RNA interference seemed poised to transform functional genomics and therapeutics with the 2001 publication of a paper by Tom Tuschl and colleagues showing that 21-base-pair (bp) RNA duplexes silence mammalian genes in a sequence-dependent manner.1 Though prior research demonstrated the effectiveness of double-stranded (ds)RNA as a posttranscriptional gene-silencing tool in plants and nematodes, its use in mammals was limited by the fact that dsRNAs larger than 30 bp can activate the interferon response through a PKR kinase-mediated pathway. After Tuschl's research was published, RNAi scientists were optimistic that so-called short interfering (si)RNAs were the answer to this problem.

Data derived from the Science Watch/Hot Papers database and the Web of Science (Thomson Scientific, Philadelphia) show that Hot Papers are cited 50 to 100 times more often than the average paper of the same type and age.

"Expression profiling reveals off-target gene regulation by RNAi," Jackson AL, Nat Biotechnol...



Bryan Williams

Subsequent research, most notably a 2004 paper by Michael Green and colleagues at the University of Massachusetts Medical School, showed that off-target effects can be concentration dependent.5 "We arrived at the conclusion that these nonspecific effects could be minimized if you used more potent siRNAs that allowed you to deplete or silence genes at lower levels," says Green. Jackson, however, maintains that these effects are most likely the result of siRNAs behaving like microRNAs and translationally repressing mRNA rather than mediating its destruction.

Further research uncovered mechanisms behind the interferon response. For example, John Rossi demonstrated that siRNAs produced by bacteriophage T7 RNA polymerase (as in Williams' paper) can trigger a potent interferon response owing to the presence of an initiating 5' triphosphate; removing this results in a loss of interferon induction.6 Likewise, nonstandard initiating dinucleotides on the hairpin vectors used in Iggo's experiment may have contributed to his group's initial results. "They had two A's instead of the normal G, and somehow that elicited the activation that they saw," says Rossi. Iggo has shown that removing the initiating AA eliminates interferon induction for some shRNAs, but not all.7 Ultimately, explaining the effect and what might prevent it has been a battle of small distinctions. "It takes a special set of conditions for the most part to activate the interferon pathway," says Rossi.


Indeed, stimulation of the interferon response is carrier-dependent and cell-specific. Gunther Hartmann and colleagues at the University of Munich, for instance, found that siRNAs containing a specific 9-bp motif induced high levels of type-1 interferon when transfected into plasmacytoid dendritic cells (PDCs) or injected into mice together with a cationic lipid carrier.8 They demonstrated that gene silencing and immunostimulation are two independent functional activities of siRNA, and that TLR7 is the critical receptor for immune recognition of siRNA, rather than PKR or TLR3 as suggested in earlier papers. An important distinction, however, is that Hartmann and colleagues looked at a primary immune cell subset whose function is to detect viral nucleic acids. Hartmann's results suggest that siRNAs with certain sequences could even be used for immunotherapy with an appropriate delivery system.

Ian MacLachlan and colleagues at Protiva Biotherapeutics in Vancouver also recently showed that stimulation of the immune response by liposome-delivered synthetic siRNAs in vivo, and in mouse and human PDCs is sequence-dependent, identifying several potential immunostimulatory motifs.9 MacLachlan's team found that these effects did not occur when they used naked siRNA, confirming a prior observation by Mark Davis and colleagues at the California Institute of Technology.10

Davis explains that his study looked at only a few sequences and that more experiments should be run before using siRNAs therapeutically. "But it at least shows that there are ways that you can have an RNAi effect in an animal without having any of these off-target immunostimulatory effects. They may be there, but they're below detection limits in the animal," he says. Davis also says he has since developed a polycation delivery system that does not show these effects. Jackson predicts, however, that more complications will reveal themselves. "The amount of off-target activity we observe at the transcript level by microarray may be only the tip of the iceberg," she says.



© 2005 Nature Publishing Group

Several cellular systems are set to recognize foreign, potentially dangerous nucleotides. In the cytoplasm double-stranded RNA can bind to PKR or RIG-1, inducing NF-κB translocation to the nucleus and production of inflammatory cytokines as well as TLR-independent IFN-1. Endosome-associated TLR3 recognizes dsRNA inducing IRF translocation. TLR7 and TLR8 recognize single-stranded RNA to induce NF-κB translocation, and TLR7 can signal for IRF translocation. (From M.A. Robbins, J.J. Rossi, Nat Med, 11:250–2, March 2005).

Early on, researchers assumed that a single RNAi knocked down a single gene causing a phenotype. "What we know now is that single siRNA was probably affecting tens if not hundreds of genes," says David Brown of Austin, Texas-based Ambion, an siRNA supplier. One way to overcome this problem is to use multiple, single siRNAs targeting different regions of the gene of interest as a control, with the expectation that each of the siRNAs will have different off-target effects because they have different sequences. "If each of the siRNAs causes the same phenotype ... the effect that you're seeing is specific to the target gene, and not to the single siRNAs that you're using," Brown says. Another potential way to control off-target effects is to mitigate them through chemical modifications to the siRNA backbone at critical locations, according to Jackson.

Choosing the correct sequence or delivery system could help as well. "There may be potential to design siRNAs that avoid the induction of innate immunity, yet are effective at mediating RNAi, and that's what we've shown in our paper," says MacLachlan.

Davis cautions that the RNAi picture is still "muddy" and that concentration, delivery, and sequence all need to be considered in the further development of the technology and its application to therapeutics. "I think it's too early to say that there are any global conclusions yet."

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?