100-fold Improvements in Liposomal siRNA Delivery Potencies in just 2 Years

By applying new cationic lipids to the basic SNALP/LNP formulation developed by Protiva (now Tekmira), liposomal siRNA delivery has now reached IC50 potencies of around 10microgram/kg in mice and non-human primates, a roughly 100-fold improvement over the last 2 years alone. This has profound implications for the types of clinical uses of this leading RNAi systemic delivery approach, particularly by predicting a critical improvement in the therapeutic index and lowering cost of goods and length of administration. Tekmira’s SNALP-ApoB program, for the chronic condition of severe hypercholesterolemia, is a good example where these types of improvements could prove critical for clinical success.

Before I go into the nitty-gritty of what will also include the politics behind liposomal siRNA delivery, a topic that I am sure is of interest to many readers here, let me state unambiguously: these are the types of scientific discoveries that could make a big difference to the clinical trajectory of RNAi Therapeutics and are extremely positive. All this can easily be forgotten when going into scientific and commercial details.

While these developments have been anticipated by comments made in conference calls and presentations at academic conferences by Alnylam and its collaborators from the MIT, Tekmira, and the University of British Columbia, the first scientific paper on this has just been published in the Proceedings of the National Academy of Sciences (Love et al.: Lipid-like materials for low-dose, in vivo gene silencing). This paper is a continuation of the so-called ‘lipidoid’ theme first

published on 2 years ago (Akinc et al.: A combinatorial library of lipid-like materials for delivery of RNAi therapeutics) in which Alnylam and their collaborators at the MIT have been screening new lipids that had been generated in a high-throughput fashion by combinatorial chemistry for liposomal siRNA delivery in vivo.

In the 2008 Nature Biotech paper, a library of alkyl acrylates or acrylamides coupled to amines yielded ‘98N12-5’ as the lead lipid with an IC50 in mice of around 1mg/kg. In that paper, the authors noted that unlike the SNALP formulation by Alnylam-Protiva/Tekmira published in Nature in 2006, the 98N12 system required only 1/3 of the lipid content and that it could be used in vivo without the need for an

additional helper lipid typically employed in SNALP formulations. Certainly, the less lipids, including the number of lipids in the mix, the simpler and better. Moreover, extrusion technologies which are difficult to scale up for clinical utility were used in that study. As we will see, the formulation technology as well as the lipid mix can make a huge difference in what liposomes are actually generated and somewhat takes away from the allure/necessity of high-throughput approaches to lipid discovery such as practiced here.

The new paper, however, is not a direct continuation from the initial screen, but is based on an entirely new library of amino-alcohols generated by epoxide chemistry. The stated motivation for doing so was the comparatively cleaner nature of the chemistry, circumventing the need for laborious purifications for initial tissue culture screens. While the 2008 library consisted of around 1200 compounds, this one was an order of magnitude smaller. Like before, the positively charged lipids were first screened in tissue culture by simple direct complexation with the siRNAs, i.e. not as a liposomal formulation. 12 of the most potent lipids from that screen were then carried forward into mouse studies. Although it soon became apparent that there is only little predictive value of a tissue culture screen for in vivo studies, there were 3 formulations that achieved almost complete Factor VII ablation at 3mg/kg in mice. A dose-response revealed that behind the complete knockout data at relatively high dosages were concealed vastly different IC50s of 0.01 (C12-200), 0.3, and 1mg/kg. It therefore appears that the researchers got a lucky break that they included C12-200 in the 12-compound in vivo screen, a level of experimentation that can arguably be considered to be the real bottleneck for liposomal delivery research.

Since Alnylam is increasingly talking about second-generation liposomal nanoparticles (LNPs), including with cationic lipids, and regularly refers to Tekmira’s SNALPs as first generation in this context [example from press release on PNAS paper.: “ALN-VSP and ALN-TTR both utilize a first generation lipid nanoparticle formulation known as stable nucleic acid-lipid particles (SNALP), developed in collaboration with Tekmira Pharmaceuticals Corp.”], as do the ‘lipidoid’ papers contrast their formulations with the SNALPs, it is worth going into the formulation details here again.

For the initial in vivo screen, again no helper lipid was noted (cationic lipid/'lipidoid' +cholesterol +PEG-lipid only). As to the formulation method, the following can be found in the methods section: “siRNA at a concentration of 10 mg∕mL in 50 mM sodium acetate was added to empty liposomes at a weight ratio of 10:1 total lipids:siRNA and the mixture was incubated at 37 °C for 30 min. Formulations were then dialyzed…” While no mention is made of extrusion, they reference the 2008 Nature biotech paper for further details on that formulation method. If not, I would almost have to come to the conclusion that this description would lead to Silence Therapeutics-type lipoplexes and not liposomally encapsulated siRNAs[update 15Jan10: I have looked up the incubation method, and it does appear to result in encapsulating liposomes with the right pHs and EtOH concentrations]. Be that as it may, the 3-lipid formulation and method differs from that used by Tekmira. In the case of C12-200 this yielded a 140nm particle for the initial in vivo screen that had the IC50 for the liver-expressed Factor VII of ~0.01mg/kg.

When it came to the monkey studies, however, the formulation and physical nature of the C12-200-containing particle was changed dramatically. The 2008 advantage of having only 3 lipids in the mix, was given up in favor of the addition of the now industry-standard helper lipid DSPC. Moreover, for the formulation method itself the rapid mixing-ethanol dilution method was employed with the T-shaped apparatus- both invented by Protiva/Tekmira. If anything, the what is referred to now as non-SNALP ‘lipidoid’ formulations are starting to look more and more like gold-standard SNALPs both chemically and formulation technique-wise. Also not insignificant is the fact that while the crude 3-lipid nanoparticles were 140nm in size, the SNALP formulation that was then used for the non-human primate studies was only 80nm in size (smaller is better here), a huge difference and complicating translational studies. Scale-up needs to be considered from the early stages.

Nevertheless, despite the big differences in the physical properties of the particles, the SNALP formulation containing the C12-200 cationic lipid silenced the monkey TTR gene by 70% at 0.03mg/kg following a single low-volume 15-minute infusion (note: Alnylam has filed for an TTR IND at the end of last year using 'first generation SNALP technology'). It thus appears that a potent C12-200 cytoplasm-accessing ability for hepatocytes can mask variable SNALP shapes. In an attempt to elucidate the mechanism of this, in vitro cell uptake studies were performed, although it is unclear from the paper which exact formulation mix was used for this purpose. Based on these studies, the authors argue for a macropinocytic uptake mechanism distinct from classical endosomal uptake thought to be relevant for other types LNP/SNALP formulations. While the data are consistent with this, I would not classify them as proof that this is indeed the productive uptake mechanism, and a side-by-side comparison with a ‘SNALP’ would have been helpful when arguing for unique uptake mechanisms. This also reminds me of a presentation by Tekmira's CSO Ian MacLachlan on a visit to Stanford where only slightly changed SNALP formulations exhibited vastly different biodistributions within the liver. It therefore appears that while the variability of uptake mechanisms is not a property of ‘lipidoids’ per se, they may be one (of several) factor that can account for the greatly varying knockdown potencies of these formulations.

This paper and similar conference reports on the rational design of lipids (= approach taken by Tekmira and Alnylam's collaborators at the University of British Columbia and AlCana) demonstrate that the technological development of RNAi Therapeutics is still on a steep trajectory which should help reinvigorate interest in the field by the financial markets, especially as the worst has been avoided and the longer-term comes into focus again. The next steps will be to further characterize the pharmacology of these formulations and the application of scalable formulation methods. Both of this and the further development of liposomal delivery for receptor-targeted and (gain-of-function) immune applications as highlighted in the JP Morgan presentation by Alnylam CEO John Maraganore, should be sped up by good relationships between Tekmira and Alnylam, at a time that I feel Alnylam is more and more distancing itself from and down-playing the importance of Tekmira when in fact Alnylam is increasingly adopting methods invented by Protiva/Tekmira. The JP Morgan Healthcare meeting where both companies are present should be a good opportunity to patch up relationships. One potentially mutually beneficial solution may be for Alnylam and Roche to take a majority stake in Tekmira with a simultaneous cash infusion, but let Tekmira operate independently to keep their minds sharp, similar to the Roche-Genentech example.

Note added in proof: January 2010 Nature Biotech biotech paper by Tekmira and Alnylam on significantly improved SNALP formulations through the rational design of lipids (link and press release).