The Confusing World of AtuRNAi, Stealth siRNAs and mdRNAs (Part II)

(RNAi IP discussion continued from previous entry)

If long dsRNA had worked exactly in humans as it did in the worm and plants, then you would have expected an immediate flood of publications reporting the same. Long dsRNA for gene silencing, however, were impractical for most vertebrate cell applications due to the induction of non-specific cytokine responses that essentially shuts down most gene expression and therefore does not allow for targeted gene silencing. An exception may be embryonal cells which lack an interferon response and for which long dsRNA was reported to induce specific gene silencing first in zebrafish in 1999 (Wargelius et al. Biochem Biophys Res Commun. 263:156) and then in mice (a mammal) in late 2001 by the Filipowicz group (Basel, Switzerland).

These latter findings, however, were overshadowed earlier in 2001 by a publication from the Tuschl group in Germany, representing the culmination of a body of work he first started as a post-doc in the Bartel/Sharp labs during his time at the MIT, and then as an independent investigator at the Max-Planck Institute in Goettingen. While small RNAs were then known to derive from long dsRNAs, their molecular role in guiding the recognition and destruction of target mRNAs was only hypothesized and their structure mostly unknown. A breakthrough towards this understanding came by establishing a biochemical system in Drosophila (fly) lysates that recapitulated RNAi in the test tube (1999, MIT). One year later, they reported that during this reaction both the long dsRNA as well as the target mRNA is cut at 21-23 nucleotide intervals (MIT, 2000), thereby providing a link between dsRNA processing and mRNA targeting. In early 2001, then at the MPI, Elashir and colleagues in Tuschl’s lab further delineated the relationship between dsRNA processing and target mRNA cleavage in the Drosophila system, including the observation that the mRNA is cut around 10 nucleotides from the 5’ end of a 29 base-pair dsRNA (kind of Dicer-substrate). Importantly, by sequencing the 21-23 nucleotide RNAs by borrowing a cloning technique developed for the discovery of microRNAs around the same time, they found that the small RNAs were clustered consistent with long dsRNA processing into 21-23 base-pair DUPLEX RNAs. Moreover, the small RNAs were found to contain 5’ monophosphates and 3’ hydroxyl groups all consistent with the notion that long dsRNA was processed by an RNase III enzyme into 21-23 base-pair duplexes (reported to be the Dicer enzyme by Hannon in Cold Spring Harbor in the same month).

This led them to test whether small duplex RNAs were crucial functional intermediates between long dsRNA and mRNA cleavage, by synthesizing duplex RNAs and adding them to the Drosophila system. I quote: “Perhaps the 21-nt RNAs are present in double-stranded form in the endonuclease complex, but only one of the strands can be used for target RNA recognition and cleavage”. Indeed, this prediction turned out to be correct and synthetic duplex RNAs could silence mRNAs in this system, and duplexes with 2-3 nucleotide 3’ overhangs, the hallmark of the hypothesized RNase III-type processing, worked best. These data then formed the basis for the Tuschl I patent series to which Alnylam, RXi, and Sirna Therapeutics obtained co-exclusive licenses. My guess is that Alnylam actually would not mind if this patent wasn’t issued after all, since for some obscure reason UMass, unlike the Whitehead Institute, MIT, and MPI decided to grant RXi and Sirna co-exclusive licenses. Equally curious is the fact that while in the January 2001 paper the duplex RNAs were shown to work only in fly lysate, in the Tuschl I series, out of the blue, human cell studies are described. I could well imagine that the ultimately issued Tuschl I patent will be solely focused on the fly data, so that the first human siRNA description would be exclusive to the Tuschl II series (I would encourage you to read my 27 May, 2007 Blog “2007RNAi Therapeutics IP: The Importance of Being Tuschl” on this issue).

Clearly, work in human cells was ongoing at the time in Tuschl’s lab, and they conclude the fly paper in Genes and Development with the ominous statement: “The siRNAs may be effective in mammalian systems, where long dsRNAs can not be used because they activate the dsRNA-dependent protein kinase (PKR) response (Clemens 1997). As such, the siRNA duplexes may represent a new alternative to antisense or ribozyme therapeutics.” The compositions, methods, and uses of synthetic siRNAs in human cells are described in excruciating detail in the Tuschl II patent series, much of which has issued in the EU and US and is exclusively licensed to Alnylam.

A lot of the claims by other companies such as Silence Therapeutics and Invitrogen’s Stealth siRNAs center around the fact that Tuschl II emphasizes the 3’ overhangs of siRNAs, and that blunt-end siRNAs are therefore not subject to Alnylam’s IP estate, but completely ignore the fact that Tuschl, both in his fly and human work indeed tested blunt-end siRNAs, just that they did not perform as well as the overhang siRNAs. I speculate that the reason why Alnylam has not come out and spelt out this fact is because they may think that their equally exclusively licensed Kreutzer-Limmer patent series (use of short dsRNAs for gene silencing in mammals) provides even better coverage for the use of blunt-end siRNAs. Alnylam’s competitors, including Merck, have therefore focused their efforts of fighting Alnylam’s IP dominance on narrowing the scope of Kreutzer-Limmer, particularly in Europe, and have succeeded in doing so last summer to reduce the covered length to 15-21 nucleotides. However, a so called divisional patent application based on the Kreutzer-Limmer patent was granted in Europe in 2005 and has even broader claims than the original patent (15 to 49 base-pair duplexes). It is further ironical that a weakening of Kreutzer-Limmer would only strengthen Tuschl II’s scope. In addition, Tuschl II further covers modifications and conjugations to siRNAs, a claim which Alnylam has cemented by obtaining an exclusive license to the Crooke modification patent estate from ISIS.

While RXi may have marketed their recent StealthTM siRNA license from Invitrogen for therapeutic purposes, in my mind “StealthTM” siRNAs are nothing more than a marketing gimmick disguising the fact that these are 25 base-pair, blunt-end siRNAs with a supposedly magical pattern of base modifications. I would therefore not be surprised if Stealth failed to fulfill the non-obviousness criteria, in addition to the fact that I have not seen any evidence that Stealth, per se, performs any better, if not worse than the classical Tuschl siRNA design. What RXi probably won’t tell you is that Invitrogen has deemed it necessary to gain access to the Kreutzer-Limmer patents through a licensing agreement with Alnylam for the use of Invitrogen’s siRNAs for research applications only.

With regards to Dicer-substrate, licensed by both Nastech and Dicerna, I see practical value in that Dicer-substrates may be beneficial for RNAi delivery purposes in that they provide increased flexibility in covalently conjugating the Dicer-substrate to the delivery carrier, while siRNAs have to be reversibly conjugated, e.g. via disulfide linkages, to achieve the same. However, this does not guarantee the uniqueness of Dicer-substrate since a lot of the duplex length and the conjugation idea is subject to the pre-dating Kreutzer-Limmer and Tuschl patent series. Moreover, in his fly experiments with 29 base-pair duplexes, Tuschl already demonstrated “RNase III-substrate”. Hannon should also have relevance for Dicer substrate in that he was the first to describe Dicer to be the enzyme that mediates dsRNA processing in flies, and likely humans.

Hannon continued his work on the practical implication of dsRNA processing and was one of the first to explicitly describe the use of Dicer-substrates in humans in the form of DNA-directed hairpin expression cassettes driven by a Pol III promoter. While this 2002 paper in Genes and Development was as much inspired by the newly emerging knowledge on microRNA processing as much as by Tuschl’s 2001 findings, “Tuschl and colleagues first showed that short RNA duplexes, designed to mimic the products of the Dicer enzyme, could trigger RNA interference in vitro in Drosophila embryo extracts”), the European group (Brummelkamp et al.) that published in Science on the same subject the same month were mostly inspired by Tuschl: “We report here a new vector system, named pSUPER, which directs the synthesis of small interfering RNAs (siRNAs) in mammalian cells.” In any case, both Alnylam and RXi have gained access to the Hannon patents which touch on both DNA-directed RNAi and Dicer-substrate (synthetic or DNA-directed).

In a confusing turn of events, however, Hannon reported in 2005 that hairpins, in this case synthetic versions though, with longer duplex regions often worked better (= more potent and reliably) than the classical 19 base-pair hairpin design. This could be explained by the observation that the efficiency of RNAi should be enhanced by requiring a Dicer processing step since this is coupled to the RiSC-mediated gene silencing step. The fact that the original 19 base-pair hairpins, first thought to be processed by Dicer, were found to be inferior could be explained by their inefficient processing into siRNAs by some RNases not normally related to RNAi. Essentially the same conclusion was reached by a paper in the same issue of Nature Biotechnology from John Rossi’s group at the City of Hope, this time, however, by employing synthetic 25-30 base-pair duplexes (licensed to Nastech and Dicerna) instead of synthetic hairpins. For a discussion of Dicer-substrate science and IP, please refer to my October 31, 2007 Blog: “A new player in RNAi Therapeutics: Dicerna Pharmaceuticals”.

Besides RNA polymerase III-driven small hairpins, DNA-directed RNAi can also be initiated through more microRNA-like constructs. This was enabled by the elucidation of the microRNA silencing pathway and the trick basically is to design DNA vector constructs that will mimic one of the RNA intermediates during microRNA processing. These methods have the advantage that RNA polymerase II promoters can be employed with potential tissue-specific or other regulation. This should allow for potentially safer DNA-directed RNAi, although RNA polymerase III constructs have extreme knockdown potencies. Brian Cullen’s (Duke) and particularly Narry Kim’s (Seoul, Korea) groups have spear-headed these efforts, but I have not heard from companies yet specializing on the use of such RNAi constructs for therapeutic purposes. It is further likely that the original DNA-directed RNAi patents will be quite important for the commercialization of these later methods.

In addition to employing RNAi triggers that funnel into the RNAi pathway upstream of siRNAs, it is also theoretically possible to make use of at least two more intermediates functioning downstream of siRNA generation: single-stranded guide RNA that recognizes target mRNA within RiSC and a 3-stranded intermediate in which the passenger (=non-targeting) strand is interrupted based on findings from a number of groups, again almost simultaneously about 2 years ago, that the passenger strand was cut prior to guide RNA RiSC loading, analogously to how target mRNAs are cleaved.

The single-strand siRNA method is mostly investigated by ISIS for commercial purposes, probably because it feels that their IP position on single-stranded antisense RNAs would make them the dominant player in single-stranded RNAi. It should be kept in mind, however, that it was again the Tuschl group, known to be close to Alnylam, that first reported on single-stranded RNAi inducers (Martinez et al., 2002) and patents have been filed. Moreover, evidence so far suggests that single-stranded RNAs are only very inefficiently recognized by the endogenous RNAi machinery and it is doubtful that any potential advantages of single-stranded RNAs versus duplex RNAs would ever make up for the inferior potency. Patents covering 3-stranded siRNAs have been filed for by Nastech, although they have not been associated with any of the initial reports on siRNA passenger strand cleavage. It will therefore be interesting to determine the priority dates of the various discoveries, and probably more importantly, data as to the efficiency of these “meroduplex RNAs” (same initials as Nastech’s planned RNAi spin-out mdRNA) compared to other RNAi inducers. Such 3-stranded siRNAs may offer certain advantages with regard to conjugation chemistries and the fact that short RNA strands are cheaper to synthesize than larger ones, but until this is proven it will look just like another thinly disguised patent work-around attempt.

In summary, it is clear that it was Fire and Mello’s discovery on long dsRNAs as RNAi inducers in worms and Tuschl’s extensive body of work leading to the delineation of the classical duplex siRNA that opened up RNAi for therapeutic use. Many of the other developments are directly derived from both of these fundamental discoveries and require appropriate IP licenses. The combination of Tuschl II, Kreutzer-Limmer, and all the other patents it has either exclusive or non-exclusive access to, makes Alnylam the gate-keeper of RNAi Therapeutics. The exact terms of companies wishing to commercialize RNAi Therapeutics will vary depending on their co- or non-exclusive access to some of these fundamental patents, how far removed their exact siRNA derivatives are from the classical siRNA design as well as the ability to prove their utility. While this dominant IP position by Alnylam may make them unpopular and almost look like a bully, one should not forget that concentration and clarity of IP encourages investments particularly in the risky business of drug development and therefore will increase the likelihood of maximizing the therapeutic potential of the technology. It should also be said that Alnylam has been pretty good in de-risking RNAi technology, thereby benefitting the whole field of RNAi Therapeutics, in addition to granting access to RNAi technology through their licensing policy, although the terms will increase the longer you wait. Outside this core RNAi IP, other IP, particularly relating to delivery, but also access to validated targets will prove valuable, albeit much more fragmented.

In my last blog on this RNAi IP series, I would like to briefly discuss individual companies according to technology strength and IP position. Alnylam’s view on this issue will be presented in a special IP-focussed investor presentation on November 28 at the 19th Annual Piper Jaffray Health Care Conference and can be followed live or recorded by webcast on the company’s website.

Erratum: Please note that in my discussion of Dicer-substrate in the October 31, 2007 Blog: “A new player in RNAi Therapeutics: Dicerna Pharmaceuticals”, I mistakenly stated that Hannon’s long hairpin RNAs were DNA-directed, when they actually studied synthetic versions of these hairpins.

The Confusing World of AtuRNAi, Stealth siRNAs and mdRNAs (Part I)

“Next-generation” RNAi is all the rage in the world of the ambitious RNAi start-up. The definition (note that everything is allowed under the guise of the satire): “An RNAi-inducing molecule derived from the classical Tuschl siRNA design, however with a magical pattern of modifications and variations in the exact length of the RNA duplex, sometimes an NA duplex, with overhang or not. As important as the chemistry that may sound impressively inventive to the lay (investing) public is that a proper name is chosen to further accentuate its apparent uniqueness. This is intended to suggest freedom-to-operate with the ultimate aim of attracting investments from people hoping the company will eventually catch up to the market cap of Alnylam (why would you invest in any of their direct rivals otherwise?).”

Considering that it has become commonplace to hear CEOs talk about their RNAi being so unique and advanced that they are now operating in parallel universes, the staid Tuschl siRNA must have really lost its relevance for the development of RNAi Therapeutics. While I think that some select siRNA derivatives given names such as StealthTM RNAi or Dicer-substrate definitely warrant further investigation, as it is yet unclear how well they will perform relative to the simple, but fundamental siRNA design, what I would like to do is to cut through the marketing fog and provide a brief overview of the types of RNAi inducers currently being used at the bench or in the clinic and how I think they relate to each other in terms of IP. In this post I will lay the foundation by giving a summary account of the history of RNAi as a tool, including some of the fundamental patents (and applications), before dissecting some of the Next-generation siRNA designs in a follow-up posting.

Studies on RNAi-related gene silencing really started in the early 90’s in plants with the observation of co-suppression whereby genes that share sequence similarity inhibited each others’ expression. Usually, this was triggered by the inappropriate processing of one of the gene products, typically from an introduced designer gene that is recognized as aberrant and therefore as a threat by the plant RNAi surveillance system. While the mechanism by which this occurs is a scientifically very interesting question, it cannot be used for gene silencing in humans and therefore has little or no relevance to RNAi Therapeutics IP. Parallel work on gene silencing in worms by Fire and Mello, of course, discovered that it was long double-stranded RNA (dsRNA) that was central to inducing RNAi and patents were filed covering dsRNAs longer than 25 base-pairs for gene silencing. This patent can be licensed non-exclusively by almost anybody that wants it, and despite it being based on work in worms and the long dsRNA nature in the stated claims, it is nevertheless considered to be a license that you should add to your IP portfolio anyway, I guess just because it has proven so fundamental to the understanding of RNAi in general and nobody would want to argue that. I also think this highlights the fact that real fundamental scientific insight will be credited by the patent courts even if the exact length of the duplex or modification pattern was not spelt out in the claims letter by letter.

Shortly after Fire and Mello published their research, Kreutzer and Limmer from the University of Bayreuth in Germany reasoned that short dsRNAs may have similar gene silencing effects in mammalian cells. This prediction, as we know, turned out to be true and now forms the basis of the Kreutzer-Limmer patents claiming short dsRNA of around 15-49 base-pairs for the induction of gene silencing in mammalian cells, although the exact length is the subject of patent challenges, including Merck’s opposition in Europe. This early work was considered important enough by Alnylam for them to acquire Ribopharma AG, the company founded on the Kreutzer-Limmer patents. Although I consider Tuschl’s subsequent work to be quite a bit more fundamental to the use of RNAi in mammals, Alnylam understood that it was important to remove any uncertainty as to the dominance of their RNAi IP position given the relative timing and overlapping content of Kreutzer-Limmer and Tuschl.

Around the same time, Hamilton and Baulcombe discovered that small RNAs were generated during plant RNAi. While they were prescient in predicting that these may mediate RNAi, they did not formally prove it and the structure of the siRNA that was detected in those experiments remained unknown. A world away, in Australia, DNA-directed hairpin vectors for reliably inducing RNAi were being described by Waterhouse and colleagues from the CSIRO. Based on the utility and impact of these vectors on plant research, the patents derived from these studies should give the CSIRO a strong position in the agricultural uses of RNAi. In many ways, the commercial development of plant RNAi is more progressed than therapeutic RNAi as traits can now be altered relatively quickly without having to resort to lengthy breeding and selection. I guess the most important question will be how uniform these knockdown phenotypes will be across a field of crops. The CSIRO patents also form partly the basis for Benitec’s claims to the therapeutic uses of DNA-directed hairpin RNAs. The Graham patents form the other pillar of Benitec’s contested patent estate describing the use of DNA cassettes driving the expression of various forms of dsRNAs, although I find these patents to be quite theoretical in nature and wonder whether most of the described non-Pol III expression cassettes would actually work for gene silencing in most mammalian cell types (to be continued…).


Two noteworthy developments last week that I would briefly like to comment on:

1) ISIS released further phase II data for their ApoB-targeting antisense compound mipomersen. The 200mg/week dose reduced by about half the level of bad cholesterol in patients already on stable statin therapy. This looks quite impressive and if no safety issues come up in the larger phase III trials, then this has the potential to become a blockbuster. I’ve been quite critical about mipomersen in the past, particularly due concerns about fatty liver which many scientists in the field would have expected to observe following ApoB knockdown. Safety data for the 200mg dose, based on liver enzyme measurements, however do not indicate this to be a problem. Ultimately, the proof is in the pudding and I would be happy to ultimately have to admit to have been wrong on this issue. ISIS explains the absence of fatty liver due to transcriptional compensatory changes in fat metabolism. Overall, these data augur well for the development of all RNA-targeting platform technologies, including RNAi Therapeutics, as it suggests that minor off-targeting should be well tolerated in many cases.

2) Pfizer announced the acquisition of Coley Pharmaceuticals for almost triple of Coley’s market cap before the offer. Coley Pharmaceuticals is an oligonucleotide therapeutics company that exploits the immunostimulatory properties of oligonucleotides for applications such as boosting vaccines or in the fight against cancer. Actually, I’ve been quite impressed by their OTS presentation in Berlin, particularly their vaccine program. This comes only days after a blog this month where I asked the question when Pfizer will make its big move in RNAi Therapeutics (11 Nov 07 Blog: “When Will Pfizer Finally Make its Big Move in RNAi Therapeutics?”). It is notable that Coley is a Massachussetts company and I would like to think that the proximity to Alnylam will not be an impediment to Pfizer’s new biotech initiative that also appears to more and more focus on oligonucleotide therapeutics. With the new oligo expertise in-house (note that Alnylam does not have another subsidiary to throw into the next deal) and plans to add more staff to a research facility in Cambridge (the headquarters of Alnylam) in addition to a possible biotech incubator near Boston, the plot thickens.