Roche Impresses with Effective RNA Editing of Polyglutamine Repeat mRNA

Roche has shown interest in RNA Editing through its 2021 partnership with Shape Therapeutics.  The goal of this partnership was to use Shape’s AAV-delivered, DNA-directed RNA editing nucleic acids for neuroscience and rare disease applications.

Readers of this blog will know that I have not been a great fan of DNA-directed approaches to ADAR editing, not least because the expressed editing RNAs are unmodified.  This means that they do not benefit from chemistry to optimize efficacy.  In terms of specificity, the simple, but very effective strategy of modifying the base opposite non-target adenosines (e.g. 2’-O-methyl) to abolish off-target editing is not available to DNA-directed RNA editing.  

To compensate the efficacy disadvantage, the concomitant gene therapy-directed overexpression of ADAR enzymes has been attempted.  Unfortunately, this is a no-go since it causes extensive genome-wide off-targeting.  

It therefore comes as no surprise that Roche has also been evaluating synthetic editing oligonucleotides as revealed earlier this month in patent publication WO2023/052317A1.  This patent application addresses CAG/polyglutamine repeat expansion diseases such as Huntington’s disease, but also other neurodegenerative polyGln diseases including a number of the spinal cerebellar ataxias.  Since the number of polyGln repeats critically determines whether a person will manifest the disease and is correlated with protein aggregation, disrupting stretches of CAG-encoded uncharged glutamines with even a few positively charged, CGG-encoded arginines may stop the pathogenic process and is thus a highly attractive therapeutic hypothesis.

Beyond CAG triplett expansion diseases, similar logic may apply to diseases caused by repeat expansions in non-coding regions- as long as the repeat contains an ‘A’ such as in Friedreich’s ataxia (frataxin GAA repeat in intron 1).  Regardless of the specific disease-causing mechanism, disrupting the repeat is likely to be beneficial.  

While attractive in theory, I had been wondering how easy it actually would be to target these repeats by ADAR editing as the target sequence is quite unusual in its repetitiveness which may result in impenetrable higher-order structures.  The use of repetitive oligonucleotides as therapeutic agents is also unusual because of potential structural and manufacturing issues.  Finally, once one of the target adenosines has been converted to an inosine, the target mRNA sequence is altered (=mismatch) and consequently may become a weaker target site.

On the other hand, long repeats may turn out to be excellent targets in that they provide for a high local concentration of target sequence.

Actual data

Unfortunately, conducting casual molecular biology experiments in the basement of private homes is frowned upon in Germany and fraught with legal risks (this has to change), so it’s nice that Roche has actually conducted initial tissue culture experiments to find out about the practicality of the approach. 

Employing ~50-60nt long CUG repeats (the complement of CAG), their editing oligonucleotides were above the typical length of ~30nt as now generally practiced by the leading RNA Editing companies ProQR and Wave Life Sciences.  These were transfected into HeLa cells expressing ATXN3 mRNA with 21-22 repeat CAGs all in the apparent absence of ADAR overexpression.  

The oligonucleotides were modified with 2’-o-methyl only in the 5 nucleotides on the 5’ and 3’ ends each; phosphorothioation of the backbone was also practiced at the wings of the oligos, but extended further into the center than the 2'-o-methyls.  The central part consisted of pure RNA. 

An orphan C was placed towards the 3’ end of the targeting oligo.  This creates a mismatch to the target A as is commonly practiced in the field.  Interestingly, an inosine follows 3’ of the orphan C and this is also practiced by some other companies as e.g. evidenced in last year’s high-profile paper on long-lived and potent ADAR editing in non-human primates by Wave Life Sciences in Nature Biotech.

Remarkably, robust 20-50% AàI conversions were seen for many As in the ATXN3 CAG repeat with more pronounced editing towards the 5’ end of the repeat region consistent with the 3’ placement of the orphan C in the targeting oligonucleotide.  Moreover, less than 2% of the ATXN3 mRNAs was unmodified for each editing oligo.  If you consider that a huntingtin allele with say 33 CAG repeats does not result in Huntington’s disease, but one with 37 repeats typically does, you can imagine the impact that just a single or two successful editing events should have on pathogenicity of the resulting protein.

This experiment thus is an important de-risking step for RNA Editing in repeat expansion diseases and should whet the appetite of Roche which is already heavily invested in oligonucleotide therapeutics for Huntington’s through its collaboration with Ionis Pharmaceuticals (RNaseH mechanism), including research on improving the convenience and efficacy of intrathecal oligo administration.

As an investor in ProQR I was, of course, pleased to see that when discussing the prior art of ADAR editing in general, all 5 patent applications cited by Roche referred to ones controlled by ProQR. 

Looking forward to the next chapter in this story.

RNAi Also Conquers the Central Nervous System

In the span of a day, RNAi Therapeutics have gone from a mechanism widely viewed as being constrained to the liver only, to a major therapeutic modality for many targets and indications in a variety of tissues.  Due to the demonstrations in the liver, lung (yesterday), and today the central nervous system to potently and specifically knock down genes with infrequent dosing, RNAi will play a prominent role in today’s precision medicine-oriented drug development.

Employing C16 lipid-conjugated, chemically stabilized RNAi triggers, Alnylam and their partner in CNS drug development Regeneron achieved 84-90% maximal target gene knockdown with knockdown persisting at >70% for at least 3 months after a single dose.

Since chemical stability has been key to the successes in the lung, CNS, and also liver, it seems very likely that similar breakthroughs will be achievable for muscle, kidney, adipose tissues, and (in the words of Alnylam's President) 'even tumors' that Alnylam and Arrowhead are working on.

The initial target in the phase I study of ALN-APP was amyloid beta precursor protein (APP). Unlike the armada of antibodies that have targeted every known aggregation form of abeta for the treatment of Alzheimer’s, ALN-APP reduces them all and before they are even made thereby offering a unique angle to this important target.  An even more exciting near-term application of ALN-APP in my opinion is for cerebral amyloid angiopathy (CAA) where abeta accumulation near blood vessels can lead to intracerebral hemorrhage.  Studies with antibodies in Alzheimer’s have actually led to fatal damage to those very intracerebral blood vessels by causing local inflammation, and thereby make them a bad choice for CAA.

Beyond abeta and tau for Alzheimer’s, the CNS in particular abounds with otherwise difficult-to-drug important targets for diseases like Parkinson’s, Huntington’s, ALS, spinocerebellar ataxias for which RNAi is ideally suited.

The prolonged and robust knockdown observed is significantly better than what has been observed for previous RNaseH antisense candidates such as against SOD1 and tau (~50% target gene lowerings).  Safety also appears to be superior to the broadly phosphorothioated antisense molecules with no changes in neuronal markers of damage and inflammation seen with ALN-APP compared to placebo.

The US FDA though slapped a clinical hold on the multi-dose part of the trial based on findings in standard preclinical animal tox studies at doses well above what will be needed in the clinic.  The single dose exploration study, however, has been allowed to continue, and Canada has already allowed the multi-dose part to go ahead.  It therefore seems highly unlikely that the findings could derail ALN-APP or even this technology approach at this point.

With the recent news, the pharmaceutical landscape has changed and Big Pharmaceutical companies will have to think hard whether not having a stake in RNAi as a platform is viable.  The achievement is also one of delivery and stabilization chemistry which can be more broadly applied to other oligonucleotide therapeutics modalities in the CNS.

Oligonucleotides Break Through to the Lung

It is days like today that I live for.  Days when new platform technology data is revealed that will change the practice of medicine and benefit patients for a number of diseases of high unmet need. In this case asthma, IPF, COPD etc. 

Almost a decade after GalNAc started to revolutionize oligonucleotide therapeutics delivery to the liver (hepatocytes) and turned oligonucleotides into the important therapeutic modality it has become today, Arrowhead Pharmaceuticals just reported the equivalent for the lung (lung epithelial cells to be precise).

Employing inhaled delivery of αvβ6 integrin-targeted stabilized RNAi triggers in healthy volunteers, the company found robust, -80% mean maximum target gene (RAGE) knockdown after 2 doses spaced a month apart. 

Since the knockdown reading was based on RAGE protein in serum (sRAGE), the true knockdown in the desired lung epithelium is likely higher.  This is also supported by the observation that more direct bronchoalveolar lavage measurements revealed -75% knockdown after just a single 92mg dose when the corresponding reading in the serum indicated -56%.  Further dose escalation to 184mg is ongoing and there are first indications that the long-lived pharmacodynamic response observed in animals will hold up in the clinic.

RAGE is a key player in pro-inflammatory signaling in the lung and thought to play a central role in related pulmonary disorders such as asthma.

In addition to clearing the efficacy hurdle, safety seemed excellent, or in the words of the company ‘no patterns of adverse changes in any clinical safety parameters’.

As some may remember, an earlier RNAi candidate targeting the lung (ENaC for Cystic Fibrosis) was shelved by Arrowhead due to preclinical findings in chronic tox studies in the rat.  The reason is thought to be that the sheer amount of material delivered to rat lungs overwhelmed and inflamed the macrophage-based particle clearance system.

What is different this time is that ARO-RAGE utilizes improved stabilization chemistries and therefore only a fraction of the overall tissue exposure is required to achieve the same knockdown. 

This is reminiscent of the early days in GalNAc conjugate-based delivery to the liver when a first-generation GalNAc-TTR RNAi trigger had to be discontinued by Alnylam due to adverse safety in the clinic.  Improved GalNAc RNAi drugs of increased metabolic stability (and reduced 3'-fluoro content) are now well established medicines.

Beyond RNAi Therapeutics, today’s results have important implications for oligonucleotide therapeutics applications in the lung in general, including RNA Editing. 

Most importantly, they establish αvβ6 integrin as a valid target receptor for oligo conjugates.  Moreover, some of the chemistries should be directly translatable for stabilization purposes and together with ARO-ENAC Arrowhead should now have good insights into the chemistry-safety relationship. 

ProQR and Partner Eli Lilly Demonstrate Oligonucleotide-induced RNA Editing in the CNS: A Major De-risking Event for the Industry

With every new oligonucleotide therapeutics modality that feeds into an endogenous cellular mechanism comes uncertainty as to whether the mechanism is sufficiently robust to be of therapeutic utility.  

This is especially true for RNA Editing as in its early days targeted AàI editing was only shown with the concomitant DNA-directed overexpression of ADAR along with an targeting RNA or the introduction of recombinant ADAR-antisense conjugates of little direct therapeutic use.  Similarly, simply introducing into a cell a chemically synthesized antisense oligonucleotide hybridizing to the area surrounding the target adenosine in an mRNA will only give you minute editing efficiencies in cell culture without further structural and chemical optimization.

The liver and CNS, due to their gene target richness and the demonstrated clinical feasibility of delivering oligonucleotides to these organs, are of particular importance to the RNA Editing industry.  The demonstration byscientists from Wave Life Sciences of oligonucleotide-directed RNA Editing in non-human primates was therefore an enormous de-risking event in that it showed that RNA Editing is sufficiently robust in living primate livers.

Of similar importance was the revelation by ProQR and their partners from Eli Lilly last week that this also holds true for the primate nervous system following the intrathecal administration of an editing oligonucleotide.  10-30% editing were seen in the brain depending on the anatomical location investigated.  In both the mice (intracerebroventricular delivery) and cynomolgous monkeys, editing was highest in the cortex.  Even higher editing levels, up to 50%, were observed in the spinal cord of non-human primates.

The spinal cord (motor neurons) also happens to be the location of the most successful oligonucleotide therapeutic currently on the market: SPINRAZA (nusinersen) for spinal muscular atrophy.  Since RNA Editing is quite new and many do not fully appreciate what 10-50% editing efficiencies mean, SPINRAZA can serve as a good example for how impactful such target engagements can be particular for gain-of-function approaches.

SPINRAZA is a splice modulator and works through gain-of-function by obscuring an intronic splice silencer element in the SMN2 pre-mRNA.  Typically, only 10-20% of SMN2 mRNA is ‘correctly’ spliced to yield a functional full-length protein.  With 12mg of SPINRAZA in infants (same dose used for the RNA editing studies in cynomolgous monkeys), this increases 2-3x.  This means that an approximately 10-40% successful target engagement can save babies from certain death and, if given early enough, may allow children with the type I SMA mutations to grow up almost normally.

In the case of the (undisclosed) target gene that Eli Lilly is looking at, these types of target engagements with RNA editing resulted in 5-25x increases in protein function.  Because of the above and because gain-of-function is a particular competitive strength of RNA editing, this application should be prioritized in target selection of industry pipelines.