First I'd like to understand if there's a reason for not doing it the traditional way -- by using a conditional suicide construct. For example you could us a pro-drug (gancyclovir) or a dimerizer (rapa) to activate the suicide gene. Those systems let you keep your epithelial promoter, and you can package safely, without supplying the drug. That would be a much better trodden path than non-epithelial cell packaging. Any reason against?

I am developing both an inducible (FK506-based dimerization) and constitutive (leucine zipper) system, as this meets the needs of my in vivo experimental approach. Of course packaging the inducible system will not be a problem, but I need to be able to package the constitutive suicide construct as well. I have considered using a small molecule caspase-3 (my suicide gene) inhibitor during packaging in 293T cells, but if another non-epithelial packaging line is a viable option then that seems like a good alternative.