Synthetic Dicer-Substrate siRNAs as Triggers of RNA Interference - RNA Interference from Biology to Therapeutics - page 40

2¢-modifications in common use include 2¢ - fl uoro (2 ¢-F) or locked nucleic acids
(LNAs), which are a bicyclic nucleic acid with a methylene bridge linking the
2¢ - and 4 ¢-positions in the ribose ring.
The 2 ¢-OMe modification can be placed in the sense strand, antisense strand, or
both strands of a siRNA [ 44- 48 ]. Complete modification usually results in an inac-
tive siRNA, and use of alternating (or less) 2¢-OMe groups is commonly employed.
Although not necessary, the 2¢-OMe modification is often employed in conjunction
with other modifications, such as 2¢-F residues. The 2¢-F modification is not natural;
however, it appears to be generally safe to administer to cells or live animals and can
help stabilize siRNAs and improve function [ 49- 52 ]. In particular, use of 2¢ -OMe
purines with 2¢-pyrimidine residues can result in a highly stabilized siRNA with
improved performance in vivo [ 53, 54 ]. The relative potency of siRNAs having this
kind of extensive modification pattern shows sequence dependence and thus may
not work effectively at all sites. LNA modifications have an even greater impact on
structure and potency of siRNAs and thus are generally used sparingly as
modifications. Synthetic oligonucleotides that are heavily LNA modified can show
some hepatic toxicity in mice [ 55 ], although this effect appears to be sequence
dependent and some LNA-modified oligonucleotides are well tolerated [ 56- 58 ] .
The modification strategies discussed above were developed and validated using
21-nt siRNAs. Longer dsRNAs, such as DsiRNAs, appear to naturally show greater
resistance to nuclease degradation than short siRNAs [ 59, 60 ]. This may in part be
due to the higher thermodynamic stability seen for longer duplexes, which may limit
the amount of transient single-stranded character in AU-rich regions that are more
susceptible to attack by endogenous endoribonucleases, such as RNase A [ 61 ] . The
same modification strategies employed in 21-nt siRNAs can generally be directly
applied to DsiRNAs, except that a small internal domain needs to remain unmodified
for Dicer cleavage to occur (the only kind of nuclease attack that is actually desired).
It is possible to synthesize DsiRNA duplexes that show high levels of serum stability
while retaining the ability to be processed by Dicer. A structural map of the different
functional DsiRNA domains is schematically shown in Fig. 2.5 . In this figure, the top
strand is the passenger strand, and sequence to the left of the Dicer cleavage site
comprises the final 21-nt siRNA. This region can generally be modified in ways
similar to other synthetic 21-nt siRNAs, as described above. Sequence to the right of
the Dicer cleavage site can also be modified; note that this short sequence is dis-
carded and is not part of the final product that enters RISC. As shown in Fig. 2.5 , the
favored site to add bulky end modifications to a DsiRNA is at the 3¢-end of the pas-
senger strand (labeled “ligand”); modifications can also be added to the 5¢-end of the
guide strand. Using this approach, a bulky modifying group such as a fluorescent dye
or ligand that may aid delivery (such as cholesterol, cell-penetrating peptides, etc.)
can be attached to the RNA duplex in a way that it is “disposable”; any group con-
nected to this end is cleaved off the DsiRNA and discarded and so does not remain
on the mature siRNA and thus does not enter RISC and cannot affect RISC loading.
Collingwood and colleagues reported a systematic survey of various modification
patterns in DsiRNAs, focusing on the use of the 2¢-OMe and 2¢ -F modi fi cations
[ 62 ]. The initial survey was performed at a site in the human STAT1 gene (corre-