There is little question that Alnylam enjoys considerable control over arguably the most desirable type of synthetic RNAi triggers: 19-21 base-pair dsRNAs with 3’ overhangs- some of this control, of course, being somewhat subject to the outcome of the Tuschl Tussle and the review of the Tuschl II patent in Europe. In any case, this had caused a number of competitors to try and avoid this IP estate by designing around these Tuschl-type siRNAs. One of the more popular workaround designs are (blunt-ended) 25 base-pair dsRNAs, being developed by the likes of RXi, Intradigm (now part of Silence Therapeutics), and Invitrogen (RXi has a deal with Invitrogen, most likely covering aspects of RXi’s 25mers). RXi now presents data (Salomon et al.: Modified dsRNAs that are not processed by Dicer maintain potency and are incorporated into the RISC) on the performance of these 25mer siRNAs that are consistent with the notion that it is possible to find highly potent siRNAs with such structures, and that, interestingly, modified 25bp duplexes may often not be subject to Dicer processing before being loaded into the RISC effector complex.
One concern with the use of 25mers is that, yes, they, like a lot of other structures can harness the remarkably robust RNAi gene silencing pathway, but is its potency sufficient for its therapeutic translation? Because RNAi triggers have to satisfy a few other requirements besides potency, as a platform, it is not sufficient to find one or two highly potent sequences, but maybe 5-10 to choose from. The importance of such choice arises from the need to minimize off-targeting and immunostimulatory potential, related to this the ability to modify a siRNA without compromising too much activity, and finding sequences that are conserved in pre-clinical animal models etc.
Sufficient potency to me means 50% knockdown (aka IC50) values of at least 300pM in standard tissue culture experiments (Alnylam’s siRNA in TTR01 e.g. has a single-digit pM potency). Salomon and colleagues provide examples of 2 siRNAs with IC50s of 50-150pM and add, but without showing the actual data, that they ‘have identified duplexes with potency in the picomolar range for over 10 genes and at multiple sites per gene’. So while I am convinced that highly potent 25mer siRNAs can be discovered, the question about the ease of finding those remains an open one. Support that they may not be that rare comes from the gene-specific 25mer siRNA patents filed by Intradigm/Silence Therapeutics that show silencing results from relatively detailed siRNA tiling assays.
It should be possible to increase the pool of potent 25mers by applying design rules like those by Zamore and controlled by Silence Therapeutics. Salomon and colleagues for example report that some of the modifications in RXi's particular 25mer design (unmodified guide strand, 4 5' and 3' terminal 2'-O-methyls in the passenger) affected the unwinding and loading of the passenger strand which may therefore be subject to the Zamore end-stability claims. Moreover, the value of the 3’ terminal mismatches of the guide strand for enhancing RISC turnover, as covered under the first issued Zamore patent, may be particularly high for long guide strands (25mers, see below) because such extended base-pair complementarity (=thermodynamic stability) could negatively impact RISC detachment from the cleaved message and therefore the efficiency with which it can be recycled for the next round of target cleavage. In that sense, Intradigm may have been quite clever in realizing the synergy of the 3' terminal mismatches particularly for the long duplexes as practiced by the company.
Another interesting aspect of the study, also relevant for IP purposes, is that the 25mers were loaded onto RISC without prior Dicer processing. 25mers may have been expected to be subject to Dicer processing to generate Tuschl-type siRNAs. Although this particular legal theory remains to be tested, this may not be the ideal patent for the 25mer companies because Alnylam may argue that Dicer substrates just serve as pro-drugs to the Tuschl-type siRNAs and therefore would require a license. Moreover, if they were indeed processed by Dicer, they might conflict with Dicerna which specializes on Dicer-substrate siRNAs (Dicerna typically uses 27mers). It is therefore of interest that this publication demonstrates quite convincingly that 25base-pair duplexes, whether modified or not, are not processed by Dicer prior to loading into cleavage-competent RISC. Accordingly, the 25mers may well steer clear from infringing either Alnylam’s Tuschl-type siRNAs or Dicerna's Dicer substrates.
As I stated, Alnylam’s RNAi trigger estate is scientifically the most desirable in a number of regards. Extending the length of the dsRNA for example creates additional challenges as it relates to innate immune stimulation, and reduced delivery efficiency for certain siRNA-conjugate delivery approaches. On the other hand, it is certainly possible to find therapeutic candidates for workaround designs like the 25mers and the judicial application of design rules may address some of the disadvantages in the efficiency of finding potent duplexes.
As Alnylam’s Chinese Wall in RNAi trigger is being tested and faces key challenges later this year, both in the US (Tuschl Tussle) and in Europe (oral hearing on T-II at the EPO following challenges by the likes of Merck and Silence), it is delivery more than ever that will be key to Alnylam in strengthening its position in deal negotiations. In particular, assuming that Big Pharma cannot afford to wait forever before making their moves in RNAi Therapeutics, SNALP delivery should continue to be their key for commanding the next $100M+ licensing terms. I'm not really sure what is holding them up as the present situation must have been predictable for the company and I still feel that a move coordinated with Tekmira and one Big Pharma company could be the solution.
PS: While this paper has only come to my attention this week via my pubmed alerts, some of the data contained therein has been previously presented by RXi at scientific conferences. Moreover, whereas the paper appeared in the June edition of NAR, it first appeared online already in early 2010.
Afterword
When I started this blog, it was initially from the viewpoint of an academic scientist with a keen interest in RNAi drug development. As such I had a strong bias towards patents built on ground-breaking scientific discoveries and somewhat sneered at workaround technologies. As the years have passed, I have however come to the realize that while an academic scientist might be embarrassed to report to peers that a workaround may come close to the real thing, workarounds are a well accepted, even encouraged business practice and Big Pharma will take this into account in their RNAi trigger deal-making. Moreover, because publishing cutting-edge research as quickly as possible is the objective of academic research, in many cases, despite the ‘more worthy’ nature of such science, it has often proven difficult to build strong patents on many of these discoveries (issues such a pre-mature disclosures etc). Biotech companies, on the other hand, can conduct research in a way to maximize its IP value.
