Friday, May 23, 2008
RNAi Therapeutics: Dicer-substrate Identity Crisis
Since the discussion will be, by necessity, quite technical and probably cannot easily be followed by every reader of this blog, here the take-home messages upfront:
1) Dicer may enhance gene silencing by short, less than 23 base-pair siRNAs without, as expected, actually cleaving them and, in the broad sense, such siRNAs could be therefore regarded Dicer-substrates, too;
2) Long >23bp dsRNAs may function in gene silencing without having been cleaved first, similar to how classical short siRNAs are thought to function; since typically the majority of synthetic Dicer-substrate RNAs recovered from cells following delivery remain unprocessed, the question arises what fraction of the silencing effect in “Dicer-substrate” gene silencing experiments is actually caused by products of endonucleolytic Dicer processing;
3) The 2 nucleotide 3’ overhangs structure is the consequence of Dicer processing and may not be a required feature of RNAi triggers;
4) A large-scale comparative study comparing the effect of length, Dicer endonucleolytic processing, and 3’ overhang on RNAi activity, is warranted;
5) Lawyers are certain to profit where biological boundaries blur.
To understand the fundamental importance of size in RNAi Therapeutics, one has to appreciate that while in 1998 Fire and Mello made the seminal discovery that double-stranded RNA (dsRNAs) was the inducer of RNAi, it was Tuschl and colleagues three years later that discovered that it was 19-23 base-pair dsRNAs with 2 nucleotide 3’ overhangs, the products of Dicer processing of long dsRNAs, that were the more immediate mediators of mRNA targeting during endogenous RNAi in Drosophila. Importantly, these siRNAs allowed RNAi for the first time to be used in mammalian cells since, unlike long dsRNAs which had been used until then in lower eukaryotes, were not efficient triggers of the non-specific and cytotoxic interferon response. And it was only then that the adoption of RNAi as a tool for human gene function analysis and its development as a therapeutic really took of (see page 8 of RNAi IP study).
By adding to cells the structural mimics of long dsRNA processing by Dicer, the use of siRNAs of less than 23 base-pairs was therefore thought to circumvent the need for Dicer processing and that the RNAi trigger would be directly incorporated into RiSC (RiSC is the effector enzyme complex containing the endonucleolyic Argonaute protein at its center that uses the guide single-stranded RNA of the siRNA for seeking out and destroying mRNAs with base pair complementarity).
However, the way most biological pathways work efficiently is by connecting the individual steps, and so it probably didn’t come as a great surprise that more and more data emerged showing that the dsRNA processing and downstream gene silencing stages in RNAi would be both physically and functionally coupled.
The very first dsRNA processing studies by Tuschl himself (2001) provided “evidence that the direction of dsRNA processing determines whether sense or antisense target RNA can be cleaved by the siRNA-protein complex”, meaning that although in isolation two Dicer products may be chemically identical, their activity may differ depending on how they were generated by Dicer, thereby indicating coupling.
In 2003 then, Doi and colleagues demonstrated that when Dicer activity was suppressed, the efficacy of gene silencing by <23 base-pair siRNAs was also suppressed as a consequence, similarly indicating coupling between Dicer and RiSC, but this time separable from actual Dicer cleavage. It therefore challenges the notion that only longer dsRNAs would benefit from biological coupling since only they would function as endonucleolytic substrates of Dicer. This notion has partly been based on results from work in ES cells in which Dicer had been completely eliminated and short siRNAs were still functional. I would argue that while this may well be the case, it in no way excludes a contributory role for Dicer in RiSC loading of the siRNA.
Another interesting study in 2002, again by the Tuschl group, showed that single-stranded RNAs could also induce RNAi, albeit at a much reduced efficacy compared to siRNAs. With the benefit of hindsight, maybe this was not too surprising as well since the double-stranded RNA at one point had to become single-stranded to guide target mRNA cleavage. What I, however, found a particularly interesting observation in that study and in a related patent application was that single-stranded RNAs of various lengths, from 19 to at least 29 nucleotides if not longer, could function as guides in RNA silencing. This for the first time suggested that while classical siRNAs and related microRNAs of similar length and structure may well be the natural inducers of RNA silencing in humans and their processing was functionally coupled to the RiSC effector stage, RiSC may actually also be able to use RNAs of different size and structure.
The somewhat blurred distinction between classical Tuschl siRNAs and longer dsRNAs should also have implications for the interpretation of RNAi-related IP. For example, would a company like Dicerna have to proof that the majority of the observed silencing effect is mediated by siRNAs that had been processed from longer precursor dsRNAs? And does the definition of Dicer substrates include Dicer-mediated RiSC loading in the absence of cleavage? There is also, not surprisingly, some discrepancy in the literature. Whereas some groups report that only longer dsRNAs benefit from the presence of Dicer (e.g. Gregory et al.), other reports (e.g. Doi et al.) suggest that Dicer promotes the activity of short siRNAs as well. Also, the observation that so-called “rxRNAs”, RXi Pharmaceuticals’ quite heavily modified blunt-end 25mers that are not cleaved by Dicer (Keystone poster, 2008), can still function in RNAi, further supports that differences in RNAi activity dependent on the dsRNA length may not necessarily be a consequence of Dicer processing.
Regardless of the practicalities and relative safety of using dsRNAs of various lengths and end structures for RNAi Therapeutics, it would be interesting to conduct a detailed, large-scale comparison of RNAi functionality of the various RNAi triggers, including an investigation of the contribution to silencing activity by Dicer, both in RiSC loading and endonucleolytic processing, through the use of Dicer knockout cells.
Disclaimer: This blog is not intended for distribution to or use by any person or entity who is a citizen or resident of, or located in any locality, state, country or other jurisdiction where such distribution, publication, availability or use would be contrary to law or regulation or which would subject the author or any of his collaborators and contributors to any registration or licensing requirement within such jurisdiction. This blog expresses only my opinions, they may be flawed and are for entertainment purposes only. Opinions expressed are a direct result of information which may or may not be accurate, and I do not assume any responsibility for material errors or to provide updates should circumstances change. Opinions expressed in this blog may have been disseminated before to others. This blog should not be taken as investment, legal or tax advice. The investments referred to herein may not be suitable for you. Investments particularly in the field of RNAi Therapeutics and biotechnology carry a high risk of total loss. You, the reader must make your own investment decisions in consultation with your professional advisors in light of your specific circumstances. I reserve the right to buy, sell, or short any security including those that may or may not be discussed on my blog.