Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Biomedical Sciences


Cancer and Developmental Biology

Research Advisor

James R. Downing, M.D.


Suzanne Baker, Ph.D. Dario Campana, M.D., Ph.D. Lawrence Pfeffer, Ph.D. Gerard Zambetti, Ph.D.


Acute lymphoblastic leukemia (ALL) is the commonest pediatric malignancy and comprises several distinct subtypes each with its own unique pathogenesis, clinical behavior, and response to therapy. Chromosomal aberrations are a hallmark of ALL but alone fail to induce leukemia. Pediatric ALLs can be divided into several categories based on the expression of several genetically conserved chromosomal translocations including the t(9,22)[BCR-ABL1], t(1,19)[TCF3-PBX1], t(12,21)[ETV6-RUNX1], MLLrearranged leukemia’s, hyperdiploid and hypodiploid karyotypes, and T-lineage leukemia. Each translocation confers a characteristic transforming phenotype within the cell in which it originates but is alone insufficient to induce overt leukemia. In order to identify oncogenic lesions that cooperate with the aforementioned initiator lesions, we have performed genome-wide analysis of leukemia cells from 242 pediatric ALL patients using high resolution, single-nucleotide polymorphism microarrays. Our analysis revealed deletion, amplification, point mutation, and structural rearrangements in genes encoding principal regulators of B lymphocyte development and differentiation in 40% of B-progenitor ALL cases. The PAX5 gene was the most frequent target of somatic mutation, being altered in 31.7% of cases. The identified PAX5 mutations resulted in reduced levels of PAX5 protein or the generation of hypomorphic alleles. Deletions were also detected in TCF3 (also known as E2A), EBF1, LEF1, IKZF1 (IKAROS), and IKZF3(AIOLOS). These findings suggest that direct disruption of pathways controlling B-cell development and differentiation contributes to B-progenitor ALL pathogenesis. Moreover, these data demonstrate the power of high-resolution, genome-wide approaches to identify new molecular lesions in cancer.

The Philadelphia chromosome, a chromosomal abnormality that encodes BCR-ABL1, is the defining lesion of chronic myelogenous leukemia (CML) and a subset of ALL. To specifically define oncogenic lesions that cooperate with BCR-ABL1 to induce ALL, we subsequently performed genome-wide analysis of diagnostic leukemia samples from 304 individuals with ALL, including 43 BCR-ABL1 B-progenitor ALLs and 23 CMLcases. IKZF1 (encoding the transcription factor Ikaros) was deleted in 83.7% of BCR-ABL1 ALL, but not chronic-phase CML. Deletion of IKZF1 was also identified as an acquired lesion at the time of transformation of CML to ALL (lymphoid blast crisis). The IKZF1 deletions resulted in haploinsufficiency, expression of a dominant-negative Ikaros isoform, or the complete loss of Ikaros expression. Sequencing of IKZF1deletion breakpoints suggested that aberrant RAG-mediated recombination is responsible for the deletions. These findings suggest that genetic lesions resulting in the loss of Ikaros function are an important event in the development of BCR-ABL1 ALL.

In order to assess the contribution of the loss of B-cell developmental regulatory genes with BCR-ABL1 we performed bone marrow transplant assays. Pax5haploinsufficiency was shown to cooperate with BCR-ABL1 during leukemogenesis. Furthermore, as seen in human ALL, both Pax5 and p19Arf haploinsufficiency further cooperate during leukemogenesis. Pathological analysis of the leukemias revealed a B-lymphoid phenotype suggesting that this model results in the development of ALL. Secondary transplant studies confirmed that this was ALL and not a lymphoproliferative disorder. Immunophenotypic analysis confirmed the B-ALL phenotype and further revealed striking differences between Pax5+/+, Pax5+/-, and Pax5-/- leukemias. The leukemias that have lost either one or both Pax5 alleles revealed a more immature immunophenotype that was most pronounced in those with bi-allelic Pax5 loss. The wild-type leukemias were consistent with a Hardy fraction C immunophenotype while the Pax5 null leukemias were akin to Hardy fraction A with no expression of any definitive B-cell surface antigens except B220 and CD43.

The Pax5+/+ and Pax5+/- leukemias were monoclonal while the Arf+/- and the compound heterozygous (Pax5+/- Arf+/-) leukemias were oligoclonal suggesting that the loss of p19Arf confers greater leukemogenic properties to a cell than does Pax5 loss. This is substantiated by the data that Arf heterozygous animals are tumor prone alone while Pax5 heterozygous animals live full normal lives.

Genomic analysis of 50 murine leukemias (15 WT, 25 Pax5+/- and 10 Arf+/-) revealed that some of the same genomic abnormalities found in human ALL also develop in our murine ALL model. This finding suggests that our model is accurately recapitulating the development of human ALL. Furthermore, we have found through extensive gene expression studies that our mouse BCR-ABL1 leukemias share a gene expression profile similar to human BCR-ABL1 leukemias further supporting our murine model of ALL.

The gene expression studies that we have performed have also given us insight into the role that Pax5 haploinsufficiency may be playing during leukemogenesis. Recently D.J Wong et al performed an exhaustive study of embryonic and adult stem cell gene expression profiling. He found that there are modules of genes that are common to either embryonic or adult stem cells. He went further to delineate a group of genes that are commonly expressed in both murine and human embryonic stem cells and call this the core ESC-like module. We performed gene set enrichment analysis (GSEA) using this core ESC-like module and found that this geneset is significantly enriched in our murine leukemias. In addition, we found that this core ESC-like module was significantly enriched in normal Hardy fraction B cells, but not in A, or C-F. We also found, using a principal component analysis method, that Pax5+/+ leukemias are most similar to Hardy fraction C while leukemias that have lost either one or both Pax5 alleles are most similar to Hardy fraction B suggesting that the loss of Pax5 blocks B-cell development and results in cells that are more similar to a stage that shares similar expression of embryonic stem cell genes. Taken together this data suggests that by losing Pax5 the leukemia becomes more stem cell like and may gain advantages that other B-cells do not because they continue down the road of differentiation.