Synthetic Dicer-Substrate siRNAs as Triggers of RNA Interference - RNA Interference from Biology to Therapeutics - page 43

modifications that are common in mammalian cells, such as 2¢-OMe RNA, pseudou-
ridine, and others, typically help the RNA molecule avoid detection [ 66 ] . In fact,
2¢-OMe RNA is a competitive inhibitor of TLR7 and can block recognition of
unmodified RNAs by this receptor, even in trans [ 67 ] . Thus, employing the 2 ¢ -OMe
RNA modification, even in only one strand, can block recognition of a siRNA by
TLR7. Other 2¢ -modi fi cations, such as 2 ¢-F and LNA, also help avoid immune
detection. Additional sensors exist in the cytoplasm that can also trigger a response
to foreign RNAs, such as PKR, OAS, RIG-I, and MDA5. Many mammalian cells
that have adapted to growth in cell culture have lost the ability to respond to foreign
nucleic acids; however, the risk of triggering an immune response in vivo, where all
cell types are present, is quite high. Strategies to evade immune detection are essen-
tial to all methods that use synthetic nucleic acids in mammals, including siRNAs.
The reader is referred to some excellent recent reviews on this topic for more
information [ 68, 69 ] .
Like all synthetic RNAs, DsiRNAs can trigger an unwanted immune response
when introduced into mammalian cells. In fact, long RNAs are generally more
potent at stimulating an immune response than are short RNAs. Thus, there was
initially some concern that dsRNAs in the 27-nt length range would present a greater
risk for triggering immune responses than traditional short siRNAs. In support of
this idea, Reynolds and colleagues studied the ability of different dsRNAs to trigger
immune responses in a variety of cell types and found that cell type, the length of
the synthetic RNA, and method of delivery were all contributing factors to immune
stimulation and other off-target effects [ 70, 71 ]. Reynolds studied blunt 27-nt dsRNAs
and found that these structures showed a significantly higher propensity to trigger
immune responses in several cell types than 21-nt siRNAs (with a 2-base 3¢ -over-
hang on both ends). However, it later became evident that end structure is an addi-
tional important feature to consider in triggering immune responses. Marques and
colleagues reported that transfection of 27-nt dsRNAs with 3¢-overhangs on both
ends into T98G glioblastoma cells did not trigger an immune response while use of
blunt 27-nt dsRNAs resulted in a brisk immune response [ 72 ] . The optimized
DsiRNA design (asymmetric 25/27 nt with a single 3¢-overhang on one side and a
blunt end with two 3¢-DNA bases on the other end) was a relatively weak immune
trigger. Collingwood and colleagues further demonstrated that addition of 2¢ -OMe
residues to the optimized asymmetric 25/27-nt DsiRNA design prevented stimula-
tion of interferon-a (IFN- a) secretion from human peripheral blood mononuclear
cells (PBMCs) [ 62 ]. PBMCs are considered a sensitive cell population to employ
when testing the potential for a synthetic nucleic acid to trigger an immune response
as PBMCs are a mixed population of cells which comprise a wide variety of normal
immune-competent white blood cells. Using cationic lipid-mediated transfection,
unmodified 21-nt siRNAs and 25/27-nt DsiRNAs both stimulated significant levels
of IFN-a secretion in human PBMCs, whereas 2¢ -OMe-modi fi ed versions of the
same sequences did not. Thus, the combined use of chemical modification and opti-
mized end structure enabled safe use of 25/27-nt DsiRNAs.