A convenient method for positive selection of retroviral producing cells generating vectors devoid of selectable markers

Abstract

Early retroviral vectors containing both a therapeutic gene and a dominant selectable marker gene, offered some distinct advantages with respect to gene therapy, in that they simplified the generation, isolation, and titration of retroviral producer cell clones, as well as the evaluation and selection of successfully targeted cells. However, a number of problems were engendered by this strategy: the promoter driving the selectable marker gene could interfere with transcription of the therapeutic gene, and immune responses could be induced to cells expressing foreign proteins of selection marker origin. Simplified retroviral vectors, which lack a selection marker gene, were constructed to address these problems, but the inability to use a selection marker has made identification and cloning of virus producing transfected cells a heavy burden. To maintain the benefits of simplified retroviral vectors, while providing a facile means to select packaging cells transfected with retroviral DNA, we cloned the bacterial selection marker gene encoding neomycin phosphotransferase (neo) into the plasmid backbone of the vector, but outside of the provirus, resulting in efficient selection of transfected packaging cells and generation of packaged virus which lacks the neo gene. This novel approach generates greater numbers of high infectious titer producing clones, after selection in G418 media, than does a co-transfection approach, due to integration of higher construct copy numbers per cell. No transmission of the selection marker gene to target cells was observed following retroviral transduction. Thus, our strategy eliminates the adverse consequences of a selection-based method, while diminishing the burden of identification of packaging cells transfected with vectors devoid of selectable markers.

Introduction

Retroviral vectors used in early gene transfer clinical trials often contained two genes, the therapeutic gene and a selectable marker gene. However, the interference between the 5′ viral LTR and the promoter driving the selectable marker gene resulted in low expression levels of the therapeutic gene (Emerman and Temin, 1984). Furthermore, immune responses directed against proteins produced by the selectable marker gene were shown to eliminate retrovirally-transduced cells in vivo (Riddell et al., 1996, Bonini et al., 1997). Simplified retroviral vectors addressed these deficiencies by eliminating the dominant selectable marker gene and changing the long terminal repeat (LTR) promoter to optimize expression (Onodera et al., 1998). Although these simplified retroviral vectors were an improvement in terms of enhancing viral expression and avoiding immunogenicity, they also created a heavy burden in establishing and characterizing viral producing cell clones from unselected cell populations.

Several labor intensive methods have been employed to enhance recovery of viral producing cell clones transfected with simplified vectors: large scale RNA dot blot screening, in which supernatant or viral RNA from each clone is blotted directly onto the membrane (Murdoch et al., 1997, Nelson et al., 1998, Onodera et al., 1997), the combined use of successive rounds of viral transduction of packaging cells and FACS sorting of cells with high expression of the viral envelope (Persons et al., 1997) or the use of cre-mediated recombination, in which the gene of interest replaces the selectable marker (Vanin et al., 1997, Wildner et al., 1998).

Among the possible ways to reduce the burden of selection of packaging cells transfected by simplified retroviral vectors, but avoid problems associated with early retroviral vectors, are two strategies, which we explore in this manuscript. The first method is co-transfection of the vector plasmid with a separate plasmid carrying a selectable marker gene at a high ratio of vector plasmid to selectable marker plasmid (Riviere et al., 1995). The second method, which to our knowledge has not thus far been reported, is to insert a selectable marker gene into the vector plasmid outside of the proviral genome in the vector backbone, physically linking the two, but precluding incorporation of the selectable marker gene in the retroviral transcript. In this study, these two methods were compared by assessment of their transfection efficiencies. We demonstrate that physical linkage is superior to co-transfection in producing greater numbers of high titer expressing packaging cells, via neo selection, but that neo is not transmitted to target cells transduced with retroviral vectors.