Ebola, Advantage Tekmira

These are critical weeks for Tekmira.  Not only is Tekmira fighting to get back control over its LNP delivery technology, the full-fledged termination of the Ebola biodefense program following the Stop-Work Order would have given the company serious headaches.  As much of the current cash-flow derives from the $140M biodefense contract from the Department of Defense (DoD), the company may have been forced to decide between laying off or furloughing staff until the decision in re Alnylam would provide more funding clarity (November-December 2012).

Luckily, with the DoD deciding to continue Tekmira’s Ebola program, the advantage is again with Tekmira.  My main concern with a negative biodefense decision was that it might have particularly affected the long-term ability of the company to continue to innovate on its LNP technology.  The benefits of the biodefense contract were particularly visible in manufacturing (such as scale-up and lyophilization), thereby allowing the company to build on a key competitive strength.

Of course, it is fair to say that I feel vindicated by the simultaneous decision by the DoD to discontinue Sarepta’s competing Ebola development program: Tekmira’s RNAi Therapeutics approach, now in phase I studies, had the stronger scientific data compared to Sarepta’s antisense approach which had once enjoyed a multi-year head-start (here my 2010 comparison of the two programs).  

Liposomally formulated RNAi Therapeutics also has the advantage of facilitating multi-targeting.  This opens up the prospect for targeting multiple viral, but also host factors, some of which may actually be shared by a number of hemorrhagic fever viruses (HFV). I start getting the impression that setting the stage for the development of such broad-acting HFV therapeutics appears to be an increasingly important aspect of Tekmira’s Ebola program.

I should add that Sarepta will continue to be funded under an analogous $140M contract for Marburg (also a HFV).  Here, the company seemed to have achieved important post-infectionmilestones.  Tekmira does not compete here.  Continuing to fund both an RNAi Therapeutics and an antisense program may also be a strategic risk-mitigating decision by the DoD in case one approach encounters terminal platform issues.

Nevertheless, comments by David Hough from the DoD indicate that the decision was mostly based on the scientific merits:

"Our job is to ensure we're making the most out of every dollar we spend," he noted. "We evaluated each contractor's efforts independently.  The evaluation was conducted using all current data and in accordance with the criteria set forth in the awarded contracts and as stated in the initial solicitation.  This was certainly a tough decision but the final decision was made to move forward on continuing the development of the drug candidate that represents the best value to the Government based on what we know today."

Roche Reveals Antibody-Targeted DPC and SNALP Data

Arrowhead Research has stated that it is about to lift the veil on the RNAi Therapeutics research at Roche through a number of publications.  As you remember, Arrowhead Research made a daring splash a year ago when it acquired much of the Roche RNAi assets for about a cent on the dollar invested in RNAi Therapeutics by Roche.  I will be covering the revelations on this blog as I had once considered Dynamic PolyConjugates one of the more promising systemic RNAi delivery technologies- albeit at least 3-4 years behind Tekmira's SNALP technology in terms of clinical translation and validation.

It was actually Roche (and not Arrowhead/Madison) that fired the opening shot with a study on the use of bispecific antibodies for targeted siRNA delivery (Schneider et al. Molecular Therapy- Nucleic Acids: Targeted siRNA delivery and mRNA knockdown mediated by bispecific digoxigenin-binding antibodies). 

Antibody-targeted siRNA delivery has been a concept that has been around for a while.  The overall industry sentiment on this topic is that while it is still worthwhile pursuit, there have been problems with replicating some of the early data concerning simple antibody-siRNA conjugates and electrostatic complexes.  Stability and cytoplasmic penetration have been the two main issues.

The latter obstacle was also encountered when Roche simply added siRNAs conjugated to digoxigenin (DIG) to a bispecific antibody recognizing both the small molecule DIG (thereby binding the siRNA) and a cell surface protein: the siRNA was specifically delivered to the cell expressing the cell surface protein, but there was no gene silencing, presumably due to lack of endosomal escape into the cytoplasm.

This, of course, is not all that surprising, although, in all fairness, it is not just the old antibody-siRNA literature, but also work from the aptamer-siRNA field and the GalNAc data by Alnylam that suggest that certain cell surface receptor uptake pathways allow for such simple delivery.

To further facilitate gene silencing, the bispecific antibody technology was then applied to DPCs and SNALPs by  linking DIG to the polymer backbone (DPC) and PEG-lipid (SNALP), respectively, with the siRNA either covalently bound to the backbone or enclosed in the aqueous SNALP interior.  The masked cationic polymer in the DPC and the cationic liposomes were thus tasked with overcoming the endosomal release challenge.

Targeted SNALPs for Gene Knockdown in Vascular Endothelia

Indeed, gene silencing could now be observed in tissue culture with both approaches.  Next, the authors tested out the concept in the much more challenging in vivo animal setting.  Data for the targeted SNALPs were reported.   

Mouse seeded with human tumor cells were administered SNALPs (employing Tekmira’s ’57.1’ formulation ratio, KC2 ionizable lipid, and in-line mixing) targeted towards the VEGFR2 receptor that is abundantly expressed on the endothelial cells of the vasculature.  Impressively, such constructs not only maintained the expected endothelial silencing efficiency of SNALPs, but greatly enhanced it: a 40% versus almost 80% gene silencing of the CD31 endothelial marker gene.

Endothelial cells are a good initial target cell population for this approach as these complex, and rather large (>100nm) nanoparticles have easy access.  The approach also does not rely on positive charge for cellular uptake which could be a safety advantage over competing approaches.

The reason why the performance of DPCs in this model was not described is unclear.  It is possible that DPCs simply did not work.  It is also possible that Roche did not want to steal Arrowhead’s thunder and held off on publishing the data.  Pointing in that direction was the fact that the siRNA sequence and, more importantly, modification was not described in the paper.  Not only would this have been crucial for interpreting the innate immune stimulation data (it seemed to be absent), but also an important factor when it comes to uptake that involves the direct exposure of the siRNA to endosomal endonucleases.

Overall, it was a fun paper to read and quite a bit more innovative than many of the RNAi research that has been conducted by Big Pharma bar Merck. Antibody-targeted RNAi delivery is certainly not dead. The cationic lipoplexes by Silence Therapeutics could get competition, although Atuplex enjoys a considerable, multi-year head-start.

Comment on Last Week’s ‘Patent Victory’ by Alnylam

Last week, a US judge ruled that, based on the contracts between Alnylam and Tekmira, Alnylam has standing in enforcing certain exclusively licensed patents.  Tekmira tried to duck this infringement lawsuit by claiming that as the licensor of these patents, it was immune from such enforcement.  The judge disagreed saying that such immunity should have been explicitly stated in their agreement.  In my opinion, this technical decision is not entirely surprising.  All it means is that the case will now be tested in more detail, representing another cash drain on Tekmira.  That financial pressure is Alnylam's main aim in filing the lawsuit is also shown by the company expanding the case into Canada, where Tekmira actually conducts its business.

The ruling does not, however, predict whether Alnylam will prevail on the merits.  This should boil down to the question of whether Alnylam’s right to the patents extend to target validation and not just RNAi therapeutics.  As the Alnylam field is defined as “the treatment, prophylaxis and diagnosis of diseases in

humans using an RNAi Product or miRNA Product", it obviously does not.  

The decision also does not reduce the damage that Tekmira could claim as part of the much more important Trade Secret litigation where Tekmira has charged financial damage arising from Alnylam trying to cut out Tekmira from the financial benefits of SNALP, e.g. when dealing with Big Pharma companies such as Takeda and Novartis.  I do not recall that Tekmira accused Alnylam of making it impossible for Tekmira to close RNAi Therapeutics deals directly with Big Pharma, without Alnylam.  An important distinction.

Alnylam’s Imminent PCSK9 Partnership Deal

A lot of the business development and corporate communications strategy that the 17 VPs at Alnylam come up with these days is, you guessed correctly, directed at building investor confidence so that it may survive the Trade Secret case with Tekmira.  So also the planted interview by John Maraganore a few weeks ago in which he suggested that a PCSK9 deal (ALN-PCS) may be imminent.

Needless to say, without having resolved the important questions about ownership and control over SNALP delivery, chances for a real deal are pretty slim.  Instead, expect a VSP-type deal with minimal upfront, and where the press release will state that development efforts will be focused on the subcutaneous GalNAc conjugates at the expense of (Tekmira’s) SNALP delivery.  

It would, of course, be extremely convenient to announce such a deal just in time for trial, as Alnylam will be able to look the jurors in the eye and state that TTR, PCS, and hemophilia are all now employing a non-Tekmira technology.   A minimal upfront would have the further advantage of depriving cash-strapped Tekmira financial ammunition for the litigation.

Of course, no IND has been filed for any of these yet and, most importantly, clinical validation has not been established leaving a big question mark behind the value (and safety) of GalNAc-conjugates.  I’m wondering what type of messaging the 4-person PR team will come up with to explain, after things have calmed down, that they are still favoring the SNALP-delivered versions.

Note: To put (the estimated) 17 VPs and 4 PR employees in context, Alnylam had 115 employees after the latest lay-offs earlier this year.

Novartis Sells $30M Worth of Alnylam Stock

Last time I’ve checked, Novartis was not desperate for cash.  The cash-flow of Big Pharma is still one of the most impressive features of their businesses, maybe more so than their innovative juices. 

It therefore strikes me as very odd that Alnylam has just provided notice that Novartis last week sold approximately $30M worth of their stock, and has a remaining holding of $80M.  If Novartis wanted to get out of Alnylam, and worse still, RNAi Therapeutics altogether, why not sell the entire stake at once and get it over with?

For Novartis, Delivery is Now or Never

Novartis has RNAi trigger IP rights from Alnylam for 31 targets.  However, the IP, especially the Tuschl II 3’ overhang one which I consider the most valuable, albeit not gate-keeping part of that decimated estate, is ageing rapidly.  Novartis therefore needs access to some clinically viable delivery fast.  Tekmira’s systemic SNALP delivery, of course, is currently industry-leading and, also considering the IP clock (~2020-2021), is an obvious technology for Novartis to utilize.   

Indeed, Tekmira’s Complaint against Alnylam includes a charge that Alnylam allegedly manufactured SNALP LNPs for Novartis, despite the Tekmira/Protiva-Alnylam Agreement that clearly state that Tekmira is the exclusive liposome manufacturer until late-stage clinical development commences.  This would seem an easy charge to verify (note: this is a Count that is independent from the MC3 story). If so, Alnylam’s very survival would also depend on the silence of Novartis (and its other current and former Big Pharma partners).  It is also known that the Head of RNAi Therapeutics at Novartis, David Morrissey, is intimately familiar with Tekmira’s technology from a hepatitis-related collaboration when still at Sirna Therapeutics (~2004-5).

Since neither Alnylam nor Tekmira have a clinically-focused hepatitis program, I suspect that at least HepC has been picked by Novartis exclusively; HepB may be co-exclusive with Arrowhead/Mirus..hence 31 and not 30 targets.

All this leads me to the following speculation (and I emphasize that this is just my speculation, nothing more): Novartis is shifting their RNAi Therapeutics budget to invest $30M or more in delivery technology.  If that involved Tekmira's technology it could also be part of an effort to resolve the hard-fought litigation.  

Novartis’ delivery strategy is one of the industry’s best-kept secrets, but with the August 2012 Marina Biotech oligonucleotide modification deal and an estimated 60-100 employees working on RNAi Therapeutics at that company and having played the game relatively smartly compared to its Big Pharma peers Merck and Roche, which leads me to believe they are not getting out of RNAi Therapeutics (also in light of the recent clinical results), some of that secret may soon be lifted.  

Syngenta Acquires RNAi Speciality Co. Devgen: Is RNAi History Repeating Itself?

An important factor ailing the pharmaceutical industry is that the managements of large companies are not equipped to independently evaluate the merits of innovative technology platforms.  Too far removed from the lowly laboring scientist, laser-focused on meeting next quarter’s numbers and increasingly with non-science backgrounds, this frequently leads to odd deal dynamics driven by herd behavior.  Needless to say, while the technologies may be deserving, much money is wasted not only by overpaying, but also by putting it to work in the wrong places setting the stage for investor backlash and lawsuits.   

We have seen this herd behavior in RNAi Therapeutics in 2006-8 culminating in Merck's $1.1B acqusition of Sirna Therapeutics and Roche's $300M+ non-exclusive platform license from Alnylam.  We now know that these companies grossly overpaid for the RNAi trigger IP and that they would soon panic realizing that they had not paid enough attention to RNAi delivery.

Those vested in the agricultural applications of RNAi gene silencing must similarly be feeling like in the land of milk and honey these days.  In fact, Alnylam’s reputation as a company that aggressively enforces what it believes (?) to be its intellectual property in RNAi triggers, has earned it a juicy $30M upfront from Big Ag Monsanto less than a month ago.  This came after a similar-sized deal for plant RNAi in May involving crop protection juggernaut Syngenta (the marriage of the plant businesses of Zeneca and Novartis) and Belgian plant RNAi specialty company deVGen.  What I did not understand about the Alnylam deal is why Monsanto would let its collaboration with Devgen expire in 2011 and instead decide to go for RNAi trigger IP.  Is Monsanto repeating here Big Pharma’s mistake of undervaluing enablement?

This week’s announcement that Syngenta in fact is now fully buying deVGen, of course, provides a good explanation for the Monsanto-Alnylam deal: Syngenta made deVGen an offer it could not refuse (over half a billion US dollars, representing a 68% premium over its market cap) and Monsanto was left scrambling for alternatives and found in Alnylam a very willing provider of IP ammunition in the fight for dominance in RNAi-based pest control. If history is any guide, one or both companies will likely regret their decision due to lack of enablement and lawsuits will fly in 3-5 years. 

Note: although deVGen is also engaged in seed business, especially rice, it's annual $50M revenue run-rate from this part of the business would unlikely warrant the $500M+ price tag. 

RNAi Therapeutics Poised for another Round of Deal Making

Although the clinical dataflow in RNAi Therapeutics has significantly improved the valuations of a number of RNAi Therapeutics companies and has also generated renewed interest in the technology among the wider pharmaceutical industry down to the VCs, we are still waiting for this important validating deal where a large pharmaceutical company will pony up a sizeable upfront (>$30M) in an RNAi Therapeutics deal.  My expectation is that such an event would spawn a number of further financially meaningful deals.  Being an eternal RNAi Therapeutics optimist, I hope that this time will be slightly different in that Big Pharma will have learned its lessons and technical enablement will feature prominently in this new round of deal making.

Fundamental Baulcombe RNAi Patents Extend Reach

I just got notice of the September issuances of two additional US patents (US 8258285 and US 8263569) belonging to the Baulcombe IP estate.  As previously reported, a first patent (US 8097710) from this series was issued earlier this year and represented a mini-shock to the RNAi Therapeutics IP landscape as it sat smack on the sweet-spot of the prototypical Tuschl-type siRNAs: siRNAs with guide/passenger strands of 20-24 nucleotides in length.  Consequently, Alnylam obtained a non-exclusive license to ‘710 shortly thereafter.

‘569 extends coverage over Dicer-substrate RNAi triggers

The claims of the two newly issued patents extend the coverage of the Baulcombe patent estate in 2 important ways.  Firstly, the ‘569 patent is almost identical to the original ‘710 methods patent.  This time, however, the lengths of the guide/passenger strands can be up to 30 nucleotides in length (20-30 instead of 20-24).  This means that companies working with Dicer-substrates like Dicerna may want to take a license from PBL.  Similarly, the ~25bp dsRNAs previously reported on by RXi and Silence/Intradigm, which curiously did not function as Dicer-substrates, would also fall under this new patent.  The saving grace: like ’710, ‘569 is a methods patent.  Methods patents are often easier to work around.

‘285 is a solid composition-of-matter patent

Having said that, the new ‘285 patent essentially turns the ‘710 20-24nt methods patent into a composition-of-matter one.  There is one important exception though: 20mers have to be unmodified, leaving, de facto (because clinical synthetic RNAi triggers are modified), open important asymmetric designs like the 19/21 and 20/22 designs which have been reported to be even more efficacious in many cases than the classical Tuschl 21/21 design.  Nevertheless, the ‘710 and ‘285 together could pose significant headaches for those trying to find holes with traditional RNAi triggers designs. 

Another interesting question is whether Alnylam will have to seek an additional license to ‘285, as in the press release on the Baulcombe license, only the ‘710 was noted as the subject of the license.  My sense is that ‘285 will be included and that as a result PBL will get a slightly increased participation.

[Update September 17, 2012: in an email, PBL confirmed that the new patents are part of their non-exclusive agreement with Alnylam.]

Classical ddRNAi also impacted?

All 3 patents share claims directed towards DNA-directed RNAi (ddRNAi).  It is therefore possible that they will impact the freedom-to-operate of Benitec which practices short hairpin RNAs from which short RNAs are generated by enzymatic processing in the cell.  Accordingly, an important question will be whether the DNA-directed guide and passenger strands covered by the Baulcombe claims would have to be directly generated by the described vector or can also be provided for in the form of a shRNA-type precursor.  I would guess 'probably', because in the Hamilton et al. work, the small RNAs that were seen and form the basis of the claims were also only indirectly generated. 

In summary, the Baulcombe patents have, quite unexpectedly (because based on plant work), emerged as the strongest RNAi trigger IP estate.  Stronger than Kreutzer-Limmer and stronger than Tuschl I.  In many ways, very deservedly so.  The main limitation is though that they are rapidly ageing.   


Addendum: I reviewed some of the prosecution history of the Baulcombe patents and it seems that for '285 to be granted it had to overcome a 'Crooke' patent (in this case US 6,107,094).  I've always found it a travesty that the Crookes often get cited during RNAi trigger patent prosecutions- although they have no scientific relationship to the biological RNAi process.  It is thus pleasing to see that the Examiner in this case saw the light that a double-stranded RNA that directly inhibits an enzyme (i.e. a PROTEIN) does not represent prior art for a dsRNA that targets an mRNA.  Duh!

ISIS ssRNAi Challenge to Gold Standard RNAi Delivery Comes up Short

The esteemed editors and reviewers of CELL got it wrong this time.  Usually, when a high-impact journal like CELL decides to publish back-to-back papers on a given topic, it believes that they mark a turning point of some sort that will be cited all over.  In this case (Lima et al.; Yu et al.), the turning point would be nanoparticle, or more precisely SNALP-formulated RNAi delivery ‘out’, and unformulated, single-stranded RNAi ‘in’.

For those new to my blog: it is the double-strand feature that is a defining property of the RNAi mechanism.  While a single-stranded intermediate is generated in the process, numerous studies, including indeed the one by Lima et al., show that these are contrived, and consequently about 100-fold less potent ‘inducers’ of RNAi gene silencing.

It is therefore surprising that CELL would publish a confirmation of this.  What is new though is that the in vivo ssRNAi data involved unformulated ssRNAi application, whereas previous in vivo ssRNAi work by e.g. Merck involved LNP-mediated delivery (Haringsma et al., 2012; also covered on this blog here).  However, as detailed below, the efficacy was not impressive. To me, the main point of interest related to chemistry and how this sheds light on the basic RNAi mechanism, which actually made this paper enjoyable to read.  For example, the metabolically stable 5'-(E)-vinylphosphonate modification and the positive effect of 2’F on Ago binding.  Nevertheless, such biochemical detail is not the groundbreaking stuff that lands you a paper in CELL, but more something for the dedicated aficionado.

Does ISIS feel threatened by SNALP delivery?

The complexity of SNALP delivery, by which actually the difficulty of re-engineering SNALP technology without access to Tekmira’s trade secrets and know-how is meant, is held against the technology also in a commercially competitive sense.  If patient outcomes is the main goal, as long as you master complexity, isn’t that a good thing, especially in terms of the all-important length of market exclusivity (note that the main cost of SNALP delivery is still the siRNA ingredient)?

A common criticism of my writings is that I connect all things to Tekmira’s SNALP technology.  But read the Lima et al. paper and see for yourself how ISIS equates formulated RNAi delivery with liposomal delivery (start of the abstract e.g.): 

‘The therapeutic utility of siRNAs is limited by the requirement for complex formulations to deliver them to tissues. If potent single-stranded RNAs could be identified, they would provide a simpler path to pharmacological agents. Here, we describe single-stranded siRNAs (ss-siRNAs) that silence gene expression in animals absent lipid formulation.’  

Or

'However, in their current state, the therapeutic utility of siRNA is limited by the requirement for complex lipid formulations to deliver siRNA to peripheral tissues (Vaishnaw et al., 2010).'

It looks  like 1-billion market cap ISIS feels threatened by $40M market cap Tekmira’s SNALP technology after all and is a very interested participant in the frivolous patent infringement lawsuit against Tekmira (frivolous for the reason alone that Alnylam expressly congratulated Tekmira on the BMS deal).

SNALP requires 1000-10.000-fold less oligonucleotides

The assessment that the ssRNAi work by ISIS does not mark a turning point in systemic RNAi delivery is based on simple math.  1000 to 10.000-fold higher amounts of oligonucleotides were required to achieve equivalent knockdowns in mice: conservative 50microgram/kg/month for SNALP vs 50mg/kg/twice a week for ISIS ssRNAi.    

As with any drug, large doses increase the risk of causing toxicities.  In this case, it is particularly the accumulations of large amounts of phosphorothioated oligonucleotides in the liver and kidney that causes such concern.  Of course, SNALP LNP delivery is not entirely without its safety issues.  For example, in the clinic it still involves the use of transient immune suppression which may e.g. be prohibitive to their use in millions of patients with less severe forms of hypercholesterolemia.

Although assuming for a moment that the amount of required oligonucleotides should be irrelevant as long as it was safe, antisense technologies still suffer from poor cost of goods.  Last week for example, the CEO of another antisense company, Sarepta Therapeutics (formerly known as AVI Biopharma), wrote in an unsettling Open Letter to the Duchenne Muscular Dystrophy community that the company essentially cannot afford the large, almost nutritional amounts of oligonucleotides that are required for attempting a therapeutic splice correction.  Closer to home, instead of acknowledging their current cost of goods, ISIS Pharmaceuticals is only providing estimates for their future oligo manufacturing cost goals.

Extrahepatic tissues, which ones please?

Another claim by ISIS related to their ssRNAi tech was that it would be applicable beyond the liver ('broadly distributed and active in multiple organs'), also following systemic administration.  However, once again, this was directly contradicted by their own data which showed a maximally 35% knockdown (the type of from 100 to 65, not from 100 to 35 mind you) for such a tissue (the kidney) when administering…100mg/kg of oligonucleotides in mice. 

The accompanying Huntington’s Disease paper achieved ssRNAi-mediated knockdown in the brain.  This, however, was observed following non-systemic, intraventricular infusion of large amounts of oligonucleotides in small mice brains.

Alnylam was right in  terminating ssRNAi collaboration

Taken together, the publications explain why Alnylam decided two years ago to terminate their ssRNAi collaboration with ISIS.  In addition to using Tekmira’s intravenously infused SNALP LNPs, Alnylam has been working on GalNAc-siRNA conjugates as a subcutaneously deliverable alternative for gene knockdown in the liver.  Although the gene knockdown achievable with that technology still pales in comparison to SNALP LNP, it is about 10-fold more potent than ISIS’ ssRNAi, which goes to show that despite the disadvantages in cellular delivery of unformulated, rigid dsRNAs, their dramatically increased potency more than compensates for it.

At least Lima and colleagues and I agree on this point (opening statement of the introduction): 'RNA interference (RNAi) is a mechanism by which double-stranded RNA triggers the loss of homologous sequence (Fire et al., 1998).' [Emphasis mine]