The first step in preparing a plasmid for siRNA experiments is to identify target sequences in the gene of interest that are susceptible to siRNA-induced degradation. We have found that a little more than half of the siRNAs provide at least a 50% reduction in target mRNA levels and approximately 1 out of 4 siRNAs provide a 75-95% reduction. The general process begins with scanning the length of the target gene for potential siRNA target sites. For siRNA expression vectors, the target sites should have 5' terminal AAs because, upon folding the siRNAs hairpins will give rise to UU overhangs. These UU overhangs will therefore be complementary to the AA in the target site. The 21 nucleotide siRNA target sequences are then compared to an appropriate genome database to eliminate sequences with significant homology to other genes. For screening, we typically test four siRNAs per target. We space the siRNAs down the length of the gene sequence to reduce the chances of targeting a region of the mRNA that is either highly structured or bound by regulatory proteins.

Constructing four siRNA expression plasmids for each target can be time-consuming and expensive. In vitro transcription provides a less formidable siRNA screening method. To ensure that siRNAs expressed from plasmids are functionally equivalent to siRNAs prepared by in vitro transcription, we prepared plasmids and siRNAs targeting four different sequences in the cyclophilin and GAPDH genes. These nucleic acids were transfected into HeLa cells. Silencing was evaluated by Northern analysis using probes specific to GAPDH, cyclophilin, and 28S rRNA. The hybridization signal from the various targets was quantitated by phosphorimager. The susceptibility of siRNA target sites to siRNA-mediated gene silencing appears to be comparable for both in vitro prepared siRNAs and RNA Pol III-expressed siRNAs. This is also true for chemically synthesized siRNAs versus RNA Pol III-expressed siRNAs. Therefore, it is not necessary to re-screen genes for which functional siRNAs have already been identified.