H Lilljebjo¨rn et al
2139 deletions and 28 gains, of which 13 deletions and 18 gains were not described in the original karyotypes. Of these, 12 imbalances were missed because of karyotypic failures (cases 2, 8, 15 and 17) and 6 because they were too small (1–12 Mb) to be identified by conventional chromosome banding. The remaining 13 imbalances, varying in size between 17 and 155 Mb, most likely escaped detection owing to either poor chromosome morphology or similar banding patterns of the involved chromosomes. Small deletions of 1 Mb at 2p11.2 and 14q32.33 were found in most cases; these regions contain the IGK@ and IGH@ loci, respectively, and the deletions are most likely the result of somatic immunoglobulin rearrangements clonotypic for the leukemic blasts.20
Table 2 Case no. 1
2
3 4 5 6
7 8
9 10
11 12 13 14
15 16 17
Gain of whole chromosomes was observed in four cases: þ X in case 14, þ 5 in case 6, þ 16 in cases 1 and 7, and þ 21 (or rather 21q; gain of 21p cannot be identified by array CGH) in cases 7 and 14 (Figure 1; Table 2). Deletions of 9p were observed in two (12%) of the cases (nos. 1 and 6). Case 1 had a 17 Mb hemizygous deletion between 9p21.2 and 9p23 with a 3 Mb homozygous deletion in 9p21.3. Case 6 had a 3 Mb hemizygous deletion between 9p21.3 and 9p22.1 with a 1 Mb homozygous deletion in 9p21.3. In both cases, the homozygous deletion included the tumor suppressor CDKN2A. Hemizygous 12p deletions, including at least 50 genes as well as ETV6 and CDKN1B, were found in three (18%) of the cases (nos. 4, 6 and 11; Figure 1; Table 2).
Genomic imbalances in 17 ETV6/RUNX1-positive ALLs Chromosome segments involved in imbalances 6q14.1–q27 9p21.2–p23 9p21.3–p21.3 16p13.3–q24.3 Xq21.31–q28 11q22.3–q25 12p13.2–p13.33 21q21.2–q22.12 Xq21.32–q28 12p13.2–p13.33 21q11.2–q22.12 Xq25–q28 12p12.1–13.33 12p13.2–p13.33 21q11.2–q22.12 5p15.33–q35.3 9p21.3–p22.1 9p21.3–p21.3 12p13.2–p13.33 12p12.1–p13.2 21q11.2–q22.12 10p11.1–p15.3 16p13.3–q24.3 21p11.2–q22.3 5q34–q35.3 6q14.1–q27 11q14.1–q25 Xq21.31–q28 Xq25–q28 6q14.1–q22.1 6q24.3–q27 8q13.3–q24.3 13q13.1–q31.1 12p13.2–p13.33 21q11.2–q22.12 12p12.3–p13.2 19q13.32–q13.33 No imbalances No imbalances 18p11.23–18p11.32 18p11.23–q23 21p11.2–q22.3 Xp22.33–q28 12p13.2–p13.33 21q11.2–q22.12 No imbalances 11q14.3–q25 Xq25–q28
Position (Mba)
Size (Mb)
82–171 10–27 20–23 0–89 89–155 106–134 0–12 26–35 92–155 0–12 14–35 123–155 2–22 0–12 14–35 0–181 19–22 21–22 0–12 12–25 14–35 0–39 0–89 14–47 166–181 82–171 84–134 87–155 118–155 77–115 147–171 72–146 32–84 0–12 14–35 11–17 52–55
Gain/loss
Identified by G-banding/FISHb
89 17 3 89 66 29 12 9 63 12 21 32 20 12 21 181 3 1 12 13 21 39 89 33 15 89 50 68 37 38 24 74 52 12 21 6 3
L L L G G L G G G G G G L G G G L L G L G G G G G L L G G L L G L G G L L
Yes No No Yes No No No No No Yes Yes No Yes Yes Yes Yes No No No Yes No No No Yes No No No No No No No No No Yes Yes Yes No
0–7 8–76 14–47 0–155 0–12 14–35
7 68 33 155 12 21
L G G G G G
No No Yes No No No
89–134 123–155
45 32
L G
No No
Abbreviations: ALLs, acute lymphoblastic leukemias; BAC, bacterial artificial chromosome; FISH, fluorescence in situ hybridization. Gains involving Xq are indicated in bold type. According to BAC mapping provided by BACPAC. b FISH with locus-specific probes for the ETV6 and RUNX1 genes. a
Leukemia
Analyses of ETV6/RUNX1-positive ALL reveal frequent Xq duplications H Lilljebjo¨rn et al
2140
Figure 1 DNA copy number heat map based on array CGH of 17 ETV6/RUNX1-positive childhood ALLs. Each row represents the genome-wide copy number profile in respective case, indicated on the left. Data from the 32 433 BACs were ordered based on the chromosomal position of the BAC; chromosome numbers are indicated above the heat map. Red indicates gain of material, green indicates loss of material and black indicates no change from reference as illustrated in the test/reference log2-scale indicated on top left. (a) DNA copy number changes in 11 ALLs from male patients. Several imbalances previously described in t(12;21)-positive ALLs, such as an extra der(21)t(12;21) and deletions of 6q, 9p and 12p, are seen. Gain of the distal part of Xq is seen in six cases (cases 1, 2, 3, 8, 9 and 17). (b) Partial gain of Xq is not present in any of the female cases. Abbreviations: ALLs, acute lymphoblastic leukemias; array CGH, array-based comparative genomic hybridization; BACs, bacterial artificial chromosomes.
A 12 Mb duplication distal of 12p13.2 was present in six (35%) of the cases (nos. 2, 3, 5, 6, 10 and 15). In all these, gain of 21q was also seen, a finding consistent with an additional copy of the der(21)t(12;21), as also verified by FISH in cases 3, 5 and 10. In five of the cases (all except no. 2), the gained 21q material spanned at least 21 Mb between 21q11.2 and q22.12. As only a few BACs map to 21p, this region was not informative. In case 2, the gain involved a smaller 9 Mb region between 21q21.2 and q22.12; hence, not the entire der(21)t(12;21) was duplicated (Figure 1; Table 2). The array CGH confirmed 12 (86%) of the 14 imbalances identified cytogenetically in cases 1–11 and 14–17 (see below for nos. 12 and 13; Figure 1; Table 2). Only the two imbalances resulting from the þ der(21)t(12;21), inferred from interphase FISH analysis, in case 17 could not be confirmed by array CGH. The þ der(21) may have been present only in a small subclone or the duplicated fusion gene did not arise through gain of the entire derivative chromosome 21. As regards cases 12 and 13, none of the imbalances identified by G-banding and FISH was seen using array CGH, suggesting a selection of cells with secondary chromosomal changes during culture before the cytogenetic analysis.
Array CGH reveals duplication of Xq material in males only The previously unrecognized abnormality dup(Xq) and þ der(21)t(12;21) were the two most common aberrations found by array CGH; they were both detected in six (35%) of the cases. Interestingly, all patients with dup(Xq) were male. Thus, 6 (55%) of 11 boys displayed this abnormality (nos. 1, 2, 3, 8, 9 and 17; Table 2). In contrast, none of the six girls had a similar gain, although þ X was seen in one of them (no. 14; Figure 1; Table 2). The Xq gains had not been detected in any of the cases where G-banding had been performed; in hindsight, however, the extra Xq material in case 9 (Table 1) is most likely on the add(16)(q21). To investigate the Xq gain further, cases with available material were studied with FISH, revealing that Leukemia
the additional chromosome X material was localized on chromosome 6 (in cases 1 and 8) (Supplementary Figure 1) and on chromosome 11 (in case 17). Array CGH showed that both the cases with chromosome 6 involvement had an 89 Mb loss of material on this chromosome. Deletions involving 6q were also present in case 10; this case, however, lacked gain of Xq (Figure 1; Table 2). The gain of material on chromosome X and loss of material on chromosome 11 in case 17 were less evident, with log2 values around 0.2, possibly indicating that these imbalances were not present in all cells (Figure 1). Taken together, the FISH and array CGH results indicated the presence of a der(6)t(X;6)(q21.31;q14.1) in cases 1 and 8 and a der(11)t(X;11)(q25;q14.3) in case 17. The breakpoints on 6q in nos. 1 and 8 were both mapped to the adjacent BACs RP11705B7 and RP11-684M3, suggesting similar if not identical breakpoints at 6q. This region, however, does not contain any named genes. In addition to case 17, with the der(11)t(X;11), two cases (nos. 2 and 8) had hemizygous deletions of 11q, both accompanied by gain of Xq material. Case 2 displayed a 29 Mb deletion distal of 11q22.3, whereas case 8 had a 50 Mb deletion starting at 11q14.1 (Figure 1; Table 2). Unfortunately, no material from case 2 was available for FISH analysis; thus, a possible presence of a der(11)t(X;11) in this case could not be ascertained. In case 8, the extra Xq material was part of the der(6)t(X;6) described above. Only two of the six cases with gain of Xq material had similar proximal Xq breakpoints, but all gains included the telomeric end of Xq. The largest duplication was 68 Mb (Xq21q28) in case 8, and the smallest was 32 Mb (Xq25q28) in cases 3 and 17.
Impact of genetic imbalances on gene expression Visualization of the gene expression data using the CGH-plotter provided a tool for observing the effect of genomic imbalances on gene expression levels. This analysis showed that both deletions and duplications had a strong impact on the expression level of genes in the affected regions, as exemplified
Analyses of ETV6/RUNX1-positive ALL reveal frequent Xq duplications H Lilljebjo¨rn et al
2141 in case 1 with a der(6)t(X;6) (Supplementary Figure 2). The strong impact of the Xq duplication on gene expression was further confirmed by plotting the distribution of t-statistics for genes in the minimally gained Xq region for t(12;21)positive cases both with and without gain of Xq (Supplementary Figure 3). Data from the cDNA microarrays were also used to study the expression of suggested tumor suppressor genes in commonly deleted regions. A low expression level of CDKN2A was found in the two cases with homozygous 9p deletions, but also in nine additional cases. Two of the three cases with 11q deletions had analyzable expression data for the ATM gene, suggested to be the critical target of 11q deletions in chronic B- and T-cell neoplasms.24 Both displayed a low expression of this gene. In the three cases with 12p deletions, ETV6 was underexpressed, whereas low expression of CDKN1B was seen in only one of the cases. An additional three cases (nos. 9, 13 and 16) also displayed a low ETV6 expression; array CGH data for these cases showed log2 ratios around 0.2 for BACs near ETV6, suggesting deletions not present in all cells or small intragenic deletions.
the significantly higher expression of SPANXB (P ¼ 0.00078), HMGB3 (P ¼ 0.033) and FAM50A (P ¼ 0.023) in t(12;21)positive ALLs compared to t(12;21)-negative (one-sided Wilcoxon rank sum test; Supplementary Figure 5). In particular, expression of SPANXB was associated with t(12;21) status as this gene was clearly expressed in all t(12;21)-positive cases, but was not, or only barely, expressed in the other ALLs (Supplementary Figure 4). To investigate if the expression of SPANXB was regulated by promoter methylation, we examined the methylation status of the SPANXB promoter region in both SPANXBpositive and -negative cases using bisulfite sequencing. Previously, the expression of SPANXB, which in normal tissues is confined to testis, was shown to be regulated by demethylation of specific CpG dinucleotides within a 500 bp region of the SPANXB promoter.25 As expected, we observed that the CpG dinucleotides in the promoter region of the SPANXBnegative cases were all methylated. However, despite the observed expression of SPANXB in the t(12;21)-positive ALLs, no demethylation was seen in these cases, regardless of sex. This suggests that expression of SPANXB is regulated by other CpG dinucleotides or yet to be determined regulatory elements in the vicinity of the SPANXB promoter.
Integrative genomics identifies several putative target genes within the commonly gained Xq region
Discussion
Given the frequent occurrence (55%) of Xq gain in males and the strong impact of genomic imbalances on gene expression, we studied the expression of genes on the X chromosome by plotting smoothed (25-probe sliding mean) z-scores (Figure 2a). The six cases (nos. 1, 2, 3, 8, 9 and 17; Tables 1 and 2) with gain of Xq displayed a distinct overexpression of genes within the duplicated segment. An overexpression of genes on the entire X chromosome in case 14, with gain of the whole chromosome by array CGH, was also seen. To validate the common overexpression of Xq genes in t(12;21)-positive ALLs, an external gene expression profiling data set was used.22 In this, 7 of the 20 ETV6/RUNX1-positive cases had an enrichment of relatively overexpressed genes, consistent with a gain of Xq, in a region coinciding with the minimally 32 Mb gained region identified by array CGH (Figure 2b). Notably, six of the seven patients were predicted to be males using an expression based model (see Materials and methods), further demonstrating the male predominance of this abnormality. The external data set contained 127 of the 180 genes with assigned gene symbols in the minimally gained region, and t-statistics on the 127 genes revealed that the most differentially expressed genes between leukemias positive and negative for ETV6/RUNX1 were several members of the highly homologous SPANX family of genes (unknown function) along with the genes HMGB3 (transcription factor), FAM50A (unknown function), HTATSF1 (transcription elongation factor) and RAP2C (member of RAS oncogene family) (Supplementary Table 1). A similar overexpression, specific for t(12;21)-positive cases, was seen in our expression data18 for the genes HMGB3 and FAM50A, but not HTATSF1 and RAP2C. None of the SPANX genes were, however, present on our microarrays.18 We therefore used RT-PCR and sequencing to identify which of the known SPANX genes were expressed in five ETV6/RUNX1positive ALLs. SPANXB was clearly expressed in all t(12;21)positive cases examined (Supplementary Figure 4), but none of the genes SPANXA1, SPANXA2, SPANXC and SPANXD were expressed in any of the cases (data not shown). We then examined the expression of SPANXB, HMGB3, FAM50A, HTATSF1 and RAP2C in eight t(12;21)-negative cases and five t(12;21)-positive cases using RQ-PCR. This analysis confirmed
To characterize molecularly pediatric ALLs with ETV6/RUNX1, 17 cases were investigated with genome-wide high-resolution array CGH and cDNA microarray gene expression analyses. In accordance with previous studies, an additional copy of the normal or derivative chromosome 21 and deletions at 12p, 9p, and 6q13,14,26 were identified as common abnormalities. The expression levels of suggested target genes in these regions were also found to be altered. The two most common genomic imbalances in this study were duplication of the der(21)t(12;21) and partial gains of Xq. As regards X chromosome changes in t(12;21)-positive ALLs, previous studies have mainly reported a high frequency of losses, not gains.14 To the best of our knowledge, only eight ETV6/RUNX1-positive ALLs with partial Xq gains, as identified by metaphase CGH or multicolor FISH, have been described.15–17,27–29 All the gains in these cases were generated through unbalanced translocations involving Xq, none of which were seen by conventional cytogenetics. Notably, all cases with Xq gain in the present study were males, as were six of the previously published cases;16,17,27–29 the gender of the remaining two cases was not reported.15 Hence, this abnormality seems to be quite specific for ETV6/RUNX1-positive ALLs occurring in boys. Of the six cases in the present study, two had similar unbalanced translocations between chromosomes X and 6, and one had an unbalanced der(11)t(X;11). According to the Mitelman database of chromosome aberrations in cancer,26 only four such translocations have previously been described in childhood ALL.15–17,30 Interestingly, three of these cases were ETV6/RUNX1 positive.16,17,30 None of these abnormalities had been identified by G-banding; hence, it is possible that, owing to the similar banding pattern of the involved chromosome arms in combination with the poor banding quality often found in childhood ALL, der(6)t(X;6) and der(11)t(X;11) may go undetected unless specifically looked for by FISH. While this manuscript was being prepared for submission, a large study of childhood ALL using high-resolution singlenucleotide polymorphism arrays was published, where alterations of genes important for B lymphocyte development and differentiation were found in 40% of B-progenitor ALLs.31 Small Leukemia
Analyses of ETV6/RUNX1-positive ALL reveal frequent Xq duplications H Lilljebjo¨rn et al
2142
deletions of the PAX5 gene, the majority below the resolution of the platform used in the present study, were found to be the most common alterations. As in our study, deletions at 12p, 11q, 6q and 9p were found to be common in t(12;21)-positive ALLs. Leukemia
Gain of Xq was not discussed as a common finding in ETV6/ RUNX1-positive ALLs, but on careful examination we note that Xq gain was present in 10 (21%) ETV6/RUNX1-positive cases, confirming the high prevalence of Xq duplications in this genetic
Analyses of ETV6/RUNX1-positive ALL reveal frequent Xq duplications H Lilljebjo¨rn et al
2143 Figure 2 Relative overexpression of genes mapping to Xq25–q28 in t(12;21)-positive ALLs. The relative expression of genes located on the X chromosome in B-lineage ALLs in the data sets of Andersson et al.18 (a) and Ross et al.22 (b–d) is presented graphically. The genetic subtype and the gender of the ALL patients in the respective data sets are indicated above images (a) and (b). NK denotes normal karyotype and n/a not available. The color scale below each heat map represents z-scores; red indicates positive values, green indicates negative values and black indicates values close to zero. (a) To emphasize spatially localized expression patterns in 58 B-lineage ALLs in the data set of Andersson et al.,18 smoothed z-scores (25-probe sliding mean) were plotted for all cases. The six cases where dup(Xq) was identified by array CGH are indicated above the heat map. In all cases with gain of Xq material, the gain results in an enrichment of relatively overexpressed genes located on distal Xq, as indicated by red segments in these cases. (b) The mean z-scores over 25 probe sets were plotted for 101 B-lineage ALLs in the data set of Ross et al.22 The seven t(12;21)-positive cases indicated by arrows exhibit an enrichment of relatively overexpressed genes distal of Xq25, consistent with a gain of Xq material. Six of these cases were predicted to be males. (c) Enlarged view of the genes located distal of Xq25, indicated by * in image (b), at the single gene level. A probe set (220922_s_at) detecting several genes in the SPANX family is marked by an arrow. This probe set displays a significant overexpression in t(12;21)-positive cases, regardless of gender. (d) The relative expression detected by the probe set is plotted for each case. The y-axis indicates z-scores, whereas each case is separated on the x axis as in images (b) and (c). Cases negative for t(12;21) are colored green and cases positive for t(12;21) are colored red. A t(12;21)-specific overexpression of one or more of the genes recognized by this probe set is clearly indicated. Abbreviations: ALLs, acute lymphoblastic leukemias; array CGH, array-based comparative genomic hybridization.
subtype. Interestingly, we found that PAX5 alterations and Xq duplications were mutually exclusive in this data set, possibly indicating that the two changes represent different genetic pathways of disease evolution in t(12;21)-positive ALL. The high prevalence of Xq gains indicates that one or several genes in the 32 Mb minimally gained region are important for leukemogenesis. The striking lack of this abnormality in t(12;21)-positive ALLs in girls may be explained by the special characteristics of the X chromosome, where one of the X chromosomes is inactivated in normal female cells.32 It is tempting to speculate that re-activation of one or several genes on the inactive X chromosome in girls could result in similar effects as in boys with gain of Xq. Interestingly, re-activation of a subset of genes on the inactive X chromosome has recently been described as an important mechanism in basal-like breast cancer.33 By investigating the expression of genes in the minimally gained region in the data sets produced by us18 and Ross et al.,22 we could show that several genes were specifically expressed in ETV6/RUNX1-positive ALLs. These included a number of genes in the highly homologous SPANX family (normally expressed only in spermatozoa) and the genes HMGB3 (transcription factor affecting hematopoietic stem cell fate and Wnt signaling), FAM50A (unknown function), HTATSF1 (transcription elongation factor) and RAP2C (member of the RAS oncogene family). Because we deemed it unlikely that t(12;21)-positive ALL in males would evolve through different mechanisms than in females, we focused on genes on Xq that also showed upregulation among females. Such an expression pattern was, in particular, exhibited by the genes in the SPANX family. Using RT-PCR, we could further show that SPANXB, and not the other SPANX family members, was expressed in all t(12;21)-positive cases. The genes in the SPANX family share more than 90% sequence identity; therefore, the detection of SPANX genes other than SPANXB in the gene expression profiling data set by Ross et al.22 is most likely the result of cross-hybridization with SPANXB. Although it is presently unclear which gene(s) within the commonly gained Xq region that are important in the leukemogenic process, we conclude that of the genes identified, SPANXB provides an attractive target of the dup(Xq). This gene is normally expressed only in spermatozoa and distinguishes itself as the most differentially expressed gene between t(12;21)positive and -negative ALLs. We suggest that this could be the result of a gain of Xq material, followed by a clonal selection of cells showing an upregulation of SPANXB, in a majority of males. In females, a re-activation on both alleles would be an alternative way by which SPANXB becomes upregulated. It is, in this context, noteworthy that SPANXB expression has previously
been shown to be primarily regulated by promoter methylation,25 providing a mechanism for the re-activation in females. However, examining the methylation status of the SPANXB promoter in both t(12;21)-positive and negative cases, we observed no correlation between promoter methylation and SPANXB expression. This does not, however, exclude the possible involvement of CpG dinucleotides or yet to be determined regulatory elements in the vicinity of the SPANXB promoter as important regulators of SPANXB expression. Sporadic expression of SPANXB has previously been demonstrated in cancer cell lines and various hematologic malignancies, such as multiple myeloma, chronic lymphocytic leukemia, chronic myeloid leukemia and acute myeloid leukemia, but it has never been linked to any specific genetic subgroups.25,34 So far, the function of SPANXB or the other SPANX genes has not been studied in detail, but all five SPANX genes are present on the 750 kb Xq region linked to prostate cancer susceptibility, and deregulation of either one has been suggested as a possible mechanism for malignant transformation of prostate cells.35 This is supported by experimental models showing that ectopic expression of SPANXC induces transformation in mammalian cells.35 The association between SPANXB expression and presence of ETV6/RUNX1 may also be of importance for future treatment modalities, as SPANXB, normally present only in the immune-privileged testis, has been suggested to be an ideal target for tumor immunotherapy.34
Acknowledgements This work has been supported by grants from the Swedish Children’s Cancer Foundation, the Swedish Cancer Society, the Swedish Research Council and the Medical Faculty at Lund University. We thank the Swegene DNA Microarray Resource Center in Lund, supported by the Knut and Alice Wallenberg Foundation through the Swegene consortium, for providing BAC arrays. The expression data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus (GEO, http:// www.ncbi.nlm.nih.gov/geo/) and are accessible through GEO Series accession number GSE7186.
References 1 Gustafsson G, Schmiegelow K, Forestier E, Clausen N, Glomstein A, Jonmundsson G et al. Improving outcome through two decades in childhood ALL in the Nordic countries: the impact of high-dose methotrexate in the reduction of CNS irradiation. Nordic Society of Pediatric Haematology and Oncology (NOPHO). Leukemia 2000; 14: 2267–2275. Leukemia
Analyses of ETV6/RUNX1-positive ALL reveal frequent Xq duplications H Lilljebjo¨rn et al
2144 2 Golub TR, Barker GF, Bohlander SK, Hiebert SW, Ward DC, BrayWard P et al. Fusion of the TEL gene on 12p13 to the AML1 gene on 21q22 in acute lymphoblastic leukemia. Proc Natl Acad Sci USA 1995; 92: 4917–4921. 3 Romana SP, Mauchauffe´ M, Le Coniat M, Chumakov I, Le Paslier D, Berger R et al. The t(12;21) of acute lymphoblastic leukemia results in a tel-AML1 gene fusion. Blood 1995; 85: 3662–3670. 4 Shurtleff SA, Buijs A, Behm FG, Rubnitz JE, Raimondi SC, Hancock ML et al. TEL/AML1 fusion resulting from a cryptic t(12;21) is the most common genetic lesion in pediatric ALL and defines a subgroup of patients with an excellent prognosis. Leukemia 1995; 9: 1985–1989. 5 Raynaud S, Cave H, Baens M, Bastard C, Cacheux V, Grosgeorge J et al. The 12;21 translocation involving TEL and deletion of the other TEL allele: two frequently associated alterations found in childhood acute lymphoblastic leukemia. Blood 1996; 87: 2891–2899. 6 Loh ML, Goldwasser MA, Silverman LB, Poon WM, Vattikuti S, Cardoso A et al. Prospective analysis of TEL/AML1-positive patients treated on Dana-Farber Cancer Institute Consortium Protocol 9501. Blood 2006; 107: 4508–4513. 7 Harbott J, Viehmann S, Borkhardt A, Henze G, Lampert F. Incidence of TEL/AML1 fusion gene analyzed consecutively in children with acute lymphoblastic leukemia in relapse. Blood 1997; 90: 4933–4937. 8 Nakao M, Yokota S, Horiike S, Taniwaki M, Kashima K, Sonoda Y et al. Detection and quantification of TEL/AML1 fusion transcripts by polymerase chain reaction in childhood acute lymphoblastic leukemia. Leukemia 1996; 10: 1463–1470. 9 Wiemels JL, Cazzaniga G, Daniotti M, Eden OB, Addison GM, Masera G et al. Prenatal origin of acute lymphoblastic leukaemia in children. Lancet 1999; 354: 1499–1503. 10 Ford AM, Bennett CA, Price CM, Bruin MCA, Van Wering ER, Greaves M. Fetal origins of the TEL-AML1 fusion gene in identical twins with leukemia. Proc Natl Acad USA 1998; 95: 4584–4588. 11 Mori H, Colman SM, Xiao Z, Ford AM, Healy LE, Donaldson C et al. Chromosome translocations and covert leukemic clones are generated during normal fetal development. Proc Natl Acad USA 2002; 99: 8242–8247. 12 Fischer M, Schwieger M, Horn S, Niebuhr B, Ford A, Roscher S et al. Defining the oncogenic function of the TEL/AML1 (ETV6/ RUNX1) fusion protein in a mouse model. Oncogene 2005; 24: 7579–7591. 13 Stams WAG, Beverloo HB, den Boer ML, de Menezes RX, Stigter RL, van Drunen E et al. Incidence of additional genetic changes in the TEL and AML1 genes in DCOG and COALL-treated t(12;21)positive pediatric ALL, and their relation with drug sensitivity and clinical outcome. Leukemia 2006; 20: 410–416. 14 Forestier E, Andersen MK, Autio K, Blennow E, Borgstro¨m G, Golovleva I et al. Cytogenetic patterns in ETV6/RUNX1-positive pediatric B-cell precursor acute lymphoblastic leukemia: a Nordic series of 245 cases and review of the literature. Genes Chromosomes Cancer 2007; 46: 440–450. 15 Lu XY, Harris CP, Cooley L, Margolin J, Steuber PC, Sheldon M et al. The utility of spectral karyotyping in the cytogenetic analysis of newly diagnosed pediatric acute lymphoblastic leukemia. Leukemia 2002; 16: 2222–2227. 16 Martineau M, Jalali GR, Barber KE, Broadfield ZJ, Cheung KL, Lilleyman J et al. ETV6/RUNX1 fusion at diagnosis and relapse: some prognostic indications. Genes Chromosomes Cancer 2005; 43: 54–71. 17 Nordgren A, Heyman M, Sahle´n S, Schoumans J, So¨derha¨ll S, Nordenskjo¨ld M et al. Spectral karyotyping and interphase FISH reveal abnormalities not detected by conventional G-banding. Implications for treatment stratification of childhood acute lymphoblastic leukaemia: detailed analysis of 70 cases. Eur J Haematol 2002; 68: 31–41. 18 Andersson A, Olofsson T, Lindgren D, Nilsson B, Ritz C, Ede´n P et al. Molecular signatures in childhood acute leukemia and
19
20
21
22 23
24
25
26 27
28
29
30
31 32 33 34
35
their correlations to expression patterns in normal hematopoietic subpopulations. Proc Natl Acad Sci USA 2005; 102: 19069–19074. Andreasson P, Ho¨glund M, Be´ka´ssy AN, Garwicz S, Heldrup J, Mitelman F et al. Cytogenetic and FISH studies of a single center consecutive series of 152 childhood acute lymphoblastic leukemias. Eur J Haematol 2000; 65: 40–51. Paulsson K, Heidenblad M, Mo¨rse H, Borg A˚, Fioretos T, Johansson B. Identification of cryptic aberrations and characterization of translocation breakpoints using array CGH in high hyperdiploid childhood acute lymphoblastic leukemia. Leukemia 2006; 20: 2002–2007. ˚ , Ho¨glund M Heidenblad M, Hallor KH, Staaf J, Jo¨nsson G, Borg A et al. Genomic profiling of bone and soft tissue tumors with supernumerary ring chromosomes using tiling resolution bacterial artificial chromosome microarrays. Oncogene 2006; 25: 7106–7116. Ross ME, Zhou X, Song G, Shurtleff SA, Girtman K, Williams WK et al. Classification of pediatric acute lymphoblastic leukemia by gene expression profiling. Blood 2003; 102: 2951–2959. Andersson A, Ritz C, Lindgren D, Ede´n P, Lassen C, Heldrup J et al. Microarray-based classification of a consecutive series of 121 childhood acute leukemias: prediction of leukemic and genetic subtype as well as of minimal residual disease status. Leukemia 2007; 21: 1198–1203. Schaffner C, Stilgenbauer S, Rappold GA, Do¨hner H, Lichter P. Somatic ATM mutations indicate a pathogenic role of ATM in B-cell chronic lymphocytic leukemia. Blood 1999; 94: 748–753. Wang Z, Zhang J, Zhang Y, Lim SH. SPAN-Xb expression in myeloma cells is dependent on promoter hypomethylation and can be upregulated pharmacologically. Int J Cancer 2006; 118: 1436–1444. Mitelman F, Johansson B, Mertens F. Mitelman database of chromosome aberrations in cancer. 2007. http://cgap.nci.nih.gov/ Chromosomes/Mitelman. Ma SK, Wan TSK, Cheuk ATC, Fung LF, Chan GCF, Chan SY et al. Characterization of additional genetic events in childhood acute lymphoblastic leukemia with TEL/AML1 gene fusion: a molecular cytogenetics study. Leukemia 2001; 15: 1442–1447. Jarosˇova´ M, Holzerova´ M, Miha´l V, Lakoma´ I, Divoky´ V, Blazˇek B et al. Complex karyotypes in childhood acute lymphoblastic leukemia: cytogenetic and molecular cytogenetic study of 21 cases. Cancer Genet Cytogenet 2003; 145: 161–168. Kuchinskaya E, Heyman M, Grande´r D, Linderholm M, So¨derha¨ll S, Zaritskey A et al. Children and adults with acute lymphoblastic leukaemia have similar gene expression profiles. Eur J Haematol 2005; 74: 466–480. Poppe B, Cauwelier B, Van Limbergen H, Yigit N, Philippe´ J, Verhasselt B et al. Novel cryptic chromosomal rearrangements in childhood acute lymphoblastic leukemia detected by multiple color fluorescent in situ hybridization. Haematologica 2005; 90: 1179–1185. Mullighan CG, Goorha S, Radtke I, Miller CB, Coustan-Smith E, Dalton JD et al. Genome-wide analysis of genetic alterations in acute lymphoblastic leukaemia. Nature 2007; 446: 758–764. Gartler SM, Goldman MA. Biology of the X chromosome. Curr Opin Pediatr 2001; 13: 340–345. Richardson AL, Wang ZC, De Nicolo A, Lu X, Brown M, Miron A et al. X chromosomal abnormalities in basal-like human breast cancer. Cancer Cell 2006; 9: 121–132. Wang Z, Zhang Y, Liu H, Salati E, Chiriva-Internati M, Lim SH. Gene expression and immunologic consequence of SPAN-Xb in myeloma and other hematologic malignancies. Blood 2003; 101: 955–960. Kouprina N, Pavlicek A, Noskov VN, Solomon G, Otstot J, Isaacs W et al. Dynamic structure of the SPANX gene cluster mapped to the prostate cancer susceptibility locus HPCX at Xq27. Genome Res 2005; 15: 1477–1486.
Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)
Leukemia
