Tracing Dynamics and Clonal Heterogeneity of Cbx7-Induced Leukemic Stem Cells by Cellular Barcoding

摘要

class="head no_bottom_margin" id="sec1title">IntroductionLeukemias are proliferative diseases that arise from HSCs or progenitors that fail to obey the regulatory signals that normally restrict their self-renewal and direct differentiation. Sequential “linear” acquisition of mutations in tumor suppressor genes or oncogenes has long been thought to drive leukemogenesis, as postulated in the original clonal evolution hypothesis (). However, the use of advanced genomic techniques to investigate clonal diversity and evolution now shows that tumors are often organized in a nonlinear, branching hierarchy ().It had been shown that only a subfraction of cells within the leukemic cell population, termed leukemic stem cells (LSCs), possesses the ability to initiate and sustain disease (). Early publications demonstrating heterogeneity within the LSC population utilized the detection of viral integration sites after transduction of acute myeloid leukemia (AML) cells and showed that human LSCs differ in their self-renewal capacities after transplantation in immunodeficient mice (). However, it has recently been reported that xenotransplantation might not always reflect subclonal heterogeneity in patients’ leukemia due to species-specific selective pressures ().Later approaches employed fluorescence in situ hybridization (FISH) to monitor translocations and copy number alterations to investigate genetic heterogeneity within pediatric acute lymphoblastic leukemia (ALL) (). However, the resolution of this FISH-based method to identify heterogeneity and subclonal origin is limited. Deep sequencing genomic DNA or RNA (exome sequencing) of malignant cells from patients provides the highest possible resolution for identification of mutations or other genetic abnormalities within a tumor. Such attempts have recently been made in leukemic patients (). Although potentially very powerful, it is difficult to define (sub)clones based on a large set of genomic data from a pool of genetically diverse cells (), and it requires complex computational approaches and multiple assumptions. For example, the assumption that a certain mutation occurs only once and therefore represents a stable “unique mark” is often made. However, genomes of cancers are generally unstable, and the same mutation may have occurred twice. Another assumption is that the allelic frequency with which a particular mutation occurs depends on the time point of its origin. However, this does not always have to be the case, as primitive LSCs might actually be dormant. In addition, it remains difficult to distinguish which genetic abnormalities are causal to disease progression and which are functionally neutral passenger mutations ().Although previous studies begin to recognize the potential complex genetic architecture of leukemia, unambiguous longitudinal detection of leukemic clones remains difficult to achieve. Ideally, clones should be prospectively defined by unique labeling of cells before tumor initiation, and detection of different clones should be performed at high resolution. A recently described method, which would potentially accomplish this, is the marking of cells by the introduction of a unique, heritable mark that can be detected in its offspring experimentally (). These methods rely on the viral integration of a random “barcode” sequence of fixed length in the genome of target cells. After transplantation of barcoded cells, their offspring can be traced by quantifying the abundance of unique barcodes using deep sequencing.Previously, we have shown that overexpression of the Polycomb PRC1 member Cbx7 in bone marrow cells causes leukemia (). While typically a single oncogene causes one specific tumor type, the epigenetic modifier CBX7 causes a wide spectrum of leukemias, including T-ALL, erythroid, and undifferentiated leukemias. Since only long-term hematopoietic stems cells (LT-HSCs), short-term HSCs (ST-HSCs), and multipotent progenitors (MPPs), but not lineage-restricted progenitors are responsive to Cbx7 overexpression (), the different types of leukemias are not likely to depend on the cell of origin in which Cbx7 is overexpressed. Rather, the phenotypic variation seems to be an inherent virtue of CBX7.In the present paper, we have generated a mouse model in which overexpression of Cbx7 serves as the initial leukemic “hit” and every pre-LSC is uniquely labeled by a barcode. We show how our approach allows for the identification of LSC-derived clones in the transplanted primary and secondary recipients. We prospectively describe clonal dynamics in mice that succumb to leukemia and highlight the complexity of clonal evolution.