The Time for the U(nusual)siRNA Strategy Has Come

As Tekmira and Arrowhead Research will unveil their next RNAi Therapeutics development candidates later this year, an interesting question will be whether these will involve one of their RNAi trigger options that some consider to be unencumbered by fundamental IP related to traditional designs (esp. the Baulcombe and Tuschl II IP).  These decisions could have important strategic consequences for the competitive landscape, from targets and indications to Big Pharma involvement.   

Support for freedom-to-operate claim

One of these designs is the usiRNA from Marina Biotech.  These comprise at least one ‘unlocked’ nucleic acid monomer (UNA) in the double-stranded RNA molecule.  While I have reservations about the scientifically tenuous claim (see here why) that UNAs are not to be grouped with most of the other nucleotide modifications for RNAi use because they lack an intact ribose group, usiRNAs were held to be sufficiently non-obvious and of specific utility that the USPTO issued fairly broad claims in 2012.  Moreover, Marina Biotech once commissioned an external IP lawfirm perform a freedom-to-operate analysis on usiRNA, and (surprise, surprise) came to the conclusion that, indeed, usiRNAs have FTO.

Overcoming target picking limitations

This view seems to be shared also by others in the industry. Notably, Roche RNAi (now part of Arrowhead Research) in 2009 gained access to Marina’s usiRNAs, meroduplex siRNAs, and Dicer-substrate RNAi triggers.  This came as a surprise given that Roche had spent over $300M just two years earlier to gain access to RNAi trigger IP held by Alnylam.  Given that none of the three licensed RNAi trigger forms and related IP poses any FTO threat to traditional Baulcombe-Tuschl designs, the most likely explanation for the move is that it was about allowing the company to escape the target picking limitations under the license from Alnylam.  This included the 31 targets exclusively held by Novartis, some Tekmira exclusive target picks, and some targets pursued by Alnylam that Alnylam exempted from competition.  Whether the last of Alnylam’s Big Pharma licensees, Takeda, might pursue a similar strategy is an interesting question.

Facilitating platform partnerships

When Alnylam and ISIS sued Tekmira for infringing on their RNAi trigger IP by collaborating with Bristol-Myers Squibbs on RNAi delivery, it became a priority for them to have access to or control over non-Alnylam RNAi triggers.  As a consequence, they obtained an exclusive license to Halo-Bio’s multivalent RNAi triggers (more than two strands).  Subsequently, they gained access to Marina’s usiRNAs, including the ability to sublicense.   This now puts them in the position to engage in platform partnerships with Big Pharma companies that do not have access to Alnylam IP.

The same strategy would likely also apply to Arrowhead Research with its various RNAi trigger options that it inherited from Roche, especially if Alnylam provided Roche with only product-specific sublicensing rights, if at all.  As RNAi Therapeutics enjoys a return of pharmaceutical interest, this one-stop-shop option by the two leading delivery companies could be critical to bringing new companies into the space.

And for Alnylam, these developments would not only diminish the royalty it might earn from licensing its IP, they could undermine their own product candidates, including ALN-PCSK9 (hypercholesterolemia) and ALN-AT3 (hemophilia).  Accordingly, the preclinical data strongy suggest that subcutaneous DPCs can do everything that Alnylam’s GalNAcs can do, only much more potently and with less frequent dosing. 


 

Has mdRNA Found a Way around Alnylam’s RNAi Trigger IP Wall?

mdRNA has to be The RNAi Therapeutics corporate success story of 2009. At the brink of bankruptcy at this time of last year after problems with their former nasal delivery business sent the company into a breathtaking tailspin, the new management, critically made up of former Sirna Therapeutics executives, was able to save the company by quickly cashing in on some non-core assets, attracting upfront payments of about $12M in two non-exclusive RNAi Therapeutics deals with RNAi heavy-weights Novartis and Roche, and an opportunistic financing shortly after final approval of a partnered legacy generic (calcitonin-salmon nasal spray for osteoporosis) caused an irrational spike in mdRNA’s share price. Although I perceive mdRNA’s lack of RNAi Therapeutics publications as a clear weakness, also in light of its history of later unsubstantiated scientific claims, these developments are signs that not only is the management apparently well connected, but that the company has been able to establish a quite decent RNAi Therapeutics drug development platform and with IP that cannot be dismissed out of hand. Since the Novartis deal on delivery involved not only IP, but also the transfer of liposomal formulation know-how, but the Roche deal on RNAi triggers appears to be an IP-only deal, I have looked more into mdRNA’s claim of having freedom-to operate with its ‘proprietary’ RNAi trigger designs. Out of fairness, I should add here that a lot of this change had been set in motion with the old management.

It has become a popular game in the RNAi Therapeutics industry to design ways around Alnylam’s dominant RNAi trigger patent portfolio. While corporate strategy is the driving force behind these supposedly proprietary designs, given that intellectual property is not always guided by the spirit of science and since new trigger designs might have unexpected beneficial properties for certain applications, it is prudent to pay attention when Roche is willing to pay mdRNA non-refundable $5.0M for a non-exclusive license to mdRNA’s RNAi trigger IP claims after having just paid more than $300M for much less of Alnylam’s RNAi trigger IP. Of course, one might also view this deal, which was also largely free of downstream obligations, as a move by Roche to cover all their bases- just in case.

At the time of that deal, mdRNA had three RNAi trigger designs in its stable: certain rights to City of Hope’s Dicer substrates, the three-stranded meroduplex siRNA design after which mdRNA takes its name, and unlocked nucleic-acid-modified siRNAs (usiRNAs). Subsequent to that deal it emerged from regulatory filings that, perhaps because Dicerna appears to own most of the Dicer-substrate IP, that mdRNA has dropped Dicer substrates from their portfolio. Furthermore, since Dicerna would probably have been quite happy to make a deal with Roche, it would appear that Dicer-substrates, although part of the deal, was not the main motivation for Roche. This leaves us with meroduplexes and usiRNAs.

If you assembled five PhDs in molecular biology for a weekend, presented them with the main claims of Tuschl I and II, and asked them to come up with RNAi trigger design-arounds, in principle the result may have looked similar to what mdRNA arrived at. That the various designs should also likely to be functional biologically, at least to a certain degree, is a testament to the robustness of the RNAi pathway when it comes to relatively short double-stranded RNAs, a realization that has only sunk in subsequent to the Tuschl publications, but that was already heralded by the findings of Kreutzer and Limmer.

Both Tuschl I and II essentially claim double-stranded RNAs of 21-23 (T-I) or 18-24 (T-II) (contiguous?) NUCLEOTIDES in length. Besides size and overhangs which had been the main theater of the RNAi trigger IP war thus far, one might argue that an RNAi trigger that does contains a number of NON-nucleotides or in which the dsRNA is made up of three, and not two strands, does not literally violate the Tuschl claims.

I have long been puzzled why mdRNA would think that using unlocked nucleic acids (UNA) per se would be the solution, as I considered UNA just another nucleotide-modification alternative in the armamentarium of the siRNA synthetic chemist: UNAs are simply nucleic acids in which the C2-C3 bond has been disrupted (that is apparently also the process by which they are generated). But when the company’s CEO Michael French at the BMO Capital Markets Focus on Healthcare Conference made a point that it is the fact that UNAs supposedly are not nucleotides is indeed what underlies mdRNA’s interest in UNAs. A quick, non-representative survey among scientists that I know, however, was consistent with my initial assumption that UNAs clearly have to be considered ‘nucleotides’ as they look and behave like such. Moreover, there are numerous quotes by the inventor of UNA, Jesper Wengel of Denmark, and by the Danish company RiboTask from which mdRNA has exclusively licensed the UNAs for therapeutic applications that UNAs are modified nucleotides. The question therefore appears to boil down to whether potential partners first and then the patent offices can be convinced of that a ‘nucleotide’ that does not contain an intact ribose or deoxyribose strictly is a nucleotide no more. Although I am skeptical that any one nucleotide modification is sufficient to get around the Tuschl patents, I could imagine that the possibility alone of their literal interpretation could motivate others to come up with similar non-nucleotide siRNAs.

Aside from the IP considerations of usiRNAs, it is premature to judge the scientific value of this modification over others. The fact that ‘unlocking’ a nucleotide is a quite distinct modification should allow it to endow an siRNA with unique properties with respect to recognition by the RNAi machinery, target specificity, the potential to be recognized by innate immune receptors etc. What can be said based on the limited publication record (Bramsen et al., 2009: Large-scale screen....; Kenski et al., 2009: Analysis of acyclic nucleoside modifications...), UNAs like many other modifications are best tolerated in the passenger strand and can reduce off-targeting by the passenger-strand, whereas the use in the guide strand is much more position-dependent. Whether its judicious incorporation can indeed largely abrogate innate immune responses and/or allow for a better differentiation of on-target cleavage over microRNA-like off-targeting as claimed, remains to be seen and especially published (also because mdRNA/Nastech has a track record of making similar claims which tend to silently disappear over time). On the other hand, since Sirna Therapeutics (Merck) is evaluating UNA-modified siRNAs (Kenski et al., 2009), despite its access to Tuschl I, may be interpreted as a sign that there may indeed some unique scientific merit to usiRNAs, and mdRNA may derive value from controlling its use for RNA therapeutic applications.

Moving on to 3-stranded siRNAs, the other pillar of mdRNA’s RNAi trigger strategy, from an IP point one has to probably say that this stands a better chance of surviving Tuschl scrutiny with Tuschl II explicitly claiming double-stranded RNAs consisting of two strands. The double-strandedness of Tuschl I, however, could be more broadly interpreted in that a nick should not make a difference in determining the length of a double-stranded region. In this case, it will be interesting to find out how 3-stranded siRNAs with a gap instead of a nick would perform.

When I first heard of the 3-stranded siRNAs in 2006, I thought that this was based on the observations by a number of labs at the time that the passenger strand of the siRNA is cleaved as part of RiSC activation and that the objective of 3-stranded siRNAs thus was to hi-jack the RNAi machinery downstream of the intact siRNA. However, looking at the history of the relevant patents, it appears that this was not the case, but was merely a coincidence. There is, however, an interesting twist to the story. RiboTask, which happens to also be mdRNA’s partner in usiRNas, filed for almost identical patent protection, but with a priority date that appears to be a couple of months ahead of mdRNA’s. In the case of RiboTask, which calls the 3-stranded siRNAs not meroduplexes but small internally segmented interfering RNAs/LNAs (sisiRNA/sisiLNA), I do not want to be as dismissive about their scientific motivation, as they address a major concern of 3-stranded siRNAs related to their potential instability. Since short segments of double-stranded RNAs should be relatively unstable, they demonstrated (Bramsen et al., 2009: Improved silencing properties using small internally segmented interfering RNAs) that it is possible to restore ‘duplex’ stability by modifying it with locked nucleic acids, the mirror-image of UNAs and curiously invented by the same people. Moreover, they show that while heavy modification of both strands renders most siRNAs inactive (with the exception of 2’F and 2’-O-methyl), converting such a modified siRNA into a 3-stranded siRNA often restores at least some of the knockdown activity (also quite interesting from an RNAi mechanism point of view). The ability to heavily modify an siRNA without losing activity may be particularly useful for applications where the siRNA is exposed to nucleolytic degradation such as for conjugate-siRNA approaches. From a scientific perspective, I should also mention that the 3-stranded siRNA design is a way to abolish passenger strand microRNA-type off-targeting, although this is no more a considerable challenge for RNAi Therapeutics. It should also be of interest to evaluate how nicked, and particularly gapped siRNAs are recognized by the various innate immune receptors. The above demonstrations may also help RiboTask and mdRNA in their way through the patent offices, particularly in the US and in light of the mentioned academic research that might have started before RiboTask’s work, and I am curious whether the usiRNA relationship between RiboTask and mdRNA will be extended also to sisiRNAs/meroduplex technology.

Since IP is subject to interpretation and therefore fraught with some uncertainty, it remains to be seen whether mdRNA can achieve certain freedom-to-operate in the RNAi trigger space. Although usiRNAs are the current focus of the company, keeping the meroduplex as an alternative design option open seems prudent. More broadly, RNAi Therapeutics in general can only gain from investments in evaluating new modifications and designs as unforeseen beneficial properties could emerge from this. Until then, however, it is unlikely that any Big Pharma would want to risk their RNAi Therapeutics future entirely on mdRNA’s RNAi triggers. This also explains why mdRNA sees more partnership potential in its RNAi delivery efforts, the currently more pressing need in RNAi Therapeutics and for which the financials should continue to grow. mdRNA urgently needs such a partnership since although the company had just averted bankruptcy, with cash running out over the next 3 months, the struggle to get out of the bankruptcy quicksand has not stopped. Based on management, while a number of early-stage technology evaluations are ongoing, no significant upfront payments from partnerships can be expected until the middle of next year by which time it hopes to have more non-human primate data to satisfy a more discriminating Big Pharma audience. Given the positive developments of the company over the last year and management’s track record, investor interest should be sufficient to support a PIPE within the next 30-45 days so that the company gets a shot at achieving its partnership goal. If one indeed wanted to make the gamble on mdRNA’s management, from an investment perspective, waiting until more is learnt about the terms of the financing may be prudent, unless valuations continue to fall much further.

Final Day of Keystone RNAi Conference

There were quite a number of RNAi Therapeutics related talks on the last day of the conference. Muthiah Manoharan from Alnylam Pharmaceuticals gave an overview of some of the important chemistries and formulations the company is developing.

The liposomal delivery section of his presentation, Alnylam develops liposomes both for dsRNA as well as single-stranded antagomir delivery, proved a bit hard to follow (or maybe it was just me having difficulties concentrating towards the end of the conference) since some data related to Protiva SNALP technology, then Tekmira developed novel cationic liposome formulations, and then the next moment to lipid-like particles. It is befitting that given the overlap and complementarity in terms of biodistribution and pharmacokinetics of these technologies much of this has now been consolidated into the new Tekmira. If I have it right, the previously shown successful repeat-administration data for sustained gene knockdown in the liver made use of MIT’s lipidoid technology, while the impressive mouse liver cancer data were the result of a Protiva collaboration.

Conjugation of siRNAs to cholesterol is well known for the systemic delivery to the liver and jejunum. Less well known in the context of siRNAs is the PEG-conjugation. Since according to Dr. Manoharan this conjugation facilitates equally efficient delivery to the jejunum as cholesterol does, but largely avoids the liver, PEGylation strategies may lessen the risk for liver toxities in targeted therapies of the jejunum.

With respect to modification technology, in contrast to Merck’s (notably absent from the speaker roster) emphasis on siRNA modification (‘siNAs’), Alnylam follows the natural trend and applies modifications as little as possible. It is interesting that they are still using the classical Tuschl dTdT 3’ overhang structure, albeit substituting a thioate for a phosphate in the last linkage. Their favorite modification, however, appears to be the 2’-fluoro modification, long a workhorse for many RNA technology platforms before RNAi, which has now also proven to be very promising for RNAi Therapeutics. This modification imparts not only stability on the siRNA, but also helps in avoiding immune recognition and favors a molecular conformation that may aid in target recognition. In addition to the associated higher lipophilicity, all of this appears to result in significantly improved knockdown efficiencies.

Next on the list was the session chair of the afternoon RNAi Therapeutics workshop, Frank Bennett from the antisense company ISIS Pharmaceuticals. ISIS is developing single-stranded RNAs for the induction of therapeutic RNAi. This may be due to their IP estate in single-stranded antisense as well as the notion that single-stranded molecules may have cost and delivery advantages. The flexibility of single-stranded molecules plus its amphipathic nature, particularly when extensive phosphortioate backbone modifications are applied, should make it easier for ssRNAs to cross the cell membrane (note, however, that naked antisense delivery may not always be sufficient as their own Regulus venture is looking at systemically formulated, conjugated or liposomal, antisense for the inhibition of microRNAs). Maybe not surprising, ISIS, too, has found 2’fluoro modifications to be “magic” as this has significantly improved single-strand RNAi potency, with the IC50s in a number of cases in the low or even subnanomolar range. But his own comparisons show that since 2’fluoro also improved the potency of dsRNAi, the fold-difference in potency between single-stranded vs dsRNA RNAi has not really changed, with ssRNAi being 10-fold less efficient at best (please correct me if I am wrong here, but numbers probably based on lipoplexed oligo delivery in tissue culture). For example, the most potent ssRNAi shown molecule had an IC50 of 0.43nM, while the corresponding 2’fluoro siRNA came in at 0.01nM.

Part in me believes that ISIS has a sincere interest in developing ssRNAi for their future pipeline, but another part makes me believe that ISIS’ work is supposed to widen their claim onto IP in RNAi. Accordingly, Bennett noted that in was in 1998 that ISIS evaluated the use of single-stranded RNAs to modulate gene regulation, and as it turns out, these RNAs worked via RNAi. In any case, this was a fact-based presentation and not meant as an advertisement of single-strand RNAi or antisense technology and it will be interesting to follow just how efficiently single-stranded RNAs can be designed so that they are recognized by RNAi.

David Lewis (Mirus Bio) then gave us an update on his company’s progress with Dynamic Polyconjugates (DPC), a non-liposomal technology for the systemic delivery of RNAi to the liver. As I had
noted here before , DPCs may be particularly promising in terms of safety, as depending on the attached sugar, Kupffer cells and hepatocytes can be selectively targeted. This was corroborated by the presentation of extensive safety data, a topic that unfortunately was not really addressed in the previous two speakers’ presentations. Knockdown was remarkably sustained, and particles repeat-administrable in mice for over 100 days without loss of silencing activity, at relatively low 1.25mg/kg doses. Moreover, sustained silencing was also demonstrated for non-human primates. It is notable that by targeting ApoB in monkeys, they, like many other groups, also observe the fatty liver phenotype. The small size of DPCs should make the technology also a candidate for more difficult-to-penetrate tissues.

Beverly Davidson (Iowa) summarized her group’s experience with AAV-shRNAs for the treatment of Huntington’s Disease. Like many other laboratories have observed, they also find toxicity following U6-driven shRNA expression. Albeit very potent, it is probably recommended to stay away from the U6 promoter for future development programs, especially when high copy-numbers of expression cassettes may get into individual cells, and there are more than enough alternative potent hairpin expression technologies that do not cause toxicity. According to Davidson, incorporating the RNAi trigger within a microRNA context for example was safe. AAV delivery to the striatum, the anatomical site that matters most in HD, was “phenomenal”, particularly with a viral preparation from their collaborators at Targeted Genetics (Seattle). Before RNAi gene therapy for HD can progress into the clinic though, it will be important to determine whether knocking down both the wild-type and mutant huntingtin alleles is safe or whether allele-specific knockdown is required. Studies that address this issue are now ongoing. Overall, I am quite optimistic about the future of AAV- and lentivirally mediated RNAi gene therapy.

Judy Lieberman (Harvard), together with Phil Sharp a co-organizer of the conference, presented more data on targeting cancer stem cells by RNAi, the underlying rationale being that in order to get rid of a tumour, you really need to get rid of the cancer stem cells from which the tumour mass derives. While previous work centered on antibody-protamine siRNAs targeted to stem cell-specific surface receptors, her latest work makes use of targeted liposomes, which can be potentially more consistently manufactured as well as have the added advantage that many siRNAs can be delivered with just one liposome. Using this to deliver let-7 mimics, a microRNA that is becoming more and more clear to serve as a key differentiation factor, her group showed that breast cancer stem cells dramatically lose their ability to self-renew as well as metastasize to the lung and liver.

Equally interesting was Markus Stoffel’s presentation (ETH, Zurich) on the mechanism of cholesterol-conjugated siRNA uptake. In a study previously published in Nature Biotech and discussed here , Stoffel and his collaborators at Alnylam found that the favorable pharmacokinetics and liver uptake of cholesterol-conjugates is due to their association into lipoprotein particles of the blood, and that it is cholesterol-siRNA bound by LDL that carries the RNAi to the LDL-receptor on the hepatocytes, thereby causing gene silencing in the liver. Moreover, cholesterol-siRNA pre-formulated with LDL in vitro and then administered into mice, proved useful in enhancing the hepatic delivery of RNAi.

Extending this work, his group has now found that cholesterol-conjugated siRNAs can also be consistently formulated with Intralipid , a clinically tested fat emulsion used for nutritional purposes as well as a drug carrier for certain anaesthetics, especially for indications of the heart. Quite excitingly, and in retrospect maybe not that surprising, it turns out that cholesterol-siRNAs are not delivered to the liver (a potential safety advantage when targeting other tissues), but instead go to the lung and even better, the muscle, including heart. At reasonable 10mg/kg, they observe an 80% ApoB knockdown in the lung, but this needs to be repeated now targeting genes more highly expressed in muscle/lung. Developments such as this are all the more important, since the ability to target new tissues opens up RNAi Therapeutics to a whole new spectrum of disease (note that this is an siRNA method and gene therapy methods such as AAV-mediated RNAi may also be useful for the systemic treatment of the heart).

Represented in the poster session was work by Nastech on meroduplexes. Maybe I don’t get it, but the suggested advantages of meroduplexes are so tenuous while creating all the additional challenges in developing tri-partite siRNAs. If Nastech was really making progress on delivery as has been suggested, it would make more sense focusing on those efforts which should be sufficient to create partnering interest and financing at favorable terms. I’m sure a company like Alnylam would be more than happy to then cross-license siRNA and delivery IP.

Finally, a poster by Takanori Yokota (Tokyo) reported on the delivery of alpha-tocopherol conjugated Dicer-substrate siRNAs to the liver. Effective silencing of ApoB in mice was achieved at low mg/kg dosages and was apparently well tolerated, except for the- you hear it once again- ApoB knockdown-specific fatty liver phenotype.

Overall, this year’s Keystone has proven a must-go for those that can only go to only a limited number of scientific meetings, giving them a chance to catch up on the important developments in the field over the past year and getting a glimpse into emerging trends. Next year, the RNAi community will be split between three RNAi-related Keystone meetings, one on the basic mechanisms of RNAi, another one centered on RNAi Therapeutics, and finally one on microRNAs and cancer just illustrating how much this field has grown and diversified.