Identification of cooperative genes for E2A-PBX1 to develop acute lymphoblastic leukemia.

E2A-PBX1 is a chimeric gene product detected in t(1;19)-bearing acute lymphoblastic leukemia (ALL) with B-cell lineage. To investigate the leukemogenic process, we generated conditional knock-in (cKI) mice for E2A-PBX1, in which E2A-PBX1 is inducibly expressed under the control of the endogenous E2A promoter. Despite the induced expression of E2A-PBX1, no hematopoietic disease was observed, strongly suggesting that additional genetic alterations are required to develop leukemia. To address this possibility, retroviral insertional mutagenesis was used. Virus infection efficiently induced T-cell, B-cell, and biphenotypic ALL in E2A-PBX1 cKI mice. Inverse PCR identified eight retroviral common integration sites, in which enhanced expression was observed in the Gfi1, Mycn, and Pim1 genes. In addition, it is of note that viral integration and overexpression of the Zfp521 gene was detected in one tumor with B-cell lineage; we previously identified Zfp521 as a cooperative gene with E2A-HLF, another E2A-involving fusion gene with B-lineage ALL. The cooperative oncogenicity of E2A-PBX1 with overexpressed Zfp521 in B-cell tumorigenesis was indicated by the finding that E2A-PBX1 cKI, Zfp521 transgenic compound mice developed B-lineage ALL. Moreover, upregulation of ZNF521, the human counterpart of Zfp521, was found in several human leukemic cell lines bearing t(1;19). These results indicate that E2A-PBX1 cooperates with additional gene alterations to develop ALL. Among them, enhanced expression of ZNF521 may play a clinically relevant role in E2A fusion genes to develop B-lineage ALL.

Chromosomal abnormalities are a recurrent feature of human cancers. In human leukemias, a number of disease subtype‐specific chromosomal translocations are detected.1 These events induce aberrant expression of a translocation‐associated gene or generate a chimeric gene product fusing two different genes on different chromosomes, which is considered to play a critical role in the disease pathogenesis.

The E2A gene (also known as TCF3) encodes a basic helix‐loop‐helix transcription factor belonging to E‐box DNA‐binding proteins, which plays an essential role in B‐cell development.2 E2A is located on chromosome 19 and is the target in subsets of acute lymphoblastic leukemia (ALL) with B‐cell lineage.1 Two different chromosomal translocations involving E2A have been identified, t(1;19)(q23;p13) and t(17;19)(q22;p13), in which the E2A gene is fused to the PBX1 gene on chromosome 1 and the HLF gene on chromosome 17, thereby generating E2A‐PBX1 and E2A‐HLF chimeric gene products, respectively.3, 4, 5 In E2A‐PBX1, the transactivation domain of E2A is fused to the homeodomain of PBX1,3, 4 and in E2A‐HLF, the same domain of E2A is fused to the basic region/leucine zipper domain of HLF.5 Thus, in both fusion proteins, the DNA‐binding ability depends on the C‐terminal PBX1‐ or HLF‐derived region, whereas the transcription activation capacity resides in the N‐terminal E2A‐derived region.

To clarify the role of E2A‐PBX1 in leukemogenesis and to create a mouse model for E2A‐PBX1‐positive ALL, several different approaches have been made. Kamps et al. transduced E2A‐PBX1‐expressing retrovirus to hematopoietic progenitor cells and transplanted the cells into syngeneic mice.6 Although the recipient mice developed leukemia, the disease was exclusively classified as acute myeloid leukemia.6 Dedera et al. generated transgenic (Tg) mice expressing E2A‐PBX1 under the control of immunoglobulin (Ig) heavy‐chain promoter.7 The transgenic mice developed leukemia but the disease was mainly diagnosed as T‐cell ALL.7 Subsequently, Bijl et al. created lymphoid‐specific transgenic mice for E2A‐PBX1 and crossed the mice with CD3ε‐deficient mice to prevent the development T‐cell ALL.8 The compound (E2A‐PBX1 Tg and CD3ε‐deficient) mice mainly developed B‐cell malignancies, and the disease onset was accelerated with retroviral insertional mutagenesis, possibly in cooperation with deregulated expression of Hoxa genes.8 These results indicated that E2A‐PBX1 possesses oncogenic potential in various types of hematopoietic progenitors and suggest that E2A‐PBX1 renders oncogenicity to myeloid and T‐lymphoid cells rather than B‐lymphoid cells. Therefore, the mechanism of how E2A‐PBX1 is detected in B‐lineage ALL in the human is not yet clarified.

To address this issue and to create a more clinically relevant model for E2A‐PBX1‐positive leukemia, we generated conditional knock‐in (cKI) mice for E2A‐PBX1 in which E2A‐PBX1 is inducibly expressed under the control of the native E2A promoter.

Materials and Methods

Construction of a targeting vector and generation of cKI mice

The methods of construction of the cKI vector and generation of cKI mice were essential the same as previously described,9 except that Flag‐tagged human E2A‐PBX1 cDNA was used instead of human E2A‐HLF cDNA. WT/EPKI mice were crossed with MxCre 10 mice to generate WT/EPKI, MxCre mice. Cre activation was achieved by i.p. treatment with 500 μg polyinosinic–polycytidylic acid (pIpC) (Sigma, St. Louis, MO, USA) three times at 2 day intervals. Generation of EμSV/Zfp521 Tg mice was as previously described.9 In brief, a DNA fragment that contains EμSV enhancer/promoter,11 HA‐tagged Zfp521 cDNA,12 SV40 splicing, and polyA (pA) signals was microinjected into the pronuclei of mouse eggs. WT/EPKI mice were crossed with EμSV/Zfp521 Tg mice, and resultant offspring were subjected to pIpC treatment. Mice that have been back‐crossed to the C57BL/6N‐Ly5.2 background at least seven times were used for experiments. All the experiments were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the Hiroshima University Animal Research Committee.

Southern blot and genome PCR

Southern blot analyses and genome PCR were carried out as previously described.9

Immunoprecipitation and Western blot analyses

Immunoprecipitation and Western blot analyses were carried out as previously described.9 The anti‐Flag antibody was purchased from Sigma (#F7425).

Flow cytometric and gene rearrangement analyses

For flow cytometric analysis, cells were stained with phycoerythrin (PE)‐ or allophycocyanin (APC)‐conjugated anti‐Gr1, anti‐Mac1, anti‐Thy1.2, and anti‐B220 antibodies (Pharmingen, San Diego, CA, USA) and analyzed on BD FACSCanto II (BD Biosciences, San Jose, CA, USA) as previously described.9 For gene rearrangement analysis, DNA was digested with EcoRI and blotted with an indicated probe as previously described.9

Infection with MOL4070A and identification of retroviral integration sites

Preparation and infection of MOL4070A retrovirus were carried out as previously described.13 Newborn mice were inoculated i.p. with virus solution containing approximately 1 × 105 MOL4070A particles. Retroviral integration sites were identified by inverse PCR (iPCR) as previously described.13

Quantitative real‐time PCR

Total cellular RNA was extracted using the RNeasy Micro Kit (Qiagen, Valencia, CA, USA) and quantitative real‐time PCR (qPCR) was carried out as previously described.14 The primer sequences are as follows: mouse Evi5, 5′‐GACTCTTGGATTCTCTGGGGAAG‐3′ and 5′‐AACCAACCTCACGGTCCACTAAG‐3′; mouse Ccnd3, 5′‐CTTTGCATCTATACGGACCAGGC‐3′ and 5′‐CAACTCTGTGAGCTCATCCGCAG‐3′; mouse Cbfa2t3 h, 5′‐CCCCATCTACACCCAATGGCTTC‐3′ and 5′‐GGAAAGTTGGTGGCTTCCTGGAG‐3′; mouse Ikzf1, 5′‐CGTTGGTAAGCCTCACAAATGTG‐3′ and 5′‐CTTACGTTTGGCGACATTGCTTG‐3′; mouse Gfi1, 5′‐GTGGAGTCGGAGCTGCTTTGCAC‐3′ and 5′‐GAGCAGATGTGTGGACAGCGTGG‐3′; mouse Pecam1, 5′‐GTCGAGTAACGAGGTACAGTCTG‐3′ and 5′‐CGCAATGAGCCCTTTCTTCCATG‐3′; mouse Mycn, 5′‐GTCACCGTAGAGAAGAGACGTTC‐3′ and 5′‐GGTGAGGGTGCAGCATAGTTGTG‐3′; mouse Pim1, 5′‐TAAATGGTGCCTGTCCCTGAGAC‐3′ and 5′‐GTTGTCATTGCTGTCCCTGCATC‐3′; mouse Zfp521, 5′‐CCAGGTGTTTGAGTCACTGAGC‐3′ and 5′‐GAAATCACACCCTTCTCCATGG‐3′; and human ZNF521, 5′‐TACAATGAATGGGATATTCAGG‐3′ and 5′‐TCCCATCCCTTCGAAGCTGTGC‐3′.

Human leukemic cell lines

Human B‐progenitor cell lines with or without t(1;19) were used in this study. The cell lines with t(1;19) were KOPNK, KOPN34, KOPN36, KOPN54, KOPN60, KOPN63, YAMN90R, YAMN92, Kasumi2, and 697, and those without t(1;19) are KOPN55bi, KOPN57bi, KOPN72bi (these three lines are Ph‐positive), Reh (t(12;21)‐positive), and Nalm6.15

Results

Generation of cKI mice for E2A‐PBX1 and acquired expression of E2A‐PBX1

To investigate the leukemogenic role of E2A‐PBX1 and to create an animal model that mimics human leukemia with t(1;19)(q23;p13), we aimed to generate genetically engineered mice in which E2A‐PBX1 is inducibly expressed under the control of the native E2A promoter. To this end, we designed a knock‐in vector in which a floxed neomycin resistance (Neo) gene, Flag‐tagged E2A‐PBX1 cDNA, internal ribosomal entry site (IRES)‐GFP (IG), and a pA was inserted in the non‐coding region of exon 2 (Fig. 1a). Embryonic stem (ES) clones with homologous recombination identified by Southern blot using a 5′ probe and genomic PCR using a 3′ primer set were used to create chimeric mice (Fig. 1b), which transmitted the mutant allele and generated heterozygous (WT/EPKI) mice.

Figure 1

Generation of E2A‐ conditional knock‐in (cKI) mice and acquired expression of E2A‐. (a) Schematic illustration of the conditional KI strategy. Part of the non‐coding region of exon 2, the coding region of exon 2, intron 2, and part of the coding region of exon 3 were replaced with a floxed neomycin resistance gene (Neo), followed by Flag‐tagged E2A‐ cDNA, (), and a polyadenylation signal (). Restriction enzymes: C, ClaI; H, Hind III; N, NaeI; X, XbaI. The positions of the 5′ probe for Southern blot analysis and P1 and P2 primers for the 3′ genome PCR are shown. (b) Results of 5′ Southern blot and 3′ genomic PCR to detect homologous recombination. Positions of germline (GL) and KI allele‐derived bands by 5′ Southern blot and the PCR product by 3′ genomic PCR are indicated by arrows and an arrowhead, respectively. (c) Acquired E2A‐ expression. Proteins extracted from the spleen of and mice were immunoprecipitated with an anti‐E2A antibody and the immunoprecipitated proteins were blotted with an anti‐Flag antibody. The positions of E2A‐ protein and Ig are indicated by arrows. , diphtheria toxin A gene; Neo, neomycin resistance gene; Ip, immunoprecipitant; pBS, pBluescript; pIpC, polyinosinic–polycytidylic acid; Tg, transgenic.

The WT/EPKI mice do not express E2A‐PBX1 protein because of the existence of the Neo gene between the non‐coding region of exon 2 and E2A‐PBX1 cDNA (Fig. 1a, line 3). In contrast, crossing WT/EPKI mice with MxCre mice and treating the WT/EPKI, MxCre compound mice with pIpC produce WT/EPKI mice (Fig. 1a, line 4), which in turn express E2A‐PBX1 protein due to the excision of the Neo gene (Fig. 1a, line 4). To verify this, proteins extracted from the spleen of WT/EPKI and WT/EPKI mice were immunoprecipitated with an anti‐E2A antibody and immunoprecipitants were blotted with an anti‐Flag antibody. As expected, E2A‐PBX1 protein (molecular weight approximately 85 kDa) was detected in the spleen of a WT/EPKI mouse but not in that of a WT/EPKI mouse (Fig. 1c), indicating that our targeting strategy successfully generated cKI mice for E2A‐PBX1.

WT/EPKI, MxCre mice did not show hematopoietic diseases but developed ALL by retroviral insertional mutagenesis

The pIpC‐treated WT/EPKI, MxCre and WT/EPKI, MxCre mice (hereafter referred to as control and EPKI mice, respectively) were subjected to routine examination of peripheral blood parameters and continuously observed for any sign of illness. However, during 1 year of observation, no abnormality was detected in either type of mice (Fig. 2a, black and green lines). These results indicated that E2A‐PBX1 expression induced under the control of the E2A promoter alone is not sufficient to develop hematopoietic diseases and strongly suggest that additional genetic changes are required to produce a fully malignant phenotype.

Figure 2

Analysis of leukemias developed in MOL4070A‐infected mice. (a) Survival curves of control (Ctrl, black), (green), Ctrl+MOL4070A (blue), and +MOL4070A (red) mice and their symptomatic incidence are shown. The time points of MOL4070A infection and polyinosinic–polycytidylic acid (pIpC) injection are indicated by arrows and the diseased mice in the +MOL4070A group are numbered. (b) Pathological analysis of a tumor that developed in an +MOL4070A mouse. The infiltrated leukemic cells are indicated by arrows. (c) Flow cytometric analysis of +ML4070A tumors. The representative results of T‐cell (no. 3), B‐cell (no. 6), and biphenotypic (no. 1) acute lymphoblastic leukemia (ALL) are shown. (d) Gene rearrangement analysis. DNAs extracted from tumor tissues and a control spleen (C Spl) were blotted with a IgH or T‐cell receptor β () probe. The germline (GL) and rearranged bands are indicated by arrows and arrowheads, respectively.

To address this possibility, retroviral insertional mutagenesis (RIM) was used.16 Neonatal mice were infected with replication‐competent MOL4070LTR mouse leukemia retrovirus, a derivative of Moloney murine leukemia virus (MMLV).17 During the same observation period (1 year), all the EPKI mice infected with MOL4070A died of leukemias (Fig. 2a, red line; disease incidence, 13/13), whereas only one control mouse infected with MOL4070A developed leukemia (Fig. 2a, blue line; disease incidence, 1/12). The leukemias developed in EPKI+MOL4070A mice were highly malignant, as evidenced by marked proliferation of immature blast cells in the spleen and massive infiltration of leukemic cells in the non‐hematopoietic tissues, such as the lung, liver, and kidney (Fig. 2b).

Macroscopically, EPKI+MOL4070A mice showed splenomegaly, frequently associated with thymic enlargement and lymph node swelling (Table 1). To determine the lineage(s) and clonality of the leukemic cells, flow cytometric and gene rearrangement analyses were carried out. The results of flow cytometric analysis revealed that the leukemic cells expressed Thy1.2 (T‐cell marker), B220 (B‐cell marker), or both, but none was positive for Mac1 or Gr1 (myeloid markers) (Fig. 2c, Table 1), indicating that all the leukemias were committed to the lymphoid lineage. In addition, Southern blot analysis showed that most of the EPKI+MOL4070A tumors carried rearrangements at either or both IgH and T‐cell receptor β (TCRβ) loci (Fig. 2d, arrowheads) and the rearranged patterns mostly correspond to the results of the flow cytometric analysis (Table 1). Thus, the leukemias developed in EPKI+MOL4070A mice were diagnosed as T‐cell, B‐cell, or biphenotypic ALL and were mainly clonal in origin. These findings indicated that the expression of E2A‐PBX1 predisposes hematopoietic cells to malignant transformation and develops various types of ALL in cooperation with additional gene alterations.

Table 1

Characteristics of EPKI+MOL4070A leukemic mice

Mouse No.	Age at disease (days)	PB parameters			Macroscopic tumor sites	Surface markers	Gene rearrangements		Diagnosis
		WBC (×103/μL)	Hb (g/dL)	Plt (×104/μL)			IgH	TCRβ
1	150	27.1	14.2	17.2	Spl	Thy1.2+, B220+	G/R	G/R	Biphenotypic ALL
2	164	13.3	11.3	12.8	Spl, LN	Thy1.2+, B220+	G/R	G/R	Biphenotypic ALL
3	182	64.1	9.7	25.0	Thy, Spl	Thy1.2+	G/G	G/R	T‐lineage ALL
4†	190	ND	ND	ND	Spl	ND	G/R	G/G	B‐lineage ALL s/o
5	200	22.3	12.9	28.2	Spl, LN	Thy1.2+, B220+	G/R	G/R	Biphenotypic ALL
6	227	13.1	10.0	73.7	Spl, LN	B220+	R/R	G/G	B‐lineage ALL
7	235	17.2	13.4	26.7	Spl, LN	Thy1.2+, B220+	G/G	G/R	Biphenotypic ALL
8	250	ND	ND	ND	Spl, LN	Thy1.2+	G/R	G/G	T‐lineage ALL s/o
9	277	38.3	11.8	14.3	Spl	Thy1.2+	G/G	G/G	T‐lineage ALL
10	290	82.1	13.5	37.6	Thy, Spl, LN	Thy1.2+	G/G	G/R	T‐lineage ALL
11	318	45.2	10.3	7.1	Spl	Thy1.2+, B220+	G/R	G/G	Biphenotypic ALL
12	318	65.2	13.6	33.5	Thy, Spl	Thy1.2+, B220+	G/R	G/G	Biphenotypic ALL
13	328	10.4	12.2	13.5	Spl, LN	Thy1.2+, B220+	G/R	G/R	Biphenotypic ALL

ALL, acute lymphoblastic leukemia; G, germline; Hb, hemoglobin; LN, lymph node; ND, not done; PB, peripheral blood; Plt, platelet count; R, rearranged; s/o, suspected of; Spl, spleen; TCRβ, T‐cell receptor β; Thy, thymus; WBC, white blood cells. †Found dead.

Identification of common integration sites in leukemias developed in EPKI+MOL4070 mice

To identify gene(s) whose altered expression cooperated with E2A‐PBX1 to develop leukemia, genomic DNAs extracted from leukemic samples of EPKI+MOL4070 mice were subjected to iPCR, the method to isolate virus integration sites.16 In the iPCR products obtained from EPKI+MOL4070 tumors, we identified eight retroviral common integration sites (CIS), which were Evi5 (ecotropic viral integration site 5), Ccnd3 (cyclin D3), Cbfa2t3 h (core‐binding factor, runt domain, α subunit 2), Gfi1 (growth factor independent 1), Ikzf1 (IKAROS family zinc finger 1 (Ikaros)), Pecam1 (platelet/endothelial cell adhesion molecule 1), Mycn (v‐myc myelocytomatosis viral related oncogene, neuroblastoma derived), and Pim1 (proviral integration site 1) (Table S1, asterisks and bold type).

The viral integration sites were found in 5′ upstream and 3′ downstream regions, in introns and occasionally in exons (Fig. 3). Viral integrations are considered to contribute to disease progression through altering the expression of the affected genes.16 To investigate this possibility, expression patterns of the CIS genes in tumor tissues were examined by qPCR and compared to that in a control spleen. As shown in Figure 3, although expressional change was less than 2‐fold in the Evi5, Ccnd3, Cbfa2t3 h, Ikzf1, and Pecam1 genes, enhanced expression (>2‐fold) was detected in the Gfi1, Mycn, and Pim1 genes. In the Gfi1 gene, most of the tumors, including virus‐integrated nos. 2 and 12, showed more than 5‐fold upregulation (Fig. 3, left bottom panel). In the Mycn gene, among three virus‐integrated tumors (nos. 3, 10, and 13), two samples (nos. 10 and 13) showed significantly enhanced (>200‐fold) expression, along with one insertion‐undetectable case (no. 12) (Fig. 3, right‐hand panel, row 3). In addition, concerning the Pim1 gene, in the two virus‐integrated tumors (nos. 5 and 10), one sample (no. 10) showed more than 5‐fold upregulation (Fig. 3, bottom right panel). These results strongly suggested that, among the eight CIS genes, integrations in the Gfi1, Mycn, and Pim1 genes enhanced the expression patterns and possibly contribute to development and progression of leukemias in EPKI+MOL4070A mice.

Figure 3

Analysis of retroviral integration sites and altered gene expression patterns in tumors of +ML4070A mice. In the schematic illustrations of common integration site (CIS) genes (upper panel of each gene figure), exons are boxed, coding regions are filled, and viral integration sites are indicated by vertical arrows with the related mouse numbers. In the quantitative real‐time PCR analysis (lower panel of each gene figure), the expression levels of common integration sites in +ML4070A tumors relative to that in a control spleen (C Spl) are shown. *Tumors with viral integrations.

Identification of Zfp521 as the major retroviral integration site in a leukemic mouse with B‐lineage phenotype and cooperative oncogenicity of overexpressed Zfp521 with E2A‐PBX1 in B‐cell tumorigenesis

Among the EPKI+MOL4070A leukemic mice, no. 6 attracted our attention with two reasons: (i) the leukemia of no. 6 was diagnosed as B‐lineage ALL (Fig. 2c, Table 1), which correlates with the phenotype of human E2A‐PBX1‐positive leukemia;18 and (ii) Zfp521 (also known as Evi3) was detected as a viral integration site (Table S1, double asterisks and bold type), which we previously identified as a CIS in our E2A‐HLF cKI RIM study and reported as a cooperative gene with E2A‐HLF to develop B‐lineage ALL.9 Therefore, it is strongly postulated that Zfp521 may participate in the development of B‐lineage leukemia in EPKI cKI mice as well as in E2A‐HLF cKI mice.

The viral integration site of no.6 in the Zfp521 gene was located in the promoter region, which was very close to those detected in two EHKI+MMLV B‐lineage ALL samples (EHKI‐2 and EHKI‐4) (Fig. 4a).9 To investigate whether cells with Zfp521 integration account for the majority of the tumor, Southern blot analysis was carried out using a genomic probe adjacent to the viral integration sites (Fig. 4a, probe E).9 A rearranged band was detected in no. 6 as well as EHKI‐2 and EHKI‐4 (Fig. 4b, arrowheads), indicating that Zfp521 was the major integration site in these three tumors. We then examined the expression alteration of the Zfp521 gene by virus integration. As shown in Figure 4(c), more than 2‐fold upregulation of Zfp521 expression was detected in no. 6 as well as EHKI‐2 and EHKI‐4.

Figure 4

Analysis of Zfp521 in no. 6 and ‐2 and ‐4 tumors. (a) Schematic illustration of the viral integration sites. Exon 1 is boxed and the coding region is filled. The viral integration sites in no. 6, ‐2, and ‐4 are indicated by vertical arrows. The position of the probe for Southern blot (probe E) is also shown. (b) Southern blot to detect gene rearrangements by viral integration. The germline (GL) and rearranged bands are indicated by an arrow and arrowheads, respectively. (c) Quantitative real‐time PCR analysis for Zfp521 expression. The expression levels of Zfp521 in +MOL4070A tumors, ‐2, and ‐4 relative to that in a control spleen (C Spl) are shown. *Tumors with viral integrations. (d) Cooperative oncogenicity of overexpressed Zfp521 with E2A‐. Survival curves of , Zfp521 transgenic (Tg), and Tg compound mice are shown with thin continuous, thin dotted, and thick continuous lines, respectively. The time point of polyinosinic–polycytidylic acid injection is indicated by an arrow and the diseased /Zfp521 Tg mice are numbered. (e) Flow cytometric and gene rearrangement analyses of leukemias developed in /Zfp521 Tg mice. The representative result of flow cytometry and the results of gene rearrangement are shown in the upper and lower panels, respectively. The germline (GL) and rearranged bands are indicated by arrows and arrowheads, respectively.

To directly verify the in vivo cooperative oncogenicity of overexpressed Zfp521 with E2A‐PBX1, we crossed EPKI mice with EμSV/Zfp521 (Zfp521) Tg mice that express Zfp521 at a high level in lymphoid cells.9 The survival curves of pIpC‐treated offspring are shown in Figure 4(d). During approximately 5 months of observation, more than half of the EPKI/Zfp521 Tg compound mice developed acute leukemia, while none of the EPKI or Zfp521 Tg alone mice showed hematological abnormalities. The flow cytometric and gene rearrangement analyses of the leukemic tissues indicated that the leukemias were of B‐cell lineage (Fig. 4e). These results showed that overexpressed Zfp521 synergized with EPKI and contributed to B‐cell tumorigenesis.

Enhanced expression of ZNF521 in human leukemic cell lines with t(1;19)

We finally investigated the clinical relevance of overexpression of Zfp521 in human leukemias bearing t(1;19). To this end, the expression levels of ZNF521, the human counterpart of Zfp521, were examined in t(1;19)‐positive ALL cell lines and the results were compared with those in B‐lineage ALL lines without t(1;19). As shown in Figure 5, although the expression levels of ZNF521 varied among lines, the mean expression level in ALL lines with t(1;19) was apparently higher than that in control lines without t(1;19) and marked upregulation (more than 10‐fold of the mean of control cell lines) were observed in three lines (Fig. 5, arrows). These results strongly suggested that overexpression of ZNF521 would be implicated, at least in several cases, in the pathogenesis of t(1;19)‐positive B‐lineage ALL.

Figure 5

Quantitative real‐time PCR analysis of in human leukemic cell lines with or without t(1;19). The mRNA expression levels in five t(1;19)‐negative B‐progenitor cell lines (control) and 10 t(1;19)‐positive cell lines (t(1;19)+) relative to the mean of the control cell lines (white bar) are indicated by white and black circles, respectively. The mean of t(1;19)+ cell lines is indicated by a black bar. The relative expression ratio (vertical bar) is shown on a logarithmic scale. High expression lines with t(1;19) are indicated by arrows.

Discussion

E2A‐PBX1 is detected in approximately 5% of ALL patients and associated with B‐cell phenotype.18 To clarify the leukemogenic mechanism and to create a mouse model for E2A‐PBX1‐positive human leukemia, we generated mice in which E2A‐PBX1 was inducibly expressed under the control of the native E2A promoter (Fig. 1a). Despite the induced expression of E2A‐PBX1 in the hematopoietic tissue (Fig. 1c), no disease developed during the observation period (Fig. 2a), indicating that expression of E2A‐PBX1 per se is not sufficient to develop hematopoietic malignancies. This finding is in good agreement with our previous study using E2A‐HLF cKI mice9 and other cKI studies for leukemia‐associated fusion genes, such as RUNX1/ETO and MLL/CBP.19, 20 To identify cooperative gene(s) for E2A‐PBX1 to exert its fully oncogenic potential, E2A‐PBX1 cKI mice were subjected to RIM study, by which we have successfully isolated cooperative genes in various types of genetically engineered mice.9, 13, 14, 21, 22, 23 By MOL4070A infection, all the E2A‐PBX1 cKI mice developed leukemia; in contrast, only one control mouse showed hematopoietic malignancy (Fig. 2a), indicating that E2A‐PBX1 confers high susceptibility to leukemia development. It is of note that the leukemia that developed in control+MOL4070A was of T‐cell lineage (not shown), whereas the leukemias that developed in EPKI+MOL4070A mice were diagnosed as either T‐cell, B‐cell, or biphenotypic ALL (Fig. 2c,d, Table 1). Thus, it is strongly suggested that the expression of E2A‐PBX1 primed virus‐integrated cells to B‐lineage ALL in addition to T‐lineage ALL, as reported in previous studies.7, 8

Using iPCR, we detected eight retroviral common integration sites, Evi5, Ccnd3, Cbfa2t3 h, Gfi1, Ikzf1, Pecam1, Mycn, and Pim1 (Table S1), which have already been identified by retrovirus‐tagged mouse mutagenesis studies using various types of genetically engineered mice (Retrovirus and Transposon tagged Cancer Gene Database, http://variation.osu.edu/rtcgd/). Quantitative real‐time PCR showed that three of them, Gfi1, Mycn, and Pim1, exhibited high expression patterns (Fig. 3), leading to the idea that deregulated expression of these genes contributes to the leukemogenic process(es).

Gfi1 encodes a transcription factor and was originally identified as the gene that conferred interleukin‐2 (IL‐2)‐independent growth ability to T‐lymphocytes.24 Subsequent studies reported Gfi1 as a frequent viral integration site in T‐lymphoid tumors in MMLV‐infected mice.25, 26 Although targeted expression of Gfi1 did not efficiently induce leukemia in mice, it cooperates with other genes, such as Myc or Pim1, to develop T‐cell malignancies.27 We found that overexpression of Gfi1 (>5‐fold) in E2A‐PBX1 + MOL4070A mice with T‐cell lineage at high frequency (10/13), indicating that Gfi1 could be a new partner for E2A‐PBX1 to develop T‐cell leukemia.

Mycn is a member of the MYC family of transcription factors. Mycn was originally identified as a gene amplified in neuroblastoma cells and subsequently found to be overexpressed in various types of cancers.28 The contribution of N‐myc to leukemia development was verified by transgenic and bone marrow transplantation (BMT) studies; targeted expression of N‐myc by lymphoid‐specific transgenic enhancer/promoter induced B‐lymphoid malignancies,29, 30, 31 and retrovirus‐mediated transfer of N‐myc into hematopoietic progenitor cells developed acute myeloid leukemia.32 In addition, in a RIM study, N‐myc was found as a viral integration site and accelerated the onset of T‐cell ALL in N‐ras and Pim1 Tg mice.33, 34 Therefore, it may be possible that overexpressed N‐myc cooperates with E2A‐PBX1 and enhances its oncogenic activity to develop ALL.

Pim1 encodes a serine–threonine kinase that is involved in the regulation of apoptosis, metabolism, and the cell cycle.35 A RIM study applied to E2A‐PBX1 Tg mice7 identified Pim1 as a frequent target of retrovirus insertions in mice with accelerated diseases.36 Therefore, our study provided further evidence that deregulated expression of Pim1 cooperates with E2A‐PBX1 to develop aggressive T‐cell ALL. Given that Pim1 contributes to factor independency37 and E2A‐PBX1‐positive hematopoietic cells tend to undergo apoptosis, especially under cytokine‐depleted conditions,7 it is postulated that Pim1 confers cytokine‐independent growth ability to E2A‐PBX1‐positive cells and induces a more malignant phenotype.

Of particular interest is the identification of Zfp521 in an EPKI+MOL4070A tumor with B‐cell phenotype (no. 6; Fig. 4, Table S1). Zfp521 was originally cloned as a retroviral integration site in AKXD mice, which encodes a transcription factor with multiple zinc fingers and is highly expressed in hematopoietic progenitor cells.12 Recent studies have reported the overexpression of ZNF521, the human counterpart of Zfp521 (also known as EHZF), in human B‐cell lymphoblastic lymphoma samples38 and the detection of the Pax5/ZNF521 fusion gene in pre‐B ALL samples.39, 40 Thus, it is strongly postulated that deregulated and/or structurally altered ZNF521 is implicated in B‐cell malignancy. Our findings, together with our previous study,9 indicate that overexpressed Zfp521 cooperated with E2A‐involving fusion genes to develop B‐lineage ALL and that enhanced expression of ZNF521 is detected in human B‐lineage ALL samples bearing t(1;19) or t(17;19) (Figs 4, 5). Although the mechanism(s) of how deregulated Zfp521 contributes to B‐cell leukemogenesis has not fully been understood, studies have reported that Zfp521 impairs normal B‐cell development by inhibiting the function of EBF1,41, 42 a transcription factor required for B‐cell development.43 In addition, Zfp521 was shown to enhance pre‐B‐cell receptor signaling and interfere with the IL‐7/IL‐7 receptor‐mediated maturation pathway.38 Therefore, it is postulated that Zfp521/ZNF521 contributes to the development of B‐lineage ALL by expanding the pre‐B cell population and impairing the terminal differentiation toward mature B cells.

Recently, generation and analysis of another cKI mouse for E2A‐PBX1 was reported.44 The researchers knocked‐in PBX1 cDNA coupled with IRES‐GFP downstream of the E2A gene and inserted two loxP sites together with the Neo resistance gene, where one loxP was located between exons 12 and 13 and another between the last exon of the E2A gene and PBX1 cDNA.44 By crossing the knock‐in mice with various types of Cre (MxCre +, Mb1Cre +, and CD19Cre +) Tg mice, the compound mice (named as conditional E2A‐PBX1 Tg mice) expressed E2A‐PBX1, as in our study. However, unlike our EPKI mice, the conditional E2A‐PBX1 Tg mice spontaneously developed leukemias with B‐cell lineage.44 The phenotypical discrepancy between the two studies remains unknown, but one possibility is the difference in the knock‐in strategy. We directly knocked‐in E2A‐PBX1 cDNA in the 5′ non‐coding region of the E2A gene, but Duque‐Afonso et al. retained all the exon/intron structure of the E2A gene and knocked‐in PBX1 cDNA downstream of the E2A gene.44 Thus, it is strongly suggested that unidentified regulatory region(s) in the exon/intron(s) of the E2A gene may play a pivotal role in the spontaneous development and lineage‐determination of E2A‐PBX1‐positive (and presumably E2A‐HLF‐positive) leukemias. They showed that secondary mutations, including loss of Pax5, were frequently detected in the leukemic tissues,44 therefore it is conceivable that additional genetic aberrations are necessary for E2A‐PBX1 to fully exert its leukemogenic potential.

In this report, we generated E2A‐PBX1 cKI mice, isolated cooperative candidate genes, identified Zfp521 as a partner to develop B‐lineage ALL, and showed the frequent overexpression of ZNF521 in t(1;19)‐positive ALL samples. Our results, together with the findings of other groups,8, 44, 45 provide evidence that multistep gene alterations are required for E2A‐PBX1 to develop ALL and prove that RIM study is a valuable tool for identifying genes whose altered expression contributes to malignant transformation of hematopoietic cells.

Disclosure Statement

The authors have no conflict of interest.

Table S1. Retroviral integration sites identified by iPCR

Click here for additional data file.