Identification of Two Critically Deleted Regions within Chromosome Segment 7q35-q36 in EVI1 Deregulated Myeloid Leukemia Cell Lines

Chromosomal rearrangements involving the EVI1 proto-oncogene are a recurrent finding in myeloid leukemias and are indicative of a poor prognosis. Rearrangements of the EVI1 locus are often associated with monosomy 7 or cytogenetic detectable deletions of part of 7q. As EVI1 overexpression alone is not sufficient to induce leukemia, loss of a 7q tumour suppressor gene might be a required cooperating event. To test this hypothesis, we performed high-resolution array comparative genomic hybridization analysis of twelve EVI1 overexpressing patients and three EVI1 deregulated cell lines to search for 7q submicroscopic deletions. This analysis lead to the delineation of two critical regions, one of 0.39 Mb on 7q35 containing the CNTNAP2 gene and one of 1.33 Mb on chromosome bands 7q35–q36 comprising nine genes in EVI1 deregulated cell lines. These findings open the way to further studies aimed at identifying the culprit EVI1 implicated tumour suppressor genes on 7q.


Introduction
Chromosomal rearrangements involving chromosome band 3q26, such as translocations with various partner chromosomes or inversions of chromosome 3 are a recurrent finding in myeloid leukemias [1]. These aberrations contribute to the ectopic expression of the EVI1 proto-oncogene. EVI1 transcriptional activation has been reported in up to 10% of myeloid leukemia patients, even in the absence of 3q26 rearrangements, and is an independent indicator of adverse prognosis [2].
Retroviral integration experiments have shown that EVI1 overexpression alone is not sufficient to cause leukemia, indicating that cooperative effects are necessary for malignant transformation [3,4]. Interestingly, over fifty percent of the 3q26 rearranged leukemias also display chromosomal abnormalities involving chromosome 7, such as monosomy 7 or deletion of part of 7q [5]. This association alludes to the existence of a 7q tumour suppressor gene, which when deleted acts in concert with EVI1 overexpression to induce malignant transformation. We performed high-resolution array comparative genomic hybridization (CGH) analysis to search for 7q submicroscopic deletions in EVI1 deregulated leukemia patients in order to identify candidate 7q tumour suppressor genes.

Patients and Cell Lines
Diagnostic bone marrow samples of twelve myeloid leukemia samples and remission bone marrow samples of two patients (case 7 and case 8) were included in the study. Fluorescence in situ hybridization (FISH) for detection of EVI1 rearrangement and reverse transcription quantitative PCR (RT-qPCR) for detection of EVI1 ectopic expression was performed as previously described [6]. The study was approved by the ethics committee of the Ghent University Hospital (2003/273).
Patient and cell line characteristics and EVI1 FISH and RT-qPCR results are described in Table 1.

Array CGH
DNA of patient samples and cell lines was isolated using the QIAamp DNA mini kit (Qiagen, Belgium) or the Puregene Cell kit (Gentra Systems, Belgium) according to the manufacturer's descriptions. Array CGH analysis for submicroscopic 7q deletions was performed on a 44K custom array (Agilent technologies, Belgium) covering the entire chromosome 7 with a probe spacing of 1 oligonucleotide every 3 kb, according to the manufacturer's descriptions. In brief, DNA (400 ng) was labelled using the BioPrime Array CGH genomic labelling system (Invitrogen, Belgium) using Cy5 (control DNA; Promega, Belgium) and Cy3 (patient sample or cell line) labelled dCTPs (GE Healthcare, Belgium). Following labelling, hybridization, and washing of the slides, arrays were scanned using an Agilent DNA Microarray Scanner, quantified with Feature Extraction software 10.1 and data were further analyzed with arrayCGHbase [10] using a circular binary segmentation (CBS) algorithm taken into consideration log 2 ratios of neighbouring probes [11]. Deletions were called heterozygous when CBS ratios were #20.5 and a deletion was considered putatively homozygous when the CBS ratios were #21.2. Common copy number variations (CNVs) present in the Database of Genomic Variants (www.projects.tcag.ca/variation) were excluded from analysis.

Fluorescence In Situ Hybridization
To confirm the overlapping deletions found in the cell lines Kasumi-3, UCSD-AML1 and MUTZ-3 at 7q35-q36, FISH analysis was performed as previously described [6]. FISH probes were selected from the UCSC genome browser database (http:// genome.ucsc.edu) ( Table 2).

Results and Discussion
Based on karyotyping and FISH analysis, two critical regions, one on chromosome band 7q22 [14] and one on chromosome bands 7q32-q35 [15], have been identified as the targets for 7q deletions in myeloid leukemia. In this study, array CGH was applied to search for deletions on chromosome 7 at ultra-high resolution. Using this approach, we identified several submicroscopic 7p and 7q deletions in EVI1 overexpressing patients and cell lines (Table 3 and Figure 1) including two critically deleted regions on 7q35-q36 in EVI1 deregulated cell lines. These deletions were confirmed using FISH ( Figure 2).
Interestingly, in the Kasumi-3 cell line a 680 kb homozygous deletion on 7q encompassing the CNTNAP2 gene was identified, located within a 6.51 Mb heterozygous deleted region at 7q34-q36. The observation of homozygous deletions has been of great importance in tumour genetics as the deleted region often contains putative tumour suppressor genes. In the MUTZ-3 cell line, a 4.83 Mb deletion was present within the same chromosomal region, delineating a shortest region of overlap (SRO) of 1.33 Mb (Figure 3) encompassing nine genes (Table 4). Interestingly, in the UCSD-AML1 cell line a homozygous deletion of 0.39 Mb was detected encompassing the CNTNAP2 gene, which delineated a second SRO of 0.39 Mb on 7q35 (Figure 3).
CNTNAP2 is located in a common fragile site which is inactivated in different types of cancers such as brain, ovarian and breast tumours [16], and methylation of the promoter of this gene has been described in pancreatic adenocarcinoma [17].
However, upon RT-qPCR analysis of the CNTNAP2 gene no expression could be detected in normal bone marrow samples, CD34 + cell fractions or EVI1 overexpressing bone marrow samples (data not shown), indicating that this gene is not a straightforward candidate 7q tumour suppressor gene cooperating with EVI1 overexpression.
Subsequent screening of twelve EVI1 overexpressing patient samples with (3 cases) or without (9 cases) monosomy 7, revealed no submicroscopic alterations within the above described SROs. However, focal deletions in other genomic regions on chromosome arms 7p and 7q were detected.
Of interest is that both the Kasumi-3 and the MUTZ-3 cell line display chromosomal rearrangements involving chromosome 7. It is well known that small deletions can occur at chromosomal breakpoints [18]. Therefore, the observed 4.83 Mb deletion at 7q36 in MUTZ-3 might have originated as a consequence of the t(2;7)(q36;q36) and/or the inv(7)(p15q36) present in this cell line.
Of the nine genes located within the 1.33 Mb SRO CUL1 and EZH2 are the most promising candidates due to known function in and association with cancer. The CUL1 gene encodes a protein that, when associated with other proteins such as Skp1, forms the Skp2 E3 ubiquitin ligase E3 complex involved in degradation of different proteins [19]. In AML, the Skp2 complex has been described to target the MEF transcription factor which induces G1/S transition. Impairment of the Skp2 complex leads to a decrease in MEF degradation thereby promoting cell proliferation [20]. Deletion and mutation of other E3 ubiquitin ligase complex members such as the F-box containing FBXW7 gene has already  where it is associated with poor prognosis [21]. Moreover, in AML loss-of-heterozygosity (LOH) of the EZH2 (polycomb group) gene (enhancer of zest) was detected in 5 out of 21 patients [22]. However, further analyses such as mutation screening and functional RNA interference screens [23] are needed to identify the genes contributing to EVI1 leukemogenesis. In addition to the regions described above, other 7q deletions in the EVI1 overexpressing patients as well as in the UCSD-AML1 cell line were detected. In UCSD-AML1 two additional small 7q deletions were observed, a 53 kb (7q11.22) and 7 kb (7q21.11) deletion and containing the AUTS2 and MAGI2 genes respectively. In case 2, a small deletion of 10 kb on 7q34 was observed encompassing the KEL blood group gene. In Kasumi-3, this gene was located in a larger deleted region of 6.51 Mb on 7q34-q36. In case 9, a small homozygous deletion of 20 kb on 7q32.2 was detected. Located in this region is the NRF1 gene for which inactivating mutations have already been described in liver cancer [24].
Besides the above mentioned 7q deletions, several 7p deletions were also detected, some of which were common between different patients and/or cell lines. In case 3, a small deletion of 10 kb on 7p22 was observed encompassing the SDK1 sphingosine dependent protein kinase gene. In Kasumi-3, this gene was located in a larger deleted region of 4.37 Mb on 7p22. As involvement of the SDK1 gene in apoptosis has already been described [25], deletion of this gene could lead to a decreased rate of cell death. For patient 3 deletion of the STK17A gene at 7p12 was detected. Deletion of this pro-apoptotic gene had already been described in laryngeal squamous cell carcinoma [26]. It should be noted that it is possible that the above described small 7q and 7p deletions in patient samples are not leukemogenic as no constitutional material of these patients was available. In the remaining patient samples no chromosome 7 deletions could be detected.
In conclusion, using high-resolution array CGH, we were able to delineate two critical regions for 7q deletions in myeloid leukemias that were not detectable using standard karyotyping. Besides a 0.39 Mb SRO on 7q35 containing the CNTNAP2 gene, we identified a 1.33 Mb region on chromosome bands 7q35-q36 containing nine putative tumour suppressor genes in EVI1 deregulated cell lines. Further analysis of these candidates is warranted to investigate their role in EVI1 mediated malignant