Acquired loss of p53 induces blastic transformation in p210^bcr/abl-expressing hematopoietic cells: a transgenic study for blast crisis of human CML

Mice that were p210^bcr/abl transgenic (BCR/ABL^tg/−) were generated by using the mouse tec promoter (−1948 to +22) and the bcr/abl(p210) complementary DNA (cDNA; b3a2 type) as previously described.21 Because the founder mice were generated by using eggs derived from C57Bl/6 × DBA F2 (BDF2) mice and the transgenic progeny were generated by mating the transgenic mice with BDF1 mice, the genetic background of theBCR/ABL^tg/− mice was a mixture of C57Bl/6 and DBA. The p53-heterozygous (p53^+/−) mice were generated by using TT2 ES cells inserted with a neomycin-resistance cassette into the second exon of the p53 gene as previously described.24 Because the TT2 ES cells were established from an F1 embryo between C57Bl/6 and CBA25 and the p53-heterozygous mice were generated by mating the chimeric mice with C57Bl/6 mice, the genetic background of the p53^+/− mice was a mixture of C57Bl/6 and CBA. Therefore, the background of the mice used in this experiment was a mixture of C57Bl/6, DBA, and CBA. Identification of genotypes forBCR/ABL and p53 was carried out by hybridizingBamHI-digested tail DNA with bcr/abl cDNA andSacI-digested tail DNA with the 5′ flanking region of the mouse p53 gene, as previously described.21 24 

Autopsies were performed on dead or moribund animals. Smears and stamp specimens of leukemic tissues were stained with Wright-Giemsa stain. Tissues were also fixed in 10% neutral-buffered formaldehyde and subjected to routine light microscopical examination. All the organs were examined grossly and representative slices were prepared for hematoxylin-eosin staining.

For detection of p210^bcr/abl-transgene product, proteins were extracted by homogenizing leukemic tissues in RIPA lysis buffer (150 mmol/L sodium chloride, 50 mmol/L TRIS chloride [pH 7.4], 1% Triton X-100, 0.05% sodium dodecyl sulfate, and 1% sodium deoxycholate) with 50 U/mL of aprotinin and were probed with the anti-Abl monoclonal antibody AB3 (Oncogene Science, Manhasset, NY) as previously described.21 For detection of the kinase activity of the p210^bcr/abl-transgene product, protein aliquots were incubated with AB3 and immunoprecipitated proteins were subjected to in vitro kinase assay as previously described.21 For detection of p53 protein, nuclear fractions of the tissues were subjected to immunoprecipitation/Western blot analysis using anti-p53 antibodies (Oncogene Science) as previously described.24 

Cells were stained with fluorescein isothiocyanate-conjugated or phycoerythrin-conjugated commercial monoclonal antibodies, including anti-Thy-1.2, anti-B220, anti-CD4, and anti-CD8 (Pharmingen, San Diego, CA), according to the manufacturer's instructions. The stained cells were washed 3 times with phosphate-buffered saline and analyzed on a FACScan (Becton Dickinson, Sunnyvale, CA).

Genomic DNAs were extracted from tumor tissues and digested with restriction enzymes. Southern blot analysis was performed by using phosphorus (P) 32-dCTP–labeled mouse TCRγ as a probe as previously described.21 

PCR-SSCP analysis was performed essentially as previously described.26 In brief, genomic DNAs were amplified by using several different sets of primers with α³²P-dCTP. The PCR products were mixed with formamide dye, electrophoresed on acrylamide gels with or without glycerol, and autoradiographed.

Crossmating of p210^bcr/abl-transgenic (BCR/ABL^tg/−) mice with p53-heterozygous (p53^+/−) mice produced mice with 4 different genotypes, which wereBCR/ABL^−/−p53^+/+(wild-type),BCR/ ABL^−/−p53^+/−(p53 heterozygous),BCR/ABL^tg/−p53^+/+(p210^bcr/abl transgenic), andBCR/ABL^tg/−p53^+/−(p210^bcr/abl transgenic, p53 heterogeneous). The genotypes of the 58 total offspring were determined by Southern blot assessment using tail DNAs (Figure 1A). The numbers of the mice with each genotype were 16 forBCR/ABL^−/−p53^+/+(27.5%), 13 forBCR/ABL^−/−p53^+/−(22.4%), 15 forBCR/ABL^tg/−p53^+/+(25.9%), and 14 forBCR/ABL^tg/−p53^+/−(24.1%). These results were in good agreement with the expected ratio based on the Mendelian rule (25% each), indicating that the crossmating did not affect the embryonic development of the mice.

All mice were kept under the same maintenance conditions and were continually observed. The hematopoietic variables of the PB were routinely examined. SixteenBCR/ABL^−/−p53^+/+mice were all healthy during the observation period (Figure 1B, thin dotted line). Among the 13BCR/ABL^−/−p53^+/−mice, 1 mouse died with a gross subcutaneous tumor at 11 months of age (data not shown), but the other 12 mice survived in a healthy state for more than 1 year with no symptoms of illness (Figure B, thick dotted line). No obvious hematologic changes were observed in these mice (Figure 1C, upper panels). Among the 15BCR/ABL^tg/−p53^+/+ mice, 1 mouse died at 3 months of age from ALL and the remaining 14 gradually exhibited CML after 6 months of age, as previously reported21 (Figure 1C, left bottom panel). About half of the mice died within 1 year of age (Figure 1B, thin solid line). In contrast to the mice with these genotypes,BCR/ABL^tg/−p53^+/−mice showed high morbidity and mortality rates (Figure 1B, thick solid line). Eleven of 14BCR/ABL^tg/−p53^+/−mice died within 1 year. Although detailed analysis was not possible in 2 mice because of advanced autolysis, 9 mice were subjected to pathological, biochemical, and molecular analyses. A common macroscopic aspect of these mice was enlargement of the thymus (Table1). Thymoma was frequently associated with pleural effusion, and splenomegaly was observed in some mice (Table 1). Wright-Giemsa staining showed that blast cells with no granules and with morphologic characteristics of lymphoblasts proliferated markedly in the thymus (Figure 2A) and pleural effusion (Figure 2B).

The age at disease onset ranged from 2 to 11 months (Figure 1B, thick solid line, and Table 1). In mice with disease onset at a young age, no obvious hematologic changes were observed in PB (Figure 1C, right bottom panel), although the BM showed predominance of myeloid cells (data not shown). On the other hand, mice with disease onset in old age had typical symptoms of CML before thymomas developed; they showed leukocytosis with granulocyte proliferation in the PB (Figure 1B, right bottom panel, and Figure 2C) and marked myeloid hyperplasia in the BM (Figure 2D). Proliferation of blast cells in the PB was observed in some mice (data not shown). Pathological analysis showed that the blast cells infiltrated several major organs, such as the lung, liver, and kidney, in all mice examined (Figure 2E-2J).

To examine whether the blast cells inBCR/ABL^tg/−p53^+/− leukemic mice expressed the p210^bcr/abl-transgene product, proteins extracted from the enlarged thymus or pleural effusion were blotted with the anti-Abl monoclonal antibody AB3. As shown in Figure3A (arrow), the 210-kd band was detected in all the leukemic tissues, indicating that blast cells expressed the p210^bcr/abl-transgene product. Blotting the same samples with antiphosphotyrosine antibody showed that the expressed p210^bcr/abl was tyrosine phosphorylated (data not shown). To examine whether the expressed p210^bcr/abl had an enzymatically active kinase activity, proteins extracted from the tumor tissues were subjected to in vitro kinase assays. As shown in Figure 3B, phosphorylated 210-kd band was observed in all the leukemic tissues (arrow), indicating that p210^bcr/abl contained an increased kinase activity.

To determine the cell lineage of the leukemias, blast cells in the enlarged thymus or pleural effusion from 7BCR/ABL^tg/−p53^+/−leukemic mice were subjected to flow cytometric analysis. Staining the cells with anti-Thy1.2 and anti-B220 antibodies showed that all the tumor cells were positive for Thy1.2 antigen but negative for B220 (Table 1). Staining the cells with anti-CD4 and anti-CD8 antibodies revealed that 1 tumor was double negative (CD4⁻CD8⁻), 4 tumors were double positive (CD4⁺ CD8⁺), and the other 2 were single positive for CD4 (CD4⁺ CD8⁻) and CD8 (CD4⁻ CD8⁺), respectively (Table 1). To investigate the clonality of the blast cells, DNAs extracted from the tumor tissues were subjected to Southern blot analysis using TCR as a probe. The results showed that all the tumor tissues carried rearrangements in the TCR locus, indicating that the blast cells were clonal in origin (Figure 3C and Table 1).

The rapid proliferation of blast cells in theBCR/ABL^tg/−p53^+/−mice strongly suggested that a somatic mutation or mutations occurred in the residual wild-type p53 gene. To address this issue, PCR-SSCP analysis was performed. This assessment can detect mutations in almost all of the coding sequence (codon 5 to 356) and the splice donor and acceptor sites in introns 2 through 9 of the mouse p53 gene.26 No band shifts were observed in any regions of the p53 gene under 2 different conditions, indicating that the leukemias did not carry point mutations (data not shown). On the other hand, we found a polymorphic site in exon 5 that allowed us to distinguish the PCR product of the wild-type allele from that of the null allele. Sequencing revealed that codon 134 of the wild-type allele was GCG (coding Ala), whereas that of the null allele was GTG (coding Val).

A schematic model of the region encompassing the polymorphic site and the results of PCR-SSCP are shown in Figure4A and Figure 4B, respectively. As shown in the left 2 lanes in Figure 4B, the p53^+/− mouse was a heterozygote of alleles A and B, whereas theBCR/ABL^tg/− mouse was a homozygote of allele B. Because all the tissues examined in theBCR/ABL^tg/−p53^+/−leukemic mice harbored allele A (numbers 1-9 in Figure 4B),BCR/ABL^tg/−p53^+/−mice were considered to be heterozygotes of allele A from the p53^+/− mouse (null) and allele B from theBCR/ABL^tg/− mouse (wild-type). Interestingly, among 9 tumors that developed inBCR/ ABL^tg/−p53^+/−mice (T lanes in Figure 4B), 6 tumors (no. 2 and no. 5-9) contained only allele A (null) and clearly lost allele B (wild-type). This allelic loss was obviously a somatic event, since the normal tissues (N lanes of no. 2, 5, and 8 in Figure 4B) retained allele B. The absence of the p53 protein in the tumor tissues was confirmed by immunoprecipitation/Western blot analysis using anti-p53 antibodies (Figure 4C). These data demonstrated that the residual wild-type p53 allele was frequently and preferentially lost in the tumor tissues in theBCR/ABL^tg/−p53^+/−leukemic mice and that this led to the absence of the p53 protein. The characteristics of the leukemias that developed inBCR/ABL^tg/−p53^+/−mice are summarized in Table 1.

CML provides an appropriate model for a disease in which the gain of an acquired genetic abnormality contributes directly to the disease progression. The p53 is one of the genes whose alterations are frequently detected in the blast crisis of human CML. Therefore, the functional defect of p53 is believed to play a key role in the evolution of the disease, but no direct evidence has been shown.9-11 Skorski et al27 transfected p210^bcr/abl-expressing retroviruses into p53-deficient and wild-type BM cells. They showed that the p210^bcr/abl-positive, p53-deficient BM cells had blast-like morphologic characteristics and an increased colony-forming ability compared with p210^bcr/abl-expressing, wild-type BM cells. In addition, when injected into SCID mice, p210^bcr/abl-positive, p53-deficient cells showed more rapid disease progression than p210^bcr/abl-expressing, wild-type cells. These results demonstrated that the disruption of p53 function conferred a more aggressive leukemogenic potential on p210^bcr/abl-positive hematopoietic cells and induced a more malignant phenotype. However, because the experimental procedures, such as culture of BM cells and retrovirus transfection, were performed in vitro, the synergistic effect of p210^bcr/abl and loss of p53 in vivo could not be fully understood. We addressed this issue by generating p210^bcr/abl-transgenic, p53-heterozygous mice, in which a somatic event in the residual p53 gene directly abrogates p53 function.

The crossmating of p210^bcr/abl-transgenic mice with p53-heterozygous mice resulted in generation of mice with 4 different genotypes:BCR/ABL^−/−p53^+/+,BCR/ABL^−/−p53^+/−,BCR/ ABL^tg/−p53^+/+, andBCR/ABL^tg/−p53^+/−. TheBCR/ABL^−/−p53^+/+mice were reasonably healthy (Figure 1B and 1C). Of 13BCR/ ABL^−/−p53^+/−mice, 1 mouse died of a skin tumor (data not shown). This result is not consistent with those of previous p53^+/− studies, in which malignancies, including T-cell lymphoma, occurred at a higher ratio.28,29 Although the reason for this difference is not clear, one possibility is the genetic background of the mice, as previously suggested.30 The genetic background of the mice in our study was a mixture of C57Bl/6, DBA, and CBA, whereas other p53^+/− studies used mice containing the 129 strain28,29 and a study showed that an increased ratio of 129 strain accelerates disease development.30 BCR/ABL^tg/−p53^+/+ mice had CML-like disease and ALL (Figure 1B and 1C), as expected.21 In contrast to these mice,BCR/ABL^tg/−p53^+/−mice died in a short period (Figure 1B and 1C) and with rapid proliferation of blast cells (Figure 2A and 2B). The leukemic cells were highly malignant, as indicated by massive infiltration into tissues (Figure 2E-2J). They expressed p210^bcr/ablwith an enhanced kinase activity (Figure 3A, 3B). In addition, the residual wild-type p53 allele was frequently and preferentially lost in leukemic tissues (Figure 4B), which led to the absence of p53 expression (Figure 4C). These results demonstrate that the functional loss of p53 essentially contributed to the blastic transformation of the p210^bcr/abl-expressing hematopoietic cells in vivo.

InBCR/ABL^tg/−p53^+/−mice, the blastic transformation was observed in mice with a wide range of ages (Figure 1B and 1C and Table 1). This was possibly because the disease onset depended on when an acquired event, including loss of p53, occurred in the p210^bcr/abl-expressing hematopoietic cells. In mice with early disease onset, the BM showed a predominance of myeloid cells (data not shown), although no abnormalities were observed in the PB (Figure 1B and 1C). On the other hand, in mice with late disease onset, the PB showed excessive proliferation of granulocytes (Figure 2C) and the BM showed marked hyperplasia of myeloid cells (Figure 2D). Therefore, subclinical or clinical signs of CML already existed when acute leukemias developed in the mice. These results indicate that CML preceded the acute phase in these mice, in recapitulation of the clinical course of human CML.

It is interesting that the loss of the wild-type p53 allele was observed in the leukemic tissues at a high incidence but that normal tissues retained the wild-type allele (Figure 4B). Thus, the acquired loss of the wild-type p53 allele was not a random event; rather, there is a certain mechanism by which genetic abnormalities preferentially occurred in the hematopoietic cells. It could be supposed that the expression of p210^bcr/abl might contribute to the process or processes leading to the allelic loss. The p210^bcr/abl might accelerate the mutation rate or promote genetic instability (as reported in p190^bcr/abl transgenic mice31), which would facilitate the loss of the wild-type p53 allele and consequently cause the blastic transformation. Further studies will be required to clarify the molecular mechanism or mechanisms underlying the preferential loss of p53 in the p210^bcr/abl-expressing hematopoietic cells in the transgenic mice.

It should be noted that all the leukemias were clonal T-cell tumors, as demonstrated by flow cytometry and gene-rearrangement analysis (Figure3C and Table 1). These results indicate that the loss of p53 occurred in hematopoietic cells that had already been committed to the T-cell lineage. This idea was further supported by the finding that hematopoietic cells of other lineages, such as myeloid, contained the wild-type p53 allele (data not shown). The reason for the remarkable susceptibility of the T cells to p53 loss remains elusive. One possibility is that gene rearrangements in the TCR loci triggered gene alterations (possibly in cooperation with p210^bcr/abl) and led to loss of the wild-type p53 allele. The finding that p53 loss occurred preferentially in relatively young mice (number 2 and numbers 5-9 in Figure 4C and Table 1) supports this notion, since physiologic TCR rearrangements occur actively in the thymus of young animals. In the 3 mice in which leukemia developed at a relatively old age, no mutations were detected in the p53 gene (numbers 1, 3, and 4 in Figure 4C and Table 1). In these mice, the mechanism that caused blastic transformation remains unknown. We consider it possible that the reduction of p53 protein in p53^+/−cells enhanced the oncogenicity of p210^bcr/abl and caused malignant transformation.32 Alternatively, the decrease in p53 protein might have facilitated activation of oncogenes or inactivation of tumor-suppressor genes, with which p210^bcr/abl exerted its fully oncogenic potential.

Mouse models for clonal expansion of p210^bcr/abl-expressing hematopoietic cells have been demonstrated in BMT experiments.33-36 Serial transplantation of murine CML cells into syngeneic recipient mice frequently caused acute leukemias. The leukemic cells contained the same proviral integration site as the primary CML cells, indicating that they originated from the same clone. Although the molecular mechanisms responsible for the disease progression have not clearly been identified in BMT models, it may be that a secondary genetic event or events, such as loss of p53 as observed in our study, might contribute to the blastic transformation. One difference between our transgenic model and the BMT experiments is the disease phenotype. Our transgenic mice had T-cell leukemias exclusively (Table 1), whereas the recipient mice in the BMT experiments had myeloid and B-cell acute leukemias in addition to T-cell leukemias.33-36 This phenotypic difference might be due to the genetic backgrounds of the mice. The genetic background of our transgenic mice is a mixture of C57Bl/6, DBA, and CBA, whereas the mice used in the BMT experiments were of the BALB/C strain.33-36 An expanded study would provide more information about the phenotypic disparity between our transgenic mice and the mice in the BMT experiments.

There is a feature of our model that is dissimilar to human CML. A T-cell phenotype is rarely observed in the blast crisis of human CML.1 One possible reason is that T-cell lineage is rarely involved in human CML, whereas every type of cell in our transgenic mice has the transgene and the thymic cells might be more susceptible to a somatic event or events causing blastic transformation. An animal model for CML in which loss of p53 could be controlled in spacial and temporal ways would clarify this issue.

In this study, we showed thatBCR/ABL^tg/−p53^+/−mice had marked proliferation of blast cells in a short period, in which the residual wild-type p53 allele was frequently and preferentially lost. These results provide in vivo evidence that an acquired loss of p53 function contributed a proliferative advantage to the p210^bcr/abl-expressing hematopoietic cells and caused blastic transformation. Our findings provide insights into the molecular mechanism of blast crisis of human CML. In addition, our transgenic mice are a useful animal model for examining the biologic effect of a specific gene on the malignant transformation of p210^bcr/abl-expressing hematopoietic cells in vivo.

We thank Yoshikazu Oh-hira for preparing pathological specimens and Tsuyoshi Takahashi, Yoichi Imai, and Koichiro Yuji for technical assistance. We also thank T. W. Mak for providing the mouse TCR probe.