Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Array-based genomic resequencing of human leukemia


To identify oncogenes in leukemias, we performed large-scale resequencing of the leukemia genome using DNA sequence arrays that determine 9 Mbp of sequence corresponding to the exons or exon–intron boundaries of 5648 protein-coding genes. Hybridization of genomic DNA from CD34-positive blasts of acute myeloid leukemia (n=19) or myeloproliferative disorder (n=1) with the arrays identified 9148 nonsynonymous nucleotide changes. Subsequent analysis showed that most of these changes were also present in the genomic DNA of the paired controls, with 11 somatic changes identified only in the leukemic blasts. One of these latter changes results in a Met-to-Ile substitution at amino-acid position 511 of Janus kinase 3 (JAK3), and the JAK3(M511I) protein exhibited transforming potential both in vitro and in vivo. Further screening for JAK3 mutations showed novel and known transforming changes in a total of 9 out of 286 cases of leukemia. Our experiments also showed a somatic change responsible for an Arg-to-His substitution at amino-acid position 882 of DNA methyltransferase 3A, which resulted in a loss of DNA methylation activity of >50%. Our data have thus shown a unique profile of gene mutations in human leukemia.


Leukemias are clonal disorders of hematopoietic stem cells or immature progenitors. Several subtypes of leukemia are associated with disease-specific karyotype anomalies in the malignant blasts. Most cases of acute promyelocytic leukemia a subtype of acute myeloid leukemia (AML), for instance, are associated with a t(15;17) chromosomal rearrangement that results in the production of the PML-RARA fusion-type oncoprotein (Tallman and Altman, 2008). Similarly, another subtype of AML is associated with a t(8;21) rearrangement, resulting in the production of the oncogenic RUNX1-CBFA2T1 protein (Nimer and Moore, 2004).

The karyotype of leukemic blasts is an important determinant of the long-term prognosis of affected individuals. AML with t(15;17), t(8;21) or inv(16) rearrangements thus constitutes a subgroup of leukemias with a ‘favorable’ karyotype, with a 5-year survival rate of >60%, whereas AML with an ‘adverse’ karyotype (monosomy 7, monosomy 5 or complex anomalies) has a 5-year survival rate of only <15% (Grimwade et al., 1998). The prognosis of AML with a normal karyotype (constituting 50% of all AML cases) is substantially worse than that with a favorable karyotype, with a 5-year survival rate of 24% (Byrd et al., 2002), indicating that blasts with a normal karyotype may contain transforming genes generated as a result of (1) sequence alterations, (2) epigenetic abnormalities or (3) small chromosomal rearrangements not detectable by the G-banding technique. Indeed, several genes, including NPM1 and KIT, have been found to be mutated and activated in AML blasts with a normal karyotype (Schlenk et al., 2008).

The identification of transforming genes in AML will require large-scale resequencing of the blast genome. Although a new generation of sequencing technologies is now available, whole-genome resequencing of many samples remains a demanding task (Bentley et al., 2008; Wheeler et al., 2008). Although DNA microarray-based sequencing is suitable for analysis of multiple samples, currently available platforms are limited in the number of nucleotides that each array is able to probe. To overcome such limitations, we have now applied the extra-large arrays (‘wafers’) manufactured by Perlegen Sciences (Mountain View, CA, USA) (originally developed for typing of single-nucleotide polymorphisms) (Patil et al., 2001) to resequencing of the human genome. Our two-step analysis of human leukemia specimens (n=20) has identified a novel transforming mutation in the gene for Janus kinase 3 (JAK3) and a hypomorphic mutation in that for DNA methyltransferase 3A (DNMT3A).


Sequencing strategy

Oligonucleotide probes on the sequencing wafer for the first phase of our study were designed to detect nonsynonymous nucleotide changes in the coding exons of the genome. Intronic sequences (GT in the splicing donor sequence AG-GT and AG in the splicing acceptor sequence AG-G) adjacent to coding exons were also interrogated with the wafer to capture splicing anomalies. Genes examined by the wafer included those known to be mutated in cancer and reported in the catalog of somatic mutations in cancer (COSMIC, as of September 2006 (n=338) and those related to the regulation of DNA repair (n=419), chromatin structure (n=299), redox regulation (n=102), epigenetic regulation (n=44), cell signaling (n=2490), protein kinases (n=314), gene transcription (n=797), cell cycle (n=297), apoptosis (n=312), DNA replication (n=144) or other functions (n=92) (Figure 1a). A total of 5648 genes were thus analyzed with the wafer.

Figure 1

Resequencing of the leukemia genome with wafers. (a) Genes interrogated by the phase I wafer (n=5648) included those listed in the COSMIC database and those categorized on the basis of function of the encoded protein as indicated. (b) CD34+ and CD4+ cell fractions were purified from individuals with leukemia (n=20). Genomic DNA of the former fractions was assayed with the phase I wafer including 9 Mbp of sequence, resulting in the isolation of 9148 nonsynonymous nucleotide changes in 3403 independent genes. The phase II wafer was then constructed to analyze these 9148 changes and was hybridized with genomic DNA from both CD34+ and CD4+ fractions separately. Only 11 mutations were found to be present in the former fraction but not in the latter.

To efficiently isolate oncogenes generated by point mutation using our sequencing array, we selected leukemic blasts with a karyotype characterized by few chromosome anomalies and by few copy number variations of chromosomes, as determined by comparative genomic hybridization with single-nucleotide polymorphism-typing arrays (Supplementary Figure S1). We isolated 15 cases of de novo AML, 4 cases of AML that developed from myelodysplastic syndrome, and 1 case of myeloproliferative disorder negative for the JAK2(V617F) and MPL(W515L) mutations (Kralovics et al., 2005; Pikman et al., 2006) (Supplementary Table S1).

From each of these 20 individuals enrolled in the study, we purified immature blasts positive for the surface expression of CD34 (leukemic fraction) as well as a paired control fraction of mature T cells positive for the surface expression of CD4. Although monocytes-macrophages may also express a low level of CD4 at the cell surface, our magnetic bead-based purification system preferentially enriched mature T cells with a high level of CD4 expression; contamination of the mature T-cell fraction with monocytes-macrophages was judged to be <9% by flow cytometry (Supplementary Figure S2).

Given the potential presence of substantial numbers of unreported single-nucleotide polymorphisms in the human genome, we adopted a two-step analysis to select somatic changes (Figure 1b). In phase I, genomic DNA was isolated from the CD34+ fraction, subjected to mid-range PCR amplification and hybridized with the wafer to examine 9 Mbp of nucleotide sequence. In phase II, we constructed a smaller wafer to investigate only the nucleotides shown to be changed in phase I relative to the human reference sequence. Genomic DNA isolated from leukemic blasts and paired control fractions was then analyzed individually with the phase II wafer. We assumed that a nucleotide change was a germline polymorphism if it was observed in both leukemic and control fractions of the same individual, and that it was a somatic mutation if it was observed in the former fraction but not in the latter.

Identification of the JAK3(M511I) mutation

Screening of the leukemic blasts of the 20 individuals for point mutations in phase I yielded 9148 nonsynonymous changes among 3403 independent genes, a frequency similar to that observed in other large-scale resequencing studies performed with capillary sequencers (Sjoblom et al., 2006; Greenman et al., 2007). However, analysis of CD4+ fractions showed that most of these sequence changes were also present in the paired control genome, leaving only 11 nonsynonymous somatic mutations in 11 genes (Supplementary Table S2). Such small number of somatic mutations is in a good agreement with the eight somatic mutations found in AML through whole-genome resequencing using the Illumina Genome Analyser (Illumina, San Diego, CA, USA) (Ley et al., 2008). All of our 11 somatic changes were confirmed by analysis of both genomic DNA and cDNA of the corresponding specimens with a capillary sequencer (data not shown). These data thus support the necessity of examining paired noncancerous specimens to pinpoint somatic changes in the cancer genome.

One of the gene mutations found only in the CD34+ fractions results in a Met-to-Ile change at amino-acid position 511 of JAK3. A heterozygous JAK3 mutation responsible for the amino-acid change was confirmed in both genomic DNA and cDNA from the CD34+ fraction, but not in those from the corresponding CD4+ fraction of patient ID JM07 (Supplementary Figure S3), who had de novo AML (M1 subtype) and a normal karyotype (Supplementary Table S1). In contrast to JAK2, activating mutations in which are preferentially associated with myeloproliferative disorder, several gain-of-function mutations (such as I87T, P132T, Q501H, A572V, R657Q and V722I) of JAK3 have recently been associated with acute megakaryoblastic leukemia of children (Walters et al., 2006; Sato et al., 2008). Other JAK3 mutations (such as A573V and A593T) were also identified in the same disorder, and an M576L substitution was detected in an adult with acute megakaryocytic leukemia (AML, M7 subtype) (Kiyoi et al., 2007), although the transforming potential of these changes remains unknown.

Given that the M511I mutant of JAK3 has not previously been described and that the relevance of JAK3 to the pathogenesis of adult AML has not been extensively investigated, we first focused on the function of JAK3(M511I). The M511 residue is located in the linker region between the Src homology 2 (SH2) domain and the pseudokinase domain of JAK3 (Figure 2a). The transforming mutation Q501H that is associated with juvenile acute megakaryoblastic leukemia (Sato et al., 2008) is also located in this region. Given that JAK3 is abundant in and has an essential role in the development of lymphocytes (Russell et al., 1995), we examined the expression level of JAK3 in AML blasts. The gene was expressed at a high level in most AML specimens (n=52), with its expression level being greater than that of JAK2 in all but three cases (Supplementary Figure S3).

Figure 2

Identification of JAK3 mutants in leukemia. (a) Amino-acid substitutions detected in this study are shown relative to the domain organization of JAK3. The mutations M511I (one case) and Q501H (four cases) are located in the linker region between the SH2 and pseudokinase domains of JAK3, whereas G62S (one case), R657Q (two cases) and R918C (one case) are located in the amino-terminal region, the pseudokinase domain and the kinase domain, respectively. The KCL22 cell line also harbors an L1017M mutation within the kinase domain of JAK3. Previously known activating mutations of JAK3 (Q501H and R657Q) are indicated by red rectangles. (b) Mouse 32D cells expressing wild-type human JAK3 or the JAK3(M511I) mutant were incubated with G-CSF (0.5 ng/ml) for 14 days, stained with Wright–Giemsa solution and examined by light microscopy. Scale bars, 20 μm. (c) C57BL/6 mice were irradiated and then injected intravenously with syngeneic CD34KSL hematopoietic stem cells infected with a retrovirus encoding JAK3(M511I) or the corresponding empty virus (control). The number of white blood cells in peripheral blood was counted at the indicated times thereafter; data are means±s.d. for 10 mice in each group (upper left panel). Peripheral blood, spleen and bone marrow isolated from recipient mice 3 months after cell injection were stained with the Wright–Giemsa solution (peripheral blood) or hematoxylin–eosin (spleen and bone marrow) and were then examined by light microscopy (upper right panel); scale bars represent 10, 200 and 50 μm, respectively. Mononuclear cells isolated from peripheral blood, thymus, spleen and bone marrow of recipient mice 3 months after cell injection were subjected to flow cytometric analysis of surface expression of CD4 and CD8a (lower panel). (d) Control BA/F3 cells (−) or those expressing the indicated JAK3 mutants were cultured without IL-3 for the indicated times, after which the cell number was determined. Data are means+s.d. of triplicates from a representative experiment.

To examine the transforming potential of JAK3(M511I), we introduced the mutant or wild-type protein into the interleukin-3 (IL-3)-dependent mouse cell line 32D (Greenberger et al., 1983). Although 32D cells forced to express wild-type JAK3 underwent rapid apoptosis after withdrawal of IL-3, those expressing JAK3(M511I) continued to grow even in the absence of IL-3, although at a reduced rate compared with that of cells expressing the artificially generated, highly transforming mutant JAK3(V674A) (Choi et al., 2007) (Supplementary Figure S3). 32D cells differentiate into terminal granulocytes in the presence of granulocyte colony-stimulating factor. However, cells expressing the M511I or V674A mutant of JAK3 maintained an exponential rate of growth, without any sign of differentiation, in the presence of granulocyte colony-stimulating factor (Figure 2b, Supplementary Figure S3), supporting the notion that the M511I mutant has transforming potential.

To directly examine the leukemogenic activity of JAK3(M511I), we generated a recombinant retrovirus encoding this mutant and used it to infect murine hematopoietic stem cells. Reconstitution of the bone marrow of lethally irradiated mice with such infected cells resulted in marked lymphocytosis in peripheral blood and enlargement of the spleen in the recipient animals (Figure 2c). The cells in the peripheral blood, spleen and bone marrow of the recipients responsible for these phenotypes manifested a medium-sized, blastic morphology, and flow cytometric analysis revealed them to be CD8+ T cells (Figure 2c). The clonal nature of these proliferating T cells was further confirmed by Southern blot analysis (Supplementary Figure S4), indicative of the development of T-cell acute lymphoblastic leukemia in the recipient mice.

To assess the prevalence of JAK3 mutations in adult leukemia, we further examined the nucleotide sequence of the entire coding region of JAK3 cDNA in an additional 266 specimens of leukemic blasts. The coding region of JAK3 cDNA was successfully amplified by PCR from 83 specimens. We could further identify 4 distinct JAK3 sequence changes in 8 of these 83 samples: 1 case with G62S, 4 cases with Q501H, 2 cases with R657Q and 1 case with R918C (Figure 2a). Taking into account the 20 cases evaluated in the phase I analysis, we thus identified a total of 9 cases with a mutant form of JAK3 (3.1%) among 286 cases of leukemia (Supplementary Table S3). Our identification of known transforming JAK3 mutants (Q501H and R657Q) originally associated with acute megakaryoblastic leukemia prompted us to determine the prevalence of these two changes in another cohort of AML (n=148), revealing two cases with JAK3(Q501H) and one case with JAK3(R657Q). In addition, analysis of a hematopoietic cell line (KCL22) (Kubonishi and Miyoshi, 1983) established from a patient with chronic myeloid leukemia in BC revealed yet another mutation (L1017M) of JAK3 (Figure 2a).

To directly compare the transforming potential of these various JAK3 mutants, we introduced each protein into the IL-3-dependent mouse B-cell line BA/F3 and examined the growth properties of the resulting transfectants. Whereas all cells expressing the JAK3 mutants proliferated in a similar manner in the presence of IL-3 (data not shown), culture without IL-3 revealed marked differences in the transforming potential among the mutants. JAK3(M511I) was the most efficient oncokinase, with a transforming activity similar to that of JAK3(V674A). The frequent mutants JAK3(Q501H) and JAK3(R657Q) exhibited weaker but still pronounced transforming potential, whereas the remaining mutants (G62S, R918C and L1017M) showed an even lower potential (Figure 2d).

Somatic mutations of DNMT3A

Another somatic mutation identified in the phase II data set was a heterozygous change in DNMT3A that results in an R882H substitution in the encoded protein (Figure 3a, Supplementary Figure S5). DNMT3A, together with DNMT3B, has an essential role in de novo methylation of the human genome (Okano et al., 1999), and an aberrant methylation profile (hypermethylation of CpG islands and hypomethylation of other regions) is a hallmark of cancer cells (El-Osta, 2004). Despite a direct linkage between such methylation changes and silencing of tumor-suppressor genes in cancer, the molecular mechanism responsible for such abnormal methylation remains unknown. Our data thus provide the first evidence of somatic mutation of a DNA methyltransferase gene in cancer cells. Mutations in the catalytic domain of DNMT3B have been shown to be responsible for a hereditary syndrome characterized by ICF (immunodeficiency, instability of the centromeric region of chromosomes and facial anomalies) in humans (Ehrlich, 2003). One of the mutation sites of DNMT3B (R823) associated with the ICF syndrome corresponds to the residue of DNMT3A (R882) shown to be mutated in this study.

Figure 3

Identification of a DNMT3A mutant in leukemia. (a) Domain organization of human DNMT3A showing that the R882 residue found to be mutated in leukemia is conserved among human (h) and mouse (m) members of the DNMT3 family. DNMT3A contains a tetrapeptide PWWP domain, polybromo homology domain (PHD) and methyltransferase (MTase) domain. The R882 residue is located in the homodimerization region present within the MTase domain. (b) Wild-type (WT) and R882H forms of DNMT3A were expressed in and purified from insect cells and then subjected to an in vitro assay of methyltransferase activity. Data are means+s.d. of triplicates from a representative experiment. The P-value was determined by Student's t-test. (c) Lysates of HEK293 cells expressing Myc epitope-tagged DNMT3L and wild-type or R882H forms of DNMT3A, as indicated, were subjected to immunoprecipitation (IP) with antibodies to Myc or to DNMT3A, and the resulting precipitates were subjected to immunoblot analysis (IB) with antibodies to DNMT3A. The position of DNMT3A (wild-type or mutant) is indicated by an open arrow.

The R882 residue of DNMT3A is considered to participate in the homodimerization and activation of the protein (Jia et al., 2007) (Figure 3a). To determine whether the R882H mutation affects the catalytic activity of DNMT3A, we expressed mutant and wild-type proteins separately in insect cells, purified them to near homogeneity and subjected them to an in vitro assay of methyltransferase activity with a synthetic substrate (Suetake et al., 2003). The catalytic activity of DNMT3A(R882H) was <50% of that of the wild-type protein (Figure 3b). DNMT3L acts as a coactivator for the methyltransferase activity of DNMT3A or DNMT3B through its association with the latter proteins (Jia et al., 2007). The R882H mutation did not affect the interaction of DNMT3A with DNMT3L in transfected mammalian cells (Figure 3c) or its sensitivity to DNMT3L as examined by the in vitro assay of methyltransferase activity (data not shown). These data thus suggested that the R882H mutation directly inhibits the enzymatic activity of DNMT3A.

Screening of another cohort of leukemia cases (n=54) for mutant forms of DNMT3A revealed another two patients with a mutation of the same amino acid (R882H in one patient and R882C in the other) (Supplementary Table S4). Therefore, we identified a total of 3 cases with an R882 mutation (4.1%) among 74 cases of leukemia. Screening for mutations of DNMT3B failed to detect any somatic changes in the same individuals (data not shown), suggesting that DNMT3A is a preferential target in leukemia.

Multistep transformation in leukemia

Although >99% of nucleotide changes in the phase I data were also observed in the paired CD4+ cells, it is unlikely that all of these changes are actually germline polymorphisms because they include established oncogenic mutations. They thus include 190 nucleotide changes previously described in cancer cells (Supplementary Table S5), such as those giving rise to NRAS(Q61H) in patient ID JM17 and to FLT3(D835Y) in patient ID JM08 (Figure 4a). Given that both NRAS(Q61H) and FLT3(D835Y) are well-characterized oncoproteins (Yamamoto et al., 2001), it is unlikely that these individuals harbored such nucleotide changes in the germ line. There are at least two possible explanations for these findings. First, it is possible that purification of the CD4+ fraction was not efficient, with the result that this fraction was contaminated by CD34+ cells. However, the CD4 expression ratio for the CD4+ and CD34+ fractions of each individual was 17.1 (median=40.1) (Figure 4b), and contamination of the CD4+ fraction with CD34+ cells at such a level would not likely produce detectable changes in Sanger sequencing outputs (compare, for instance, the signal intensities of the normal and mutant alleles in Figure 4a).

Figure 4

Proposed stepwise nature of leukemogenesis. (a) Sequencing electrophoretograms for the regions surrounding codon 61 of NRAS or codon 835 of FLT3 in genomic DNA from the CD34+ and CD4+ fractions of patient IDs JM17 and JM08, respectively. Heterozygous nucleotide changes that give rise to NRAS(Q61H) or FLT3(D835Y) were detected in both fractions of the corresponding patients. (b) The amount of CD4 mRNA in the CD4+ (control) and CD34+ (leukemia) fractions of leukemia patients (with a substantial amount of control GAPDH mRNA) was quantitated by reverse transcription and real-time PCR analysis and expressed as the control/leukemia ratio. (c) Hematopoietic stem cells (HSCs) give rise to a wide range of mature blood cells. Even after the first hit (mutation) of the genome, HSCs retain their full differentiation capacity, and therefore produce differentiated cells harboring this first hit. After the second hit, the affected cell fraction undergoes full transformation to leukemia. (d) Sequencing electrophoretograms for the genome of CD34+ and CD4+ fractions from patient ID JM03 showing a heterozygous mutation for KIT(N822K) before chemotherapy but not after.

Furthermore, although CD4 expression has been occasionally observed in AML blasts (Schwonzen et al., 2007), quantitation of CD4 and CD34 mRNA within our purified CD34+ fractions failed to detect a significant level of the former message in the blasts (Supplementary Figure S6). Therefore, it is unlikely that contamination of CD4+ leukemic blasts within the purified, control CD4+ fraction substantially affected the sequencing results in our phase II experiment.

Rather, it is more likely that leukemia may develop in a stepwise manner with a substantial time interval between steps (Figure 4c). If a first hit occurs in the genome of hematopoietic stem (or progenitor) cells and if such a somatic change does not result directly in the generation of full-blown leukemia, the preleukemic clones may give rise to terminally differentiated blood cells (including CD4+ cells). After a certain period, a second (or possibly a third) hit occurs in the immature cells and triggers the rapid growth of leukemic clones without differentiation. In such a scenario, terminally differentiated ‘normal’ cells may still harbor the first hit in their genome.

Support for this latter possibility was provided by patient ID JM03, who had AML (M2 subtype) with a t(8;21) chromosome anomaly. Before chemotherapy, the genomic DNA of both CD34+ and CD4+ fractions from this patient harbored a heterozygous mutation of KIT that results in the production of a constitutively activated mutant protein, KIT(N822K) (Shimada et al., 2006) (Figure 4d). The same change was also detected in cDNA prepared from the CD34+ fraction (data not shown). Leukemic blasts in this patient were sensitive to standard chemotherapeutic regimens, and the patient underwent complete remission. Examination of CD34+ and CD4+ fractions obtained during the remission period revealed that the N822K codon change was no longer detectable not only in the CD34+ fraction but also in the CD4+ fraction (Figure 4d). These data thus support the scenario shown in Figure 4c: The N822K change represents the first hit and was present in differentiated blood cells, and the corresponding preleukemic clones were simultaneously eradicated together with the leukemic clones by chemotherapy.

On the other hand, as shown in Supplementary Tables S1 and S2, a heterozygous mutation for NRAS(G12S) was found only in the CD34+ fraction, but not in the CD4+ fraction of the patient ID JM16. Conventional chemotherapy for this patient eradicated the leukemic blasts carrying the mutation (Supplementary Figure S7), also confirming that a successful treatment results in the disappearance of cells with a (possible) ‘second hit’.

Our hypothesis of the stepwise leukemogenesis is also consistent with the previous detection of the RUNX1-CBFA2T1 oncogene in differentiated blood cells (Kwong et al., 1996; Miyamoto et al., 1996, 2000).


Our large-scale genomic resequencing of human leukemia specimens with DNA microarrays has identified recurrent nucleotide changes responsible for the generation of JAK3 and DNMT3A mutants. Whereas JAK3 mutants were unexpectedly found in adult AML, their transforming ability, and possibly their contribution to leukemogenesis, varied substantially. However, our bone marrow transplantation experiments showed that at least one of these JAK3 mutants (M511I) directly participates in the development of leukemia. Identification of the M511I mutation of JAK3 in the leukemic fraction but not in the control fraction of patient ID JM07 suggests that this mutation may be the second hit triggering AML. Given that the blasts of this patient had a normal karyotype, it is likely that the first hit is present in the genome of both fractions. Karyotyping of other patients with JAK3 mutations showed a total of three cases with a normal karyotype, one case with t(8;21), and one case with a numerical anomaly of several chromosomes (Supplementary Table S3), suggesting that JAK3 mutations may be preferentially associated with leukemia with a normal karyotype.

Although JAK3(M511I) was identified in AML, our bone marrow transplantation experiments with hematopoietic stem cells expressing this mutant yielded T-cell acute lymphoblastic leukemia. In contrast to human leukemia, in which JAK3 changes may constitute a second hit (probably in progenitor cells), JAK3(M511I) may have been expressed in all hematopoietic cells of the recipient mice. JAK3(M511I) thus likely triggered leukemia within a T-cell fraction the intracellular context of which is optimized for JAK3 signaling.

It has been frequently observed that transgenic mouse or bone marrow transplantation experiments for leukemic oncogenes do not accurately recapitulate the original leukemia subtypes (Wong and Witte, 2001). Transgenic mice expressing p210BCR-ABL1, for instance, usually develop T-cell lymphoma or acute lymphoblastic leukemia, not chronic myeloid leukemia. Furthermore, bone marrow transplantation with hematopoietic progenitor cells expressing p210BCR-ABL1 often leads to development of lymphoma, AML, acute lymphoblastic leukemia or macrophage tumors. Generation of malignancy in such systems may, thus, be elaborately influenced by mouse strains, promoter fragments for artificial expression and/or cell types to be used for gene transduction.

Our detection of recurrent DNMT3A hypomorphic mutations in leukemia clones may indicate the presence of an abnormal methylation profile in the genome of such blasts. However, given the limited amount of the specimens available, we were able to investigate microsatellite stability only at certain loci (Koinuma et al., 2005), revealing no apparent microsatellite instability (data not shown). We also generated BA/F3 cells expressing wild-type or R882H forms of DNMT3A to compare the methylation status of some CpG islands in the genome; again, we detected no discernable differences between the two cell preparations (data not shown). However, given that BA/F3 cells contained two copies of wild-type Dnmt3a in addition to multiple copies of mutant DNMT3A, whereas the leukemic blasts likely harbor one copy each of the wild-type and mutant DNMT3A alleles, the clinical relevance of the R882 mutant requires further examination under the latter condition. Cell proliferation/differentiation is indeed influenced substantially by the copy number of DNMT3 genes (Okano et al., 1999; Ehrlich, 2003).

Our observations indicate the importance of preparing paired normal fractions in large-scale resequencing projects, but they also reveal a difficulty in the preparation of bona fide ‘normal’ fractions in the case of leukemic disorders. Our data thus indicate that nonleukemic blood cells may harbor early genomic hits, rendering them inappropriate as controls. Furthermore, a substantial proportion of fingernail DNA was recently shown to be derived from donor cells among recipients of allogeneic stem cell transplants (Imanishi et al., 2007), indicating that nonblood cells may contain DNA derived from transplanted cells. Therefore, it is possible that buccal, fingernail or even hair cells may not be suitable as normal cell controls. In contrast to solid tumors, for which blood cells are appropriate as paired normal fractions, leukemic disorders require that caution be taken to discriminate somatic nucleotide changes from germline polymorphisms.

Materials and methods

Wafer sequencing

CD34+ and CD4+ fractions were isolated from leukemic individuals using CD34microbeads and CD4microbeads, respectively, and a MidiMACS separator (Miltenyi Biotec, Gladbach, Germany). All clinical specimens were obtained with written informed consent, and the study was approved by the ethics committees of both the Jichi Medical University and the Nagasaki University. DNA sequencing wafers were designed and processed at Perlegen Sciences. Genes to be interrogated on the wafers were selected from the Entrez Gene database ( by searching with various keywords characteristic to each subcategory (such as DNA repair, regulation of chromatin structure, etc.), followed by manual inspection. The final gene list for the wafers is shown in Supplementary Table S6. Construction of the wafers, quality control analysis and data processing are described in Supplementary Text.

JAK3 analysis

Complementary DNAs for JAK3 mutants were generated using a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA, USA) and ligated into the pMX retroviral vector (Onishi et al., 1996). Ecotropic recombinant retroviruses encoding each mutant were produced in BOSC23 cells transfected with the corresponding pMX-based plasmid and were used to infect BA/F3 or 32D cells as described previously (Choi et al., 2007). Both types of cell were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (both from Life Technologies, Carlsbad, CA, USA) and mouse IL-3 (Sigma, St Loius, MO, USA) at 10 Units/ml; differentiation of 32D cells was induced by culture in the presence of serum and mouse granulocyte colony-stimulating factor (Sigma) at 0.5 ng/ml. A concentrated preparation of a retrovirus with a VSV-G envelope and encoding both JAK3(M511I) and enhanced green fluorescent protein was used to infect CD34 c-Kit+ Sca-1+ Lineage-marker (CD34KSL) hematopoietic stem cells isolated from the bone marrow of C57BL/6 mice, and the infected cells were transplanted into lethally irradiated mice congenic for the Ly5 locus (Iwama et al., 2004). CD4, JAK2 and JAK3 mRNAs were quantitated by reverse transcription and real-time PCR analysis using an ABI7900HT system (Life Technologies) and with the primers 5′-CTGGAATCCAACATCAAGGTTCTG-3′ and 5′-AATTGTAGAGGAGGCGAACAGGAG-3′ for CD4, 5′-CTCCAGAATCACTGACAGAGAGCA-3′ and 5′-CCACTCGAAGAGCTAGATCCCTAA-3′ for JAK2 and 5′-GAGCTCTTCACCTACTGCGACAAA-3′ and 5′-AGCTATGAAAAGGACAGGGAGTGG-3′ for JAK3; the cDNA for GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was also amplified with the primers 5′-GTCAGTGGTGGACCTGACCT-3′ and 5′-TGAGCTTGACAAAGTGGTCG-3′. The relative abundance of the cDNAs of interest was calculated from the threshold cycle (CT) for each cDNA and that for GAPDH cDNA.

DNMT3A analysis

Recombinant His6-tagged DNMT3A or DNMT3A(R882H) was expressed in SF9 cells using the Bac-to-Bac baculovirus expression system (Invitrogen, Carlsbad, CA, USA), and each protein was purified by stepwise column chromatography as described previously (Suetake et al., 2003). The enzymatic activity of each protein was assayed with -S-adenosyl-L-methionine (GE Healthcare, Waukesha, WI, USA) and dIdC or dGdC as substrates (Suetake et al., 2003). The association between Myc epitope-tagged human DNMT3L and wild-type or R882H forms of human DNMT3A in transfected HEK293 cells was examined by immunoprecipitation and immunoblot analyses.


  1. Bentley DR, Balasubramanian S, Swerdlow HP, Smith GP, Milton J, Brown CG et al. (2008). Accurate whole human genome sequencing using reversible terminator chemistry. Nature 456: 53–59.

    CAS  Article  Google Scholar 

  2. Byrd JC, Mrozek K, Dodge RK, Carroll AJ, Edwards CG, Arthur DC et al. (2002). Pretreatment cytogenetic abnormalities are predictive of induction success, cumulative incidence of relapse, and overall survival in adult patients with de novo acute myeloid leukemia: results from cancer and leukemia Group B (CALGB 8461). Blood 100: 4325–4336.

    CAS  Article  Google Scholar 

  3. Choi YL, Kaneda R, Wada T, Fujiwara S, Soda M, Watanabe H et al. (2007). Identification of a constitutively active mutant of JAK3 by retroviral expression screening. Leuk Res 31: 203–209.

    CAS  Article  Google Scholar 

  4. Ehrlich M . (2003). The ICF syndrome, a DNA methyltransferase 3B deficiency and immunodeficiency disease. Clin Immunol 109: 17–28.

    CAS  Article  Google Scholar 

  5. El-Osta A . (2004). The rise and fall of genomic methylation in cancer. Leukemia 18: 233–237.

    CAS  Article  Google Scholar 

  6. Greenberger JS, Sakakeeny MA, Humphries RK, Eaves CJ, Eckner RJ . (1983). Demonstration of permanent factor-dependent multipotential (erythroid/neutrophil/basophil) hematopoietic progenitor cell lines. Proc Natl Acad Sci USA 80: 2931–2935.

    CAS  Article  Google Scholar 

  7. Greenman C, Stephens P, Smith R, Dalgliesh GL, Hunter C, Bignell G et al. (2007). Patterns of somatic mutation in human cancer genomes. Nature 446: 153–158.

    CAS  Article  Google Scholar 

  8. Grimwade D, Walker H, Oliver F, Wheatley K, Harrison C, Harrison G et al. (1998). The importance of diagnostic cytogenetics on outcome in AML: analysis of 1612 patients entered into the MRC AML 10 trial. The Medical Research Council Adult and Children's Leukaemia Working Parties. Blood 92: 2322–2333.

    CAS  Google Scholar 

  9. Imanishi D, Miyazaki Y, Yamasaki R, Sawayama Y, Taguchi J, Tsushima H et al. (2007). Donor-derived DNA in fingernails among recipients of allogeneic hematopoietic stem-cell transplants. Blood 110: 2231–2234.

    CAS  Article  Google Scholar 

  10. Iwama A, Oguro H, Negishi M, Kato Y, Morita Y, Tsukui H et al. (2004). Enhanced self-renewal of hematopoietic stem cells mediated by the polycomb gene product Bmi-1. Immunity 21: 843–851.

    CAS  Article  Google Scholar 

  11. Jia D, Jurkowska RZ, Zhang X, Jeltsch A, Cheng X . (2007). Structure of Dnmt3a bound to Dnmt3L suggests a model for de novo DNA methylation. Nature 449: 248–251.

    CAS  Article  Google Scholar 

  12. Kiyoi H, Yamaji S, Kojima S, Naoe T . (2007). JAK3 mutations occur in acute megakaryoblastic leukemia both in Down syndrome children and non-Down syndrome adults. Leukemia 21: 574–576.

    CAS  Article  Google Scholar 

  13. Koinuma K, Kaneda R, Toyota M, Yamashita Y, Takada S, Choi YL et al. (2005). Screening for genomic fragments that are methylated specifically in colorectal carcinoma with a methylated MLH1 promoter. Carcinogenesis 26: 2078–2085.

    CAS  Article  Google Scholar 

  14. Kralovics R, Passamonti F, Buser AS, Teo SS, Tiedt R, Passweg JR et al. (2005). A gain-of-function mutation of JAK2 in myeloproliferative disorders. N Engl J Med 352: 1779–1790.

    CAS  Article  Google Scholar 

  15. Kubonishi I, Miyoshi I . (1983). Establishment of a Ph1 chromosome-positive cell line from chronic myelogenous leukemia in blast crisis. Int J Cell Cloning 1: 105–117.

    CAS  Article  Google Scholar 

  16. Kwong YL, Wong KF, Chan V, Chan CH . (1996). Persistence of AML1 rearrangement in peripheral blood cells in t(8;21). Cancer Genet Cytogenet 88: 151–154.

    CAS  Article  Google Scholar 

  17. Ley TJ, Mardis ER, Ding L, Fulton B, McLellan MD, Chen K et al. (2008). DNA sequencing of a cytogenetically normal acute myeloid leukaemia genome. Nature 456: 66–72.

    CAS  Article  Google Scholar 

  18. Miyamoto T, Nagafuji K, Akashi K, Harada M, Kyo T, Akashi T et al. (1996). Persistence of multipotent progenitors expressing AML1/ETO transcripts in long-term remission patients with t(8;21) acute myelogenous leukemia. Blood 87: 4789–4796.

    CAS  PubMed  Google Scholar 

  19. Miyamoto T, Weissman IL, Akashi K . (2000). AML1/ETO-expressing nonleukemic stem cells in acute myelogenous leukemia with 8;21 chromosomal translocation. Proc Natl Acad Sci USA 97: 7521–7526.

    CAS  Article  Google Scholar 

  20. Nimer SD, Moore MA . (2004). Effects of the leukemia-associated AML1-ETO protein on hematopoietic stem and progenitor cells. Oncogene 23: 4249–4254.

    CAS  Article  Google Scholar 

  21. Okano M, Bell DW, Haber DA, Li E . (1999). DNA methyltransferases Dnmt3a and Dnmt3b are essential for de novo methylation and mammalian development. Cell 99: 247–257.

    CAS  Article  Google Scholar 

  22. Onishi M, Kinoshita S, Morikawa Y, Shibuya A, Phillips J, Lanier LL et al. (1996). Applications of retrovirus-mediated expression cloning. Exp Hematol 24: 324–329.

    CAS  PubMed  Google Scholar 

  23. Patil N, Berno AJ, Hinds DA, Barrett WA, Doshi JM, Hacker CR et al. (2001). Blocks of limited haplotype diversity revealed by high-resolution scanning of human chromosome 21. Science 294: 1719–1723.

    CAS  Article  Google Scholar 

  24. Pikman Y, Lee BH, Mercher T, McDowell E, Ebert BL, Gozo M et al. (2006). MPLW515L is a novel somatic activating mutation in myelofibrosis with myeloid metaplasia. PLoS Med 3: e270.

    Article  Google Scholar 

  25. Russell SM, Tayebi N, Nakajima H, Riedy MC, Roberts JL, Aman MJ et al. (1995). Mutation of Jak3 in a patient with SCID: essential role of Jak3 in lymphoid development. Science 270: 797–800.

    CAS  Article  Google Scholar 

  26. Sato T, Toki T, Kanezaki R, Xu G, Terui K, Kanegane H et al. (2008). Functional analysis of JAK3 mutations in transient myeloproliferative disorder and acute megakaryoblastic leukaemia accompanying Down syndrome. Br J Haematol 141: 681–688.

    CAS  Article  Google Scholar 

  27. Schlenk RF, Dohner K, Krauter J, Frohling S, Corbacioglu A, Bullinger L et al. (2008). Mutations and treatment outcome in cytogenetically normal acute myeloid leukemia. N Engl J Med 358: 1909–1918.

    CAS  Article  Google Scholar 

  28. Schwonzen M, Diehl V, Dellanna M, Staib P . (2007). Immunophenotyping of surface antigens in acute myeloid leukemia by flow cytometry after red blood cell lysis. Leuk Res 31: 113–116.

    CAS  Article  Google Scholar 

  29. Shimada A, Taki T, Tabuchi K, Tawa A, Horibe K, Tsuchida M et al. (2006). KIT mutations, and not FLT3 internal tandem duplication, are strongly associated with a poor prognosis in pediatric acute myeloid leukemia with t(8;21): a study of the Japanese Childhood AML Cooperative Study Group. Blood 107: 1806–1809.

    CAS  Article  Google Scholar 

  30. Sjoblom T, Jones S, Wood LD, Parsons DW, Lin J, Barber TD et al. (2006). The consensus coding sequences of human breast and colorectal cancers. Science 314: 268–274.

    Article  Google Scholar 

  31. Suetake I, Miyazaki J, Murakami C, Takeshima H, Tajima S . (2003). Distinct enzymatic properties of recombinant mouse DNA methyltransferases Dnmt3a and Dnmt3b. J Biochem 133: 737–744.

    CAS  Article  Google Scholar 

  32. Tallman MS, Altman JK . (2008). Curative strategies in acute promyelocytic leukemia. Hematol Am Soc Hematol Educ Program 2008: 391–399.

    Article  Google Scholar 

  33. Walters DK, Mercher T, Gu TL, O'Hare T, Tyner JW, Loriaux M et al. (2006). Activating alleles of JAK3 in acute megakaryoblastic leukemia. Cancer Cell 10: 65–75.

    CAS  Article  Google Scholar 

  34. Wheeler DA, Srinivasan M, Egholm M, Shen Y, Chen L, McGuire A et al. (2008). The complete genome of an individual by massively parallel DNA sequencing. Nature 452: 872–876.

    CAS  Article  Google Scholar 

  35. Wong S, Witte ON . (2001). Modeling Philadelphia chromosome positive leukemias. Oncogene 20: 5644–5659.

    CAS  Article  Google Scholar 

  36. Yamamoto Y, Kiyoi H, Nakano Y, Suzuki R, Kodera Y, Miyawaki S et al. (2001). Activating mutation of D835 within the activation loop of FLT3 in human hematologic malignancies. Blood 97: 2434–2439.

    CAS  Article  Google Scholar 

Download references


We thank D Cox, KA Frazer, DG Ballinger, J Montgomery, H Tao, C Chen, L Stuve, J Kwon, J Sheehan and Y Zhan for discussion on the wafer experiments, as well as JN Ihle, T Kitamura and SB Baylin for human JAK3 cDNA, the pMX plasmid and human DNMT3A cDNA, respectively. This study was supported in part by a grant for Third-Term Comprehensive Control Research for Cancer from the Ministry of Health, Labor, and Welfare of Japan, and by a grant for Scientific Research on Priority Areas ‘Applied Genomics’ from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

Author information



Corresponding author

Correspondence to H Mano.

Ethics declarations

Competing interests

The authors declare no conflict of interest.

Additional information

Supplementary Information accompanies the paper on the Oncogene website

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Yamashita, Y., Yuan, J., Suetake, I. et al. Array-based genomic resequencing of human leukemia. Oncogene 29, 3723–3731 (2010).

Download citation


  • resequencing
  • AML
  • JAK3
  • DNMT3A

Further reading


Quick links