Reversible model of spheroid formation allows for high efficiency of gene delivery ex vivo and accurate gene assessment in vivo.

The native three-dimensional architecture of carcinomas, which governs numerous autocrine-paracrine interactions related to tumor progression, cannot be faithfully recreated in most in vitro models. Even when the three-dimensional architecture is recreated in artificial scaffolds such as soft agar, this approach does not truly recreate the natural microenvironment of the tumor. Multicellular spheroids can reasonably recreate in vitro the natural three-dimensional architecture of carcinomas, but even the most efficient gene delivery vectors will penetrate only the outer layers of these structures and hence only a small fraction of cells will receive the gene of interest. If the multicellular spheroids are disrupted into a single-cell suspension in order to achieve high transfection efficiency, the single-cell production may have so altered the gene expression profile of the spheroid as to bring into question its present relevancy to in vivo tumor progression. Our laboratory has developed a human-SCID (severe combined immunodeficient) mouse model of inflammatory breast cancer, MARY-X, which grows as tight multicellular spheroids in vitro and as lymphovascular emboli in vivo. The spheroids, which express only low levels of surface sialyl-Lewis(x/a) (sLe(x/a)), are able to form compact homotypic cell clumps mediated by an intact, overexpressed E-cadherin/alpha,beta-catenin axis. The spheroids can be fully disrupted by trypsin proteolysis, anti-E-cadherin antibodies, or Ca(2+) depletion. Of these approaches the disruption with depleted Ca(2+), complete after 30 min, is fully reversible by the readdition of Ca(2+) within 6 hr. This time interval allows for a transfection "window" in which successful gene delivery can be achieved before spheroid reformation. Retroviruses (10(6)-10(7) CFU/ml) carrying the gene encoding either green fluorescent protein (GFP), a dominant-negative E-cadherin mutant (H-2K(d)-E-cad), its control (H-2K(d)-E-cad Delta C25), or alpha-1,3-fucosyltransferase III (FucT-III), an enzyme that increases surface sLe(x/a), were used to transfect either intact (wild-type) or disadhered/readhered (reformed) spheroids. There were marked differences in transfection efficiency in the reformed versus wild-type spheroids. Retroviral transfection of GFP resulted in successful delivery of this reporter gene to only the outer layer of cells of the wild-type spheroids, but to all layers of the reformed spheroids. A single retroviral transfection of H-2K(d)-E-cad, H-2K(d)-E-cad Delta C25, or FucT-III produced evidence of their respective gene expression at 72 hr throughout all layers of the reformed spheroids, but only H-2K(d)-E-cad and FucT-III produced progressive disadherence. Both H-2K(d)-E-cad-MARY-X and FucT-III-MARY-X lost their ability to form lymphovascular emboli in SCID mice. This reversible model of spheroid formation has provided us with insight into the pathogenesis of inflammatory breast carcinoma. If more broadly applied, this model could be used to examine the effects of any gene, using any gene delivery system in the three-dimensional context of native tumoral architecture.