Masashi Sanada

A Robust Algorithm for Copy Number Detection Using High-Density Oligonucleotide Single Nucleotide Polymorphism Genotyping Arrays

We have developed a robust algorithm for copy number analysis of the human genome using high-density oligonucleotide microarrays containing 116,204 single-nucleotide polymorphisms. The advantages of this algorithm include the improvement of signal-to-noise (S/N) ratios and the use of an optimized reference. The raw S/N ratios were improved by accounting for the length and GC content of the PCR products using quadratic regressions. The use of constitutional DNA, when available, gives the lowest SD values (0.16 +/- 0.03) and also enables allele-based copy number detection in cancer genomes, which can unmask otherwise concealed allelic imbalances. In the absence of constitutional DNA, optimized selection of multiple normal references with the highest S/N ratios, in combination with the data regressions, dramatically improves SD values from 0.67 +/- 0.12 to 0.18 +/- 0.03. These improvements allow for highly reliable comparison of data across different experimental conditions, detection of allele-based copy number changes, and more accurate estimations of the range and magnitude of copy number aberrations. This algorithm has been implemented in a software package called Copy Number Analyzer for Affymetrix GeneChip Mapping 100K arrays (CNAG). Overall, these enhancements make CNAG a useful tool for high-resolution detection of copy number alterations which can help in the understanding of the pathogenesis of cancers and other diseases as well as in exploring the complexities of the human genome.

Oncogenic mutations of ALK kinase in neuroblastoma

Neuroblastoma in advanced stages is one of the most intractable paediatric cancers, even with recent therapeutic advances. Neuroblastoma harbours a variety of genetic changes, including a high frequency of MYCN amplification, loss of heterozygosity at 1p36 and 11q, and gain of genetic material from 17q, all of which have been implicated in the pathogenesis of neuroblastoma. However, the scarcity of reliable molecular targets has hampered the development of effective therapeutic agents targeting neuroblastoma. Here we show that the anaplastic lymphoma kinase (ALK), originally identified as a fusion kinase in a subtype of non-Hodgkin’s lymphoma (NPM–ALK) and more recently in adenocarcinoma of lung (EML4–ALK), is also a frequent target of genetic alteration in advanced neuroblastoma. According to our genome-wide scans of genetic lesions in 215 primary neuroblastoma samples using high-density single-nucleotide polymorphism genotyping microarrays, the ALK locus, centromeric to the MYCN locus, was identified as a recurrent target of copy number gain and gene amplification. Furthermore, DNA sequencing of ALK revealed eight novel missense mutations in 13 out of 215 (6.1%) fresh tumours and 8 out of 24 (33%) neuroblastoma-derived cell lines. All but one mutation in the primary samples (12 out of 13) were found in stages 3–4 of the disease and were harboured in the kinase domain. The mutated kinases were autophosphorylated and displayed increased kinase activity compared with the wild-type kinase. They were able to transform NIH3T3 fibroblasts as shown by their colony formation ability in soft agar and their capacity to form tumours in nude mice. Furthermore, we demonstrate that downregulation of ALK through RNA interference suppresses proliferation of neuroblastoma cells harbouring mutated ALK. We anticipate that our findings will provide new insights into the pathogenesis of advanced neuroblastoma and that ALK-specific kinase inhibitors might improve its clinical outcome.

Landscape of genetic lesions in 944 patients with myelodysplastic syndromes

High-throughput DNA sequencing significantly contributed to diagnosis and prognostication in patients with myelodysplastic syndromes (MDS). We determined the biological and prognostic significance of genetic aberrations in MDS. In total, 944 patients with various MDS subtypes were screened for known/putative mutations/deletions in 104 genes using targeted deep sequencing and array-based genomic hybridization. In total, 845/944 patients (89.5%) harbored at least one mutation (median, 3 per patient; range, 0–12). Forty-seven genes were significantly mutated with TET2, SF3B1, ASXL1, SRSF2, DNMT3A, and RUNX1 mutated in >10% of cases. Many mutations were associated with higher risk groups and/or blast elevation. Survival was investigated in 875 patients. By univariate analysis, 25/48 genes (resulting from 47 genes tested significantly plus PRPF8) affected survival (P<0.05). The status of 14 genes combined with conventional factors revealed a novel prognostic model (‘Model-1’) separating patients into four risk groups (‘low’, ‘intermediate’, ‘high’, ‘very high risk’) with 3-year survival of 95.2, 69.3, 32.8, and 5.3% (P<0.001). Subsequently, a ‘gene-only model’ (‘Model-2’) was constructed based on 14 genes also yielding four significant risk groups (P<0.001). Both models were reproducible in the validation cohort (n=175 patients; P<0.001 each). Thus, large-scale genetic and molecular profiling of multiple target genes is invaluable for subclassification and prognostication in MDS patients.

Frequent inactivation of A20 in B-cell lymphomas

A20 is a negative regulator of the NF-κB pathway and was initially identified as being rapidly induced after tumour-necrosis factor-α stimulation. It has a pivotal role in regulation of the immune response and prevents excessive activation of NF-κB in response to a variety of external stimuli; recent genetic studies have disclosed putative associations of polymorphic A20 (also called TNFAIP3) alleles with autoimmune disease risk. However, the involvement of A20 in the development of human cancers is unknown. Here we show, using a genome-wide analysis of genetic lesions in 238 B-cell lymphomas, that A20 is a common genetic target in B-lineage lymphomas. A20 is frequently inactivated by somatic mutations and/or deletions in mucosa-associated tissue lymphoma (18 out of 87; 21.8%) and Hodgkin’s lymphoma of nodular sclerosis histology (5 out of 15; 33.3%), and, to a lesser extent, in other B-lineage lymphomas. When re-expressed in a lymphoma-derived cell line with no functional A20 alleles, wild-type A20, but not mutant A20, resulted in suppression of cell growth and induction of apoptosis, accompanied by downregulation of NF-κB activation. The A20-deficient cells stably generated tumours in immunodeficient mice, whereas the tumorigenicity was effectively suppressed by re-expression of A20. In A20-deficient cells, suppression of both cell growth and NF-κB activity due to re-expression of A20 depended, at least partly, on cell-surface-receptor signalling, including the tumour-necrosis factor receptor. Considering the physiological function of A20 in the negative modulation of NF-κB activation induced by multiple upstream stimuli, our findings indicate that uncontrolled signalling of NF-κB caused by loss of A20 function is involved in the pathogenesis of subsets of B-lineage lymphomas.

Gain-of-function of mutated C-CBL tumour suppressor in myeloid neoplasms

Acquired uniparental disomy (aUPD) is a common feature of cancer genomes, leading to loss of heterozygosity. aUPD is associated not only with loss-of-function mutations of tumour suppressor genes, but also with gain-of-function mutations of proto-oncogenes. Here we show unique gain-of-function mutations of the C-CBL (also known as CBL) tumour suppressor that are tightly associated with aUPD of the 11q arm in myeloid neoplasms showing myeloproliferative features. The C-CBL proto-oncogene, a cellular homologue of v-Cbl, encodes an E3 ubiquitin ligase and negatively regulates signal transduction of tyrosine kinases. Homozygous C-CBL mutations were found in most 11q-aUPD-positive myeloid malignancies. Although the C-CBL mutations were oncogenic in NIH3T3 cells, c-Cbl was shown to functionally and genetically act as a tumour suppressor. C-CBL mutants did not have E3 ubiquitin ligase activity, but inhibited that of wild-type C-CBL and CBL-B (also known as CBLB), leading to prolonged activation of tyrosine kinases after cytokine stimulation. c-Cbl-/- haematopoietic stem/progenitor cells (HSPCs) showed enhanced sensitivity to a variety of cytokines compared to c-Cbl+/+ HSPCs, and transduction of C-CBL mutants into c-Cbl-/- HSPCs further augmented their sensitivities to a broader spectrum of cytokines, including stem-cell factor (SCF, also known as KITLG), thrombopoietin (TPO, also known as THPO), IL3 and FLT3 ligand (FLT3LG), indicating the presence of a gain-of-function that could not be attributed to a simple loss-of-function. The gain-of-function effects of C-CBL mutants on cytokine sensitivity of HSPCs largely disappeared in a c-Cbl+/+ background or by co-transduction of wild-type C-CBL, which suggests the pathogenic importance of loss of wild-type C-CBL alleles found in most cases of C-CBL-mutated myeloid neoplasms. Our findings provide a new insight into a role of gain-of-function mutations of a tumour suppressor associated with aUPD in the pathogenesis of some myeloid cancer subsets.

Gain‐of‐function mutations and copy number increases of Notch2 in diffuse large B‐cell lymphoma

Signaling through the Notch1 receptor has a pivotal role in early thymocyte development. Gain of Notch1 function results in the development of T‐cell acute lymphoblastic leukemia in a number of mouse experimental models, and activating Notch1 mutations deregulate Notch1 signaling in the majority of human T‐cell acute lymphoblastic leukemias. Notch2, another member of the Notch gene family, is preferentially expressed in mature B cells and is essential for marginal zone B‐cell generation. Here, we report that 5 of 63 (~8%) diffuse large B‐cell lymphomas, a subtype of mature B‐cell lymphomas, have Notch2 mutations. These mutations lead to partial or complete deletion of the proline‐, glutamic acid‐, serine‐ and threonine‐rich (PEST) domain, or a single amino acid substitution at the C‐terminus of Notch2 protein. Furthermore, high‐density oligonucleotide microarray analysis revealed that some diffuse large B‐cell lymphoma cases also have increased copies of the mutated Notch2 allele. In the Notch activation‐sensitive luciferase reporter assay in vitro, mutant Notch2 receptors show increased activity compared with wild‐type Notch2. These findings implicate Notch2 gain‐of‐function mutations in the pathogenesis of a subset of B‐cell lymphomas, and suggest broader roles for Notch gene mutations in human cancers. (Cancer Sci 2009; 100: 920–926)

BCOR and BCORL1 mutations in myelodysplastic syndromes and related disorders

Patients with low-risk myelodysplastic syndromes (MDS) that rapidly progress to acute myeloid leukemia (AML) remain a challenge in disease management. Using whole-exome sequencing of an MDS patient, we identified a somatic mutation in the BCOR gene also mutated in AML. Sequencing of BCOR and related BCORL1 genes in a cohort of 354 MDS patients identified 4.2% and 0.8% of mutations respectively. BCOR mutations were associated with RUNX1 (P = .002) and DNMT3A mutations (P = .015). BCOR is also mutated in chronic myelomonocytic leukemia patients (7.4%) and BCORL1 in AML patients with myelodysplasia-related changes (9.1%). Using deep sequencing, we show that BCOR mutations arise after mutations affecting genes involved in splicing machinery or epigenetic regulation. In univariate analysis, BCOR mutations were associated with poor prognosis in MDS (overall survival [OS]: P = .013; cumulative incidence of AML transformation: P = .005). Multivariate analysis including age, International Prognostic Scoring System, transfusion dependency, and mutational status confirmed a significant inferior OS to patients with a BCOR mutation (hazard ratio, 3.3; 95% confidence interval, 1.4-8.1; P = .008). These data suggest that BCOR mutations define the clinical course rather than disease initiation. Despite infrequent mutations, BCOR analyses should be considered in risk stratification.

Prevalence and prognostic impact of allelic imbalances associated with leukemic transformation of Philadelphia chromosome–negative myeloproliferative neoplasms

Philadelphia chromosome-negative myeloproliferative neoplasms (MPNs) including polycythemia vera, essential thrombocythemia, and primary myelofibrosis show an inherent tendency for transformation into leukemia (MPN-blast phase), which is hypothesized to be accompanied by acquisition of additional genomic lesions. We, therefore, examined chromosomal abnormalities by high-resolution single nucleotide polymorphism (SNP) array in 88 MPN patients, as well as 71 cases with MPN-blast phase, and correlated these findings with their clinical parameters. Frequent genomic alterations were found in MPN after leukemic transformation with up to 3-fold more genomic changes per sample compared with samples in chronic phase (P < .001). We identified commonly altered regions involved in disease progression including not only established targets (ETV6, TP53, and RUNX1) but also new candidate genes on 7q, 16q, 19p, and 21q. Moreover, trisomy 8 or amplification of 8q24 (MYC) was almost exclusively detected in JAK2V617F(-) cases with MPN-blast phase. Remarkably, copy number-neutral loss of heterozygosity (CNN-LOH) on either 7q or 9p including homozygous JAK2V617F was related to decreased survival after leukemic transformation (P = .01 and P = .016, respectively). Our high-density SNP-array analysis of MPN genomes in the chronic compared with leukemic stage identified novel target genes and provided prognostic insights associated with the evolution to leukemia.

High-Resolution Genomic Copy Number Profiling of Glioblastoma Multiforme by Single Nucleotide Polymorphism DNA Microarray

Glioblastoma multiforme (GBM) is an extremely malignant brain tumor. To identify new genomic alterations in GBM, genomic DNA of tumor tissue/explants from 55 individuals and 6 GBM cell lines were examined using single nucleotide polymorphism DNA microarray (SNP-Chip). Further gene expression analysis relied on an additional 56 GBM samples. SNP-Chip results were validated using several techniques, including quantitative PCR (Q-PCR), nucleotide sequencing, and a combination of Q-PCR and detection of microsatellite markers for loss of heterozygosity with normal copy number [acquired uniparental disomy (AUPD)]. Whole genomic DNA copy number in each GBM sample was profiled by SNP-Chip. Several signaling pathways were frequently abnormal. Either the p16(INK4A)/p15(INK4B)-CDK4/6-pRb or p14(ARF)-MDM2/4-p53 pathways were abnormal in 89% (49 of 55) of cases. Simultaneous abnormalities of both pathways occurred in 84% (46 of 55) samples. The phosphoinositide 3-kinase pathway was altered in 71% (39 of 55) GBMs either by deletion of PTEN or amplification of epidermal growth factor receptor and/or vascular endothelial growth factor receptor/platelet-derived growth factor receptor α. Deletion of chromosome 6q26-27 often occurred (16 of 55 samples). The minimum common deleted region included PARK2, PACRG, QKI, and PDE10A genes. Further reverse transcription Q-PCR studies showed that PARK2 expression was decreased in another collection of GBMs at a frequency of 61% (34 of 56) of samples. The 1p36.23 region was deleted in 35% (19 of 55) of samples. Notably, three samples had homozygous deletion encompassing this site. Also, a novel internal deletion of a putative tumor suppressor gene, LRP1B, was discovered causing an aberrant protein. AUPDs occurred in 58% (32 of 55) of the GBM samples and five of six GBM cell lines. A common AUPD was found at chromosome 17p13.3-12 (included p53 gene) in 13 of 61 samples and cell lines. Single-strand conformational polymorphism and nucleotide sequencing showed that 9 of 13 of these samples had homozygous p53 mutations, suggesting that mitotic recombination duplicated the abnormal p53 gene, probably providing a growth advantage to these cells. A significantly shortened survival time was found in patients with 13q14 (RB) deletion or 17p13.1 (p53) deletion/AUPD. Taken together, these results suggest that this technique is a rapid, robust, and inexpensive method to profile genome-wide abnormalities in GBM.(Mol Cancer Res 2009;7(5):665–77)

Cloning of genes involved in chromosomal translocations by high-resolution single nucleotide polymorphism genomic microarray

High-resolution single nucleotide polymorphism genomic microarray (SNP-chip) is a useful tool to define gene dosage levels over the whole genome, allowing precise detection of deletions and duplications/amplifications of chromosomes in cancer cells. We found that this new technology can also identify breakpoints of chromosomes involved in unbalanced translocations, leading to identification of fusion genes. Using this technique, we found that the PAX5 gene was rearranged to a variety of partner genes including ETV6, FOXP1, AUTS2, and C20orf112 in pediatric acute lymphoblastic leukemia (ALL). The 3′ end of the PAX5 gene was replaced by the partner gene. The PAX5 fusion products bound to PAX5 recognition sequences as strongly as wild-type PAX5 and suppressed its transcriptional activity in a dominant-negative fashion. In human B cell leukemia cells, binding of wild-type PAX5 to a regulatory region of BLK, one of the direct downstream target genes of PAX5, was diminished by expression of the PAX5-fusion protein, leading to repression of BLK. Expression of PAX5-fusion genes in murine bone marrow cells blocked development of mature B cells. PAX5-fusion proteins may contribute to leukemogenesis by blocking differentiation of hematopoietic cells into mature B cells. SNP-chip is a powerful tool to identify fusion genes in human cancers.