Landmark Chemical Modification Study Shows RNA Editing Ready for the Clinic

At this stage, providing investors and the pharmaceutical industry with a clear line of sight that RNA Editing can be readily translated from concept into therapeutic reality is key to unlocking the next step-up in valuation.

A landmark study in March earlier this year by scientists from Wave Life Sciences (Monian et al, Nature Biotech) on chemically modifying ADAR guide RNA oligos (I will abbreviate them from now on AgRNAs due to missing consensus nomenclature) should go a long way in this regard.  It shows that applying a plethora of standard oligonucleotide stabilization chemistries (e.g. PS, PN backbones, 2’-O-methyl-, 2’-F-ribose) which are critical to enabling delivery and desirable durability do not compromise endogenous ADAR enzyme activity.

In fact, backbone stabilization for example via phosphorothioates, especially when in the SP stereopure conformation can actually greatly increase activity.  In a luciferase model system, editing activity of a fully (stereorandom) PS-modified AgRNA was 10x that of a corresponding AgRNA with an unmodified PO backbone.

Accordingly, when GalNAc-conjugated AgRNAs were tested in non-human primates, ~40% editing rates were observed for at least 2 months.  For this, a loading dose of 5mg/kg per day for 5 days was used.  This is on the higher end of what should be clinically acceptable, but as we know from experience with RNAi, what GalNAc works in non-human primates works even better in humans.

Illustrating the value of further refined chemical optimization of high-value candidates, impressive ~70% mRNA editing efficiencies were seen for a AgRNA against mutant SERPINA1 in primary mouse hepatocytes resulting in a concomitant increase in corrected protein.  SERPINA1 is also the target for Wave’s and possibly the industry’s first clinical RNAEditing program and addresses alpha-1-antitrypsin liver and lung disease.

What piqued my interest was that this AgRNA involved a 8-oxo-deoxyadenosine mismatch base opposite the adenine to be modified and a nearby inosine.  What this means will be addressed in my next blog entry...

So congratulations Wave Life Sciences on this study, but also they will admit that the study still only scratches the surface of what gains in potency will be possible with more detailed structure-activity studies.

Stabilizing RNAi Triggers against Cytoplasmic Degradation Pays Dividends

In describing the preliminary phase IIa results of ARC520 for HBV, Arrowhead Research noted that the duration of gene silencing (2 months and more) was surprisingly extended in Man compared to the preclinical experiences in rodents and non-human primates.  Alnylam hasnoticed the same with its GalNAc-siRNA conjugates, especially the highly modified ESC version. 

The extended gene silencing activities, of course, bode very well for RNAi Therapeutics in general when in the early days (~2002-2003) I was a bit apprehensive when gene silencing in my transfections of cancer cell lines persisted for only 2-3 days (as we now know largely due to their rapid cell division).  To maximize the duration of gene silencing, thereby opening up RNAi Therapeutics to new applications and increasing its competitive profile, it is important to understand the factors underlying it.

Alnylam explained the differences to the preclinical experience because rodent and monkey hepatocytes seem to have a more hostile, degradative cytosol compared to human hepatocytes (hypothesis 1).  In one experiment, only 6% full-length ESC-GalNAc-siRNA remained after a given time in rodent and monkey cytosolic extracts while in human liver cytosol extracts more than 60% persisted.

This, however, was only a correlation and I have considered it equally likely that the difference in gene silencing duration might be a function of more stable RISC complexes in humans (hypothesis 2) or increased stability in the endo-lysosomal compartment (hypothesis 3).  Especially for GalNAc-siRNAs, I would think that the reason that it works in the first place is due to them being able to accumulate in endo-lysosomes from which they only get released in the wake of natural vesicle membrane turnover.  So chemical stability here would be a critical factor since the endo-lysosomal compartment is known to be highly degradative.

DPC and SNALP: two endosomolytic technologies with different durations of gene silencing

While I still consider that endo-lysosomal stability of the naked RNAi trigger is critical for approaches like GalNAc-siRNA conjugates, the new DPC-enabled ARC520 results strongly indicate that another critical factor lies downstream of endo-lysosomes.  This is because in the DPC approach which involves strong endosomolytic activities that should activate soon after endocytic uptake, the risk of the RNAi trigger being degraded in the endo-lysosomes should be low.  Similarly, there should be little contribution to gene silencing from RNAi triggers that get released into the cytoplasm in a delayed fashion.

SNALP is another delivery technology where the RNAi triggers that become active in gene silencing get released into the cytoplasm soon after endocytic uptake.  However, while clinical data supporting 3-4 week dosing frequencies have been obtained with SNALPs (e.g. ALN-TTR02), the silencing does not appear to be as extended as with DPCs.  So given that one marked difference of the payloads used with SNALPs and DPCs is the modest degree of chemical modification historically used with SNALPs, this, too, points towards cytosolic stability of the RNAi trigger being important for the duration of gene silencing.  Parenthetically, it also suggests that Tekmira may want to similarly explore heavily modified RNAi triggers while being mindful not to step on the McSwiggen patent toes of Alnylam.

RISC-optimized ultra-stable single-strand RNAi triggers

In the case of traditional double-stranded RNAi triggers as e.g. used with DPCs, the stabilized RNAi triggers get used up over time as they are recruited into RNAi effector complex RISC.  Part of this process involves their unwinding into single-strand RNAs with the guide strand being retained.   It is known that once used, a 'normal' guide strand (or microRNA) is not recycled into another RISC complex and will likely suffer metabolic destruction once the protein components of RISC have become degraded as part of natural protein turnover.  And even if the guide strand had been stabilized, because a standard single-strand molecule that had relied on being part of a double-strand structure for RISC recognition, old age will eventually catch up here, too. 

What a waste after all this effort of getting the RNAi trigger into the cytoplasm.  So why not take a cue from the single-strand RNAi practitioners who optimize single-strand RNAi triggers also based on being able to be recognized by RISC?  If a corresponding dsRNA contained corresponding recognition elements, then the guide strand could contribute to another round of gene silencing, thus extending and enhancing knockdown.  On the other hand, the lessons learned from stabilized dsRNAi triggers should also benefit the single-strand RNAi approach as increased cytosolic stability should also increase their duration of activity: RISC-optimized ultra-stable single-strand RNAi triggers.

Leveraging New RNAi Trigger Chemistries for Gene Knockdown in Phagocytic (and Other) Cells

Much of the achievement of solid RNAi gene knockdowns in hepatocytes (liver) by non-LNP means (Arrowhead DPCs and Alnylam’s 2^nd gen GalNAc-siRNAs) has involved the use of heavy modifications that render the RNAi triggers highly stable. This is because nucleic acids that are not protected by the delivery chemistry itself would otherwise be subject to rapid degradation in extracellular body fluids.

In the case of Alnylam’s GalNAcs, they can even function in the absence of an explicit endosomal release chemistry.  Moreover, GalNAcs and DPCs have shown more sustained gene knockdowns than is achieved with LNP delivery which historically has relied on minimally modified RNAi triggers. 

All that is required is receptor-mediated uptake into the endosomal-lysosomal pathway which is an immensely degradative environment (esp. the lysosomes).

Considering the extended duration of silencing and degradative environment, it seems as if the RNAi triggers have to be able to survive for long enough in late endosomes/lysosomes so that when they get an opportunity to escape by as yet undefined mechanism(s), they are still there ready for gene silencing action.

Such chemistry progress may also be particularly useful for RNAi gene silencing in phagocytic cells of the immune system.  This is because there have been multiple reports, especially concerning the use of LNPs (e.g. Novobrantseva et al.)  where some, but not the very robust, hepatocyte-type of gene knockdown have been obtained.  It is the ability to confidently achieve robust knockdowns that opens the gate to a flood of therapeutic applications, and this is why pushing borderline-technologies over the edge is so tremendously valuable.

Uptake into phacocytes is usually not the problem as these have evolved to scavenge for foreign particles and macromolecules.  In fact, phagocytic uptake is often a nuisance in RNAi delivery both because it may cause off-target toxicity and because it can make pharmacology less predictable.

In the case of untargeted nanoparticles such as LNPs, phagocytosis is the likely uptake mechanism.  Similar to endosomal uptake, phagosomal uptake involves the fusion with lysosomal compartments meaning that phagosomal contents are also exposed to a highly hostile environment.  Since the normal endosomal escape chemistries and mechanisms do not appear to be very effective in phagosomes, one strategy besides of possibly tailoring existing mechanisms to the phagosomal environment (e.g. lipid pKas) is to simply use the same ultra-stable RNAi trigger chemistries that are showing promise in the liver.

It is also possible that ligand-targeted conjugate approaches will be useful here, whether they enter the cells via phagocytic mechanisms or not.  In fact, Arrowhead Research (then Mirus Bio) in their seminal publication on DPCs (Rozema et al.) have demonstrated efficient uptake of DPC-conjugates into liver phagocytes (Kupffer cells) by using mannose as the targeting ligand (they haven’t tested and/or shown the corresponding RNAi knockdown though and this may relate to their using much less nucleic acid modification back in 2007). 

Since the mannose receptor is expressed on phagocytes throughout the body and not just in the liver, more extended circulation times promises the application of this conjugate-targeting strategy more generally.

Similar principles may apply to the targeting of other ‘frontier tissues’ for RNAi delivery.  However, given the degradative and differing nature of phagocytosis which means that other release mechanisms are not readily applied to this process, RNAi in phagocytes should particularly benefit from the new RNAi trigger chemistry developments.  I look forward to seeing the results.

Journal Club: Structural Basis for Double-Stranded RNA Recognition by TLR3 and Activation

The recent Nature paper by Ambati and colleagues raised some concerns about whether non-specific inflammatory responses due to activation of TLR3 by double-stranded RNAs, the inducers of RNAi, would represent a significant, if not insurmountable obstacle towards the development of RNAi Therapeutics. While the short answer was no, since most in the field had already recognized the need to screen against the capacity of RNAi triggers to induce immune responses, determining the rules activating TLR3 and similar molecular patterns should facilitate the more efficient design of safe and potent RNAi triggers. Recent papers on TLR3 structural biology should do just that.

In an April edition in the journal Science, Liu and colleagues from the NIH report on the structure of mouse TLR3 with its double-stranded RNA substrate (note that mouse and human TLR3 are very similar). The structure suggests that a 40-50 base-pair double-strand RNA is optimal for binding by TLR3 thereby inducing TLR3 dimerization and downstream signaling. The fact that one of the dsRNA binding patches contains a number of pH-sensitive histidines further suggest why TLR3 signaling is most robust following uptake into endosome which provides for an acidic compartment. Overall, the structure supports previous observations that dsRNAs longer than what is typically used for siRNAs are better inducers of TLR3; however it does not explain well how smaller siRNAs may also induce such signaling.

An explanation for this is provided by structure-based mutagenesis studies by Pirher and colleagues from the University of Ljublijana (Slovenia). In this paper, they identify two dsRNA-binding patches within a TLR3 monomer and also find and explain why B-type helices, as typically found in DNA, only bind to one of the two binding patches and therefore fail to induce, and even competitively inhibit TLR3. This is in contrast to dsRNAs that typically assume A-type helical formation and bind both patches. With this the authors come up with a model to explain how TLR3 dimers bind to shorter double-stranded RNAs by assuming alternative conformations with ddimerization on the shorter dsRNAs being less efficient.

These findings immediately suggest various ways to avoid TLR3 signaling. The simplest would be to stay below 21 nucleotides, and it is well known that 20 nucleotide siRNAs are equally potent inducers of RNAi. Unlike the somewhat disgruntled 1st commentator following my previous blog would like to suggest, 15-21 nucleotide siRNAs as covered by one version of Kreutzer-Limmer in Europe are therefore therapeutically highly relevant. Generally, keeping it short is the probably easiest way to avoid non-specific immune responses. Next to sequence length, limited modifications of the siRNA at sites where they interact with TLR3 as shown by the structure should abolish any TLR3 activation and possibly serve to antagonize TLR3, similar to what has been found for 2’o-methylation and TLR7. Likewise, changing the helical shape at one end which may also have the added benefit in encouraging asymmetric RiSC loading may be a third strategy of circumventing TLR3 activation.

It is clear from reading the papers that much of the motivation for performing these is related to RNAi Therapeutics. These are findings that are not just theoretical in nature, but very much of practical relevance. The speed with which this progress has been achieved illustrates the vigor and the many tools brought to bear by the scientific community on making RNAi Therapeutics become a reality.

Next-Generation RNAi

Listening to a webcast by Rosetta Genomics last weekend, I noticed their concerted effort to brand themselves as the Next-Generation RNAi company. Although this is a misnomer as their efforts are really centred around microRNAs, and is likely driven by a desire to get the attention of Wall Street, it made me reflect on what I expect from the next generation of RNAi drugs, specifically the design of RNAi triggers.

Ideally, the next development cycle will yield siRNAs with higher specificity and potency. This should allow for the use of lower amounts of drugs in the clinic for obvious reasons of safety, but also cost. At the moment, algorithms can pretty well predict siRNA sequences that will give a decent knockdown in tissue culture experiments in the low nanomolar range. However, once in a while, we stumble across those “super-silencers” that have IC50s in the mid-to-low picomolar range, yet we do not understand what makes them so good.

I expect that the intense study of the RNAi-related pathways in both model organisms and human cells will ultimately explain their behaviour and reveal rules for designing better and better siRNAs. Exemplary are recent studies by the Zamore group in the fruit fly system that showed that small RNAs are partitioned into separate RNAi effector complexes based on their structure as double-stranded precursors prior to loading into the activated RNAi effector complex. Similar to flies and most other multicellular eukaryotes, there are also a number of related RNAi effector complexes in human cells. It is, however, still unclear how much they differ from each other or what their functional overlap is. It is therefore intriguing to speculate that it were possible, similar to what has just been demonstrated for flies, to introduce small RNAs that would specifically harness the RNAi-cleavage pathway, while remaining invisible to the complexes responsible for the non-cleavage silencing pathways. This is because the microRNA-like non-cleavage pathways are responsible for most RNAi off-target effects, and it would further minimise competition with the endogenous microRNA pathway.

The use of different RNAi triggers (PolII::sh-miR; PolIII::shRNA; Dicer-substrate; Tuschl siRNA; 3-stranded siRNA) or, possibly even more exciting from a drug development perspective, chemical modifications and structural variations to the siRNAs may allow us to introduce the desired bias into which effector complex the small RNA will be incorporated. Along these lines, Dharmacon reported not long ago the use of chemical modification at the 2nd nucleotide position of the guide RNA that would still allow for on-target cleavage activity, but almost eliminated microRNA-like off-target silencing by the siRNA in tissue culture. Although a recent abstract by Alnylam scientists for the Annual Meeting of the Society for Neuroscience suggests that this particular modification may not always be neutral to on-target activity, a combination of chemical modification guided by a deepening understanding of RNAi pathways in humans should yield next-generation RNAi molecules with higher clinical success rates.