Cancer induction by restriction of oncogene expression to the stem cell compartment

Abstract

In human cancers, all cancerous cells carry the oncogenic genetic lesions. However, to elucidate whether cancer is a stem cell-driven tissue, we have developed a strategy to limit oncogene expression to the stem cell compartment in a transgenic mouse setting. Here, we focus on the effects of the BCR-ABLp210 oncogene, associated with chronic myeloid leukaemia (CML) in humans. We show that CML phenotype and biology can be established in mice by restricting BCR-ABLp210 expression to stem cell antigen 1 (Sca1)(+) cells. The course of the disease in Sca1-BCR-ABLp210 mice was not modified on STI571 treatment. However, BCR-ABLp210-induced CML is reversible through the unique elimination of the cancer stem cells (CSCs). Overall, our data show that oncogene expression in Sca1(+) cells is all that is required to fully reprogramme it, giving rise to a full-blown, oncogene-specified tumour with all its mature cellular diversity, and that elimination of the CSCs is enough to eradicate the whole tumour.

Figure 1

Sca1-BCR-ABLp210 and Sca1-TK-IRES-BCR-ABLp210: transgene constructs, expression and survival. (A) Schematic representation of the genomic structure of the mouse Sca1 locus and the Sca1-BCR-ABLp210 and Sca1-TK-IRES-BCR-ABLp210 transgenic vectors used in this study. The HSV-TK/BCR-ABLp210 fusion gene construct is a bicistronic construct consisting of the herpes simplex thymidine kinase (TK) cDNA separated from BCR-ABLp210 by the picornaviral internal ribosome entry site (IRES) sequence. NotI sites used to excise the transgene fragments and EcoRI sites used to examine Southern blots are indicated. (B) Identification of the transgenic mice by Southern blot analysis of tail snip DNA after EcoRI digestion. Human ABL cDNA was used for the detection of the transgene. Sca1-BCR-ABLp210 and the endogenous c-Abl are indicated. Lines IS1A and IS1B are two different Sca1-BCR-ABLp210 transgenic lines. (C) Quantification of BCR-ABLp210 expression in Sca1-BCR-ABLp210 mice and quantification of thymidine kinase (TK) and BCR-ABLp210 expression in Sca1-TK-IRES-BCR-ABLp210 mice by real-time PCR in Sca1⁺Lin⁻ and Sca1⁻Lin⁺ cells. Percentage of TK and BCR-ABLp210 transcripts with reference to β-actin is shown. (D) Quantification of BCR-ABLp210 expression by real-time PCR in Sca1⁺Lin⁻ cells of Sca1-BCR-ABLp210 mice (line IS1A or IS1B) and of double Sca1-BCR-ABLp210 mice (line IS1A × IS1B). (E) Kaplan–Meier survival plots of Sca1-BCR-ABLp210 mice (line IS1A or IS1B), double Sca1-BCR-ABLp210 mice (line IS1A × IS1B) and Sca1-TK-IRES-BCR-ABLp210 mice (line IS9A). The total number of mice analysed in each group is indicated. Statistical analysis was performed using the χ² test, and the corresponding P-values are given in parentheses.

Figure 2a

Sca1-driven BCR-ABLp210 expression induces CML. (A) Cells from PB of Sca1-BCR-ABLp210 mice were analysed by flow cytometry. Representative FACS analysis demonstrating accumulation of mature myeloid cells in PB and spleen, and an increase in the myeloid fraction within BM. PB, peripheral blood; BM, bone marrow. (B) Representative histologic appearance of liver and spleen of diseased Sca1-BCR-ABLp210 mice after haematoxylin–eosin staining. Megakaryocytes in spleen and liver define myeloid metaplasia and are indicated by arrows. (C) Organ infiltration by myeloid cells. Haematoxylin–eosin-stained sections of the spleen (megakaryocytes, myeloid blasts and mature myeloid cells), liver (perivascular infiltration of the liver by blasts and mature myeloid cells), peritoneal lymph node (with myeloid metaplasia) and lung (infiltration of the lung by blasts and mature myeloid cells).

Figure 2b

(D) Phenotypic characteristics of cells from peripheral blood (PB) as determined by flow cytometry. Blast cells but not mature granulocytes are present in PB from two different leukaemic Sca1-BCR-ABLp210 mice. Note that the B-cell leukaemia is characterized by the presence of blast cells co-expressing Mac-1 and B220. (E) Liver haematoxylin–eosin-stained sections showing blast infiltration and histologic appearance of blood smears (Giemsa staining) in Sca1-BCR-ABLp210 mice in blast crisis. Blast cells infiltrate both liver and PB. (F) Representative histologic appearance of liver and spleen tissue sections stained by Masson's trichrome of Sca1-BCR-ABLp210 mice in blast crisis. The presence of fibrosis (green colour) and blast infiltration in the liver and the spleen of Sca1-BCR-ABLp210 mice demonstrate that the blast crisis takes place in the context of a myeloproliferative disease.

Figure 3

Identity of cancer stem cells in Sca1-BCR-ABLp210 mice. To identify genes associated with BCR-ABLp210-induced reprogramming of stem cells, we have compared the gene expression profiles of purified CSC populations versus normal HSCs. Both CSCs and HSCs were isolated as Sca1⁺Lin⁻ cells. (A) Comparison of the mouse and human CSCs at the molecular level. Graphical description of the expression pattern in CSCs (Sca1⁺Lin⁻ cells) from BM (n=10) and PB (n=7) of Sca1-BCR-ABLp210 mice of genes that have been previously published to be significantly regulated in the human CML CD34⁺ fraction (Kronenwett et al, 2005). The genes were similarly regulated in CSCs of Sca1-BCR-ABLp210 mice. We referred the ratios of the CSCs to the control haematopoietic stem cells (Sca1⁺Lin⁻ cells purified from control mice). Green indicates that the genes are downregulated in CSCs versus HSCs and red indicates upregulation. Green asterisks mark genes that are downregulated in human CD34⁺ CML. The remaining genes behave similarly in both human and mouse stem cells. (B) Embryonic surface proteins in CSCs of Scal-BCR-ABLp210 mice. Graphical description of the expression pattern in CSCs (Sca1⁺Lin⁻ cells) from BM (n=12) of Sca1-BCR-ABLp210 mice of previously identified upregulated cell surface markers in undifferentiated mouse embryonic stem cells (Nunomura et al, 2005). We identified genes for which signal intensities were upregulated (threshold±2) in CSCs and in at least seven CSC samples. We referred the ratios of the CSCs to the control haematopoietic stem cells (Sca1⁺Lin⁻ cells purified from control mice). Each gene (identified at right) is represented by a single row of coloured boxes; each experimental mouse is represented by a single column. Data are displayed by a colour code. Red fields indicate higher values than the median, green fields indicate lower values than the median.

Figure 4

STI571 treatment in Sca1-BCR-ABLp210 mice. (A) STI571 treatment does not modify the survival of Sca1-BCR-ABLp210 mice. Mice were randomized to treatment with either STI571 or placebo to study the in vivo efficacy in three independent experiments. The survival curve depicts the percentage of animals alive at the indicated time point. The number of mice in each arm (n) is also shown. (B) Sca1-BCR-ABLp210 mice treated with STI571 do not demonstrate a marked reduction in white blood count and spleen weight. Standard errors are depicted. ^*The data depicted are from mice in (A) and are representative of at least three independent experiments.

Figure 5

GCV treatment in Sca1-TK-IRES-BCR-ABLp210 mice. (A) Survival of Sca1-TK-IRES-BCR-ABLp210 mice following GCV administration. Diseased Sca1-TK-IRES-BCR-ALp210 mice were randomized to treatment with either GCV injected daily or placebo (saline solution treatment) to address the question of whether selective ablation of Sca1⁺ cells can be used as a therapeutic target in cancer. The survival curve depicts the percentage of animals alive at the indicated time point. The number of mice in each arm (n) is also shown and corresponds to three independent experiments. P<0.0001 for treated versus untreated animals. (B) Sca1-TK-IRES-BCR-ABLp210 mice treated with GCV show a marked reduction in white blood count and spleen weight. Standard errors are depicted. ^*The data depicted are from mice in (A) and are representative of at least three independent experiments.

Figure 6

Generation of stem cell-driven cancer in the mouse. (A) In human pathologies and in most animal models of cancer, the oncogenic alteration(s) is(are) present in all the cellular types that compose the tumoral tissue, from the cancer stem cells to the more differentiated types. (B) In our stem cell-driven cancer model, the expression of the oncogenic alteration is restricted to the progenitor compartment but is nevertheless capable of generating a full-blown tumour with all its differentiated cellular components. This model implicitly relies on the fact that the oncogenic presence in the cancer stem cell compartment originates (epi)genetic latent alterations, which are responsible for the posterior appearance of the tumoral phenotype. These alterations are represented here as repressing (red) or activating (green) marks in the chromosome, induced by the oncogene in the CSCs and inherited in the lower compartments where the oncogene is no longer expressed, so that these descendant cells are not normal and constitute a non-clonogenic leukaemic progeny.