Synthetic Dicer-Substrate siRNAs as Triggers of RNA Interference - RNA Interference from Biology to Therapeutics - page 35

there is no 3¢-overhang for PAZ binding. Further, the PAZ domain preferentially
binds RNA 3¢-overhangs, and the inclusion of DNA bases at the 3¢-blunt end may
further discourage PAZ binding at that end. These ideas are consistent with some of
the structural features and biochemistry of Dicer which were subsequently eluci-
dated [ 13, 14 ] .
Note that an asymmetric 25/27-nt RNA duplex could be designed in two ways,
both of which overlap the desired target site and both of which are predicted to
produce the same final 21-nt siRNA after Dicer cleavage: one where the passenger
strand is 27 nt and one where the guide strand is 27-nt length. Both of these sub-
strate designs were tested at different sites in several genes using the mass spec-
trometry dicing assay, and it was verified that the same 21-nt siRNA was indeed
produced by Dicer processing from these two related but different substrates. For
convenience, the asymmetric 25/27-nt RNA duplex with the 27-nt sequence on the
guide strand was called the right (“R”) form (keeping with the convention that the
sense strand is “top,” the Dicer PAZ domain binds the single 3¢-overhang on the left
side of the duplex, and cleavage proceeds to the right “R”). Conversely, the asym-
metric 25/27-nt RNA duplex with the 27-nt sequence on the passenger strand was
called the left (“L”) form (the Dicer PAZ domain binds the single 3¢ -overhang on
the right side of the duplex, and cleavage proceeds to the left “L”). In spite of the
fact that both forms result in the same 21-nt daughter siRNA species, very different
functional potencies were observed for actual knockdown of a target gene. The “R”
form was almost always more potent than the “L” form. An example of this interest-
ing observation is shown in Fig. 2.2 , which is reprinted from Rose et al . [ 31 ] . A site
in EGFP was selected for study, and the potency of a 21-nt siRNA was compared
with “L” and “R” form asymmetric 25/27-nt RNA duplexes at the same site. The
duplexes were transfected into HEK293 cells that expressed EGFP, and fluorescence
was measured 24 h after transfection. The 21-nt siRNA and the “L” form 25/27-nt
duplex showed similar potency with around a 70% reduction in EGFP signal seen
using 2 nM duplex. In contrast, the “R” form 25/27-nt duplex showed much higher
potency, with almost 80% suppression of EGFP fluorescence signal seen at 200 pM;
EGFP fluorescence signal was nearly undetectable using a 2 nM concentration of
the “R” form duplex.
Using a luciferase-based assay where either a sense target or an antisense target
was cloned into the 3¢-UTR of firefly luciferase, Rose and colleagues demonstrated
that the “L” and “R” forms of the 25/27-nt duplex exhibited differential loading of
the guide vs. passenger strands [ 31 ]. The 27-nt strand with the 3¢ -overhang gener-
ally shows a relative increase in RISC loading compared with the 25-nt strand.
Thus, the “R” form, where the 3¢-overhang is on the guide (antisense) strand, gener-
ally shows higher functional potency for suppressing expression of the target gene
than does the “L” form, which has a loading bias in favor of the passenger strand.
These design features are schematically illustrated in Fig. 2.3 . Interestingly, 3 years
later, a similar effect was reported for 21-nt siRNAs: use of an asymmetric design
with a single 3¢-overhang on the guide strand improved loading of that strand and
increased functional potency [ 32 ]. A number of factors influence which strand of an
RNA duplex loads into RISC and functions as the guide strand and which strand is