DNA was complexed with in vivo-jetPEI in 5% glucose at a N/P ratio of 6. A total amount of 1µg of shRNA per hemisphere was stereotaxically injected into the lateral ventricles of newborn mice brains. For siRNA experiments, a total amount of 1µg of nucleic acid per hemisphere was injected.

RNA-interference-driven loss of function in specific tissues in vivo should permit analysis of gene function in temporally and spatially defined contexts. However, delivery of efficient short hairpin RNA (shRNA) to target tissues in vivo remains problematic. Here, we demonstrate that efficiency of polyethylenimine (PEI)-delivered shRNA depends on the regulatory sequences used, both in vivo and in vitro. When tested in vivo, silencing of a luciferase target gene by shRNA produced from a hybrid construct composed of the CMV enhancer/promoter placed immediately upstream of an H1 promoter (50%) exceeds that obtained with the H1 promoter alone (20%). In contrast, in NIH 3T3 cells, the H1 promoter was more efficient than the hybrid construct (75 versus 60% inhibition of target gene expression, respectively). To test CMV-H1 shRNA efficiency against an endogenous gene in vivo, we used shRNA against thyroid hormone receptor alpha1 (TRalpha1). When vectorized in the mouse brain, the hybrid construct strongly derepressed CyclinD1-luciferase reporter gene expression, CyclinD1 being a negatively regulated thyroid hormone target gene. We conclude that promoter choice affects shRNA efficiency distinctly in different in vitro and in vivo situations and that a hybrid CMV-H1 construct is optimal for shRNA delivery in the mouse brain.