Cellular and Molecular Characterisation of a Novel Murine Erythroleukaemia System

The ELM erythroleukaemia system is unusual in that the erythroleukaemia cells growing in vivo in the spleen cannot be maintained in vitro except in contact with bone marrow-derived stromal cells. They are also unusual in that they retain the ability to undergo erythroid differentiation in response to physiological growth factors, such as Epo and IL-3. Clonal stroma-dependent ELM cells (ELM-D) have been isolated in the presence of stroma. By selecting for growth of ELM-D cells without stroma, a series of stroma-independent clones (ELM-Is) have also been obtained. They are more tumorigenic when injected into mice compared with their stroma-dependent parent. All of the stroma-independent clones, apart from one (i.e. ELM-I/1), are still able to differentiate into mature erythroid cells in response to Epo or IL-3. This particular cell line, that is blocked in differentiation, also displays the highest tumorigenicity in vivo. The ELM system therefore provides a unique opportunity to analyse the molecular nature of stromal control of erythropoiesis, and identify genetic events involved in confering stroma-independent growth, differentiation arrest and increased tumorigenicity. The work presented in this thesis has demonstrated that, upon stroma-withdrawal, the majority of the ELM-D population dies out with characteristics of apoptosis. However, the fact, that a-globin expression in ELM-D cells is largely increased when the stroma is removed, indicates that ELM-D cells undergo erythroid differentiation in the absence of the stroma. Therefore, at least one mechanism whereby the stromal cells maintain the long-term growth of ELM-D cells is to prevent them from further maturation. Gene expression and functional studies of the ELM system have revealed the following: (1) The stroma-independent cells display an elevated expression of an ets oncogene, erg, at the mRNA level in comparison to ELM-D cells. However, only ELM-I/1 cells seem to express detectable level of the Erg protein. (2) The expression of another ets oncogene, fli-1, in the ELM-I/1 cells is significantly activated at both mRNA and protein levels, but is absent in all the other ELM cells; (3) The bcl-2 expression at the mRNA level in the stroma-independent cells is higher than that in ELM-D cells, especially in ELM-I/1 which is the only cell line wherein the Bcl-2 protein is expressed at a detectable level. Functional analysis indicate that over-expression of bcl-2 is not sufficient to induce differentiation arrest or increased tumorigenicity in ELM-I cells. Finally, transient transfection assays have demonstrated that the Fli-1 protein can transactivate a partial bcl-2 promoter in a quail fibroblast cell line. The bcl-2 promoter activity is higher in ELM-I/1 cells, which express high levels of Fli-1 protein, than that in ELM-I/2 cells, which do not express fli-1 at all. This implies that Fli-1 may be responsible for the up-regulation of bcl-2 expression in ELM-I/1 cells. DNase I footprinting analysis has identified six binding sites for the ELM nuclear proteins within the minimum bcl-2 promoter region. However, gel retardation assays and "supershift" analysis using a Fli-1-specific antibody do not reveal the binding of Fli-1 to any of the six nuclear protein binding sites. The tentative conclusion is, therefore, that Fli-1 may transactivate the bcl-2 promoter by an indirect mechanism