Arrowhead Publishes SubQ Delivery Technology to Go Beyond the Liver

In late 2012, Arrowhead Research shocked the Oligonucleotide Therapeutics world when it presented spectacularly potent and prolonged gene knockdown data in non-human primates using a subcutaneously administered single-molecule Dynamic Polymer Conjugate (DPC) formulation.  This arguably represented the most elegant delivery technology at the time.  Moreover, also due to its small, but not too small size (10-20nm) and slight negative charge, it provided us with a glimpse into the future of systemic RNAi delivery for regulating genes beyond the liver.

It certainly got my full attention and made me invest almost 100% of my stock portfolio back then.

Unfortunately, despite the validation in non-human primates which suggested clinical readiness would not be far off, the subQ DPC technology has seemingly struggled to reach clinical/commercial maturity. Not only Arrowhead’s lead development candidate, ARC520 for HBV, but also its second development candidate, ARC-AAT for AAT-related liver disease, was still based on the intravenously administered two molecule DPC version.  Although the reasons for the delays were never really disclosed, a few comments here and there made it seem very likely that chemistry and manufacturing issues were behind the delay.

Back to the Future

Last week, Arrowhead Research finally published a paper showing that single molecule DPC is still alive and kicking (Rozema et al.2015) and is progressing towards clinical application.  In essence, the new single-molecule subQ DPC prototype comprises of a membrane-active polymer which has been masked from premature cytotoxic interactions by pegylation and cell-targeting ligands that are added via protease-sensitive bonds; as before, the highly modified/stabilized RNAi triggers are appended by disulfide chemistry. 

The DPC is made in a 4-step process followed by a purification step to remove unwanted side-products and reactants.  The latter step is apparently important when going into primates.

The new old DPCs are thus distinguished from the intravenous version not only in that it combines the RNAi trigger and endolysosomal release polymer in a single molecule, but most importantly by the nature of its triggered release mechanism.  Whereas in the former DPC generations triggered release was dependent on changes in pH such as they occur when a DPC is endosomally taken up, they are now responsive to the presence of certain proteases in lysosomes. 

pH-dependent formulations apparently suffered from instabilities both in the body and during storage.  This was adequate for targeting genes in the liver because of the ready access of macromolecules in the circulation to this organ following intravenous administration, but not when the DPC first has to reach the circulation from the subcutaneous space and when less well accessible target organs are the ultimate destination.

Accordingly, non-liver single-molecule DPCs of the latest publication had impressive circulation half-times of the intact, protected molecule of 11 hours.  Similarly, such DPCs are stable for at least a year both in solution and when lyophilized. 

The extra-hepatic potential thus facilitated by increased stability now needs to be demonstrated by finding suitable targeting ligands and I’m sure Arrowhead has been busy working on that.  It should be noted that for target tissues where high concentrations comparable to the liver are unlikely to be achieved following systemic delivery, the extra kick that comes from an explicit release chemistry could provide a critical advantage over competing approaches.  These include simple conjugates of the GalNAc-type and probably also self-delivering RNAi trigger chemistries which incorporate ‘milder’ release chemistries (like lipid tails).

Knockdown lasting for weeks and months

The most impressive demonstration of the single molecule DPC performance in the Rozema paper came from the primate studies.  Here, a single administration of 0.5mg/kg 2’-O-methyl/F-modified RNAi trigger led to a highly potent knockdown (peak knockdown >95%) of liver expressed Factor VII with >80% knockdown of 2 and 4 months following subcutaneous and intravenous administration, respectively.

Following the 2012 delays and some uncertainties around what was really new and old in the recent publication, I am somewhat hesitant to declare that subQ DPC is now fully de-risked and ready-to-go.  In that regard, it would be helpful to learn more about the tox profile of the new molecules and related to that which polymers will be eventually used (e.g. 2-molecule with melittin-like peptide, a polyacrylate in the publication).

Nevertheless, since Arrowhead has said that the new 2015 development candidate may be from the subQ line of DPCs (or if not going after a extra-hepatic target) one would think that the most important challenges have now been overcome.

Alnylam’s Scientifically Dishonest GalNAc Claims

Arrowhead Research, as the successor of Mirus Bio, can regard itself as the father of GalNAc-conjugated RNAi delivery.  Accordingly, in 2007, Rozema and colleagues published a seminal paper in which a multivalent polymer-conjugated GalNAc construct was utilized for the hepatocyte-specific delivery of RNAi gene silencing.

In 2015, Alnylam likes to be recognized as the inventor of GalNAc-oligonucleotide Therapeutics, with competitors like Isis Pharmaceuticals and Solstice Biologics playing the roles of copy-cats, and Arrowhead Research failing to get much mention at all.

This, however, is as noted in the introduction far from the truth, and a recent paper on ‘sequentially assembled’ GalNAc-RNAi triggers by Alnylam (Matsuda and colleagues, 2015) is yet another example for how they would like to re-write history to suit their (IP) goals.  History repeating itself you might think after all we’ve been through with SNALP LNP.

Matsuda re-discovering Rozema

Alnylam likes to laugh off Arrowhead’s GalNAc approach by claiming that you need a magic triantennary GalNAc ligand design with highly specific geometry to achieve tight ASGPR target receptor binding and subsequent cellular internalization.  By contrast, Arrowhead Research would be only using monovalent GalNAc which are known, in isolation, to be much poorer ASGPR binders.

The existence of the triantennary design obviously has not evaded Arrowhead Research.  Nevertheless, they have opted for monovalent GalNAcs most likely for their chemical simplicity and therefore reduced cost of goods.

The apparently high cost of triantennary GalNAc synthesis was acknowledged in the Matsuda paper and was said to be the motivation for testing RNAi triggers in which instead of a single triantennary ligand, monovalent GalNAcs were distributed along the RNAi triggers.

Short story short, having GalNAcs conjugated on 3 sequential nucleotides or every other nucleotide did not impact potency much compared to the triantennary 'parent' design.  In other words, the benefit from multivalent binding can be achieved by bringing monovalent GalNAcs together in space.

This, of course, is the same principle behind the Arrowhead approach, where GalNAcs are added to the free amines along a polymer/peptide (an RNAi trigger is just another polymer).  In the case of the melittin-like peptide, I have highlighted the basic amino acids to which GalNAcs are expected to be conjugated:

NH-LIGAILKVLATGLPTLISWIKNKRKQ-COOH

As you can see, towards the C-terminus (right hand side) of the peptide, there is a cluster of 4 positively charged amino acids that is expected to generate a multivalent ASGPR binder (note that 3 and 4 GalNAcs have similar binding affinities).

Shockingly, while masquerading as the inventors of GalNAc Oligonucleotide Therapeutics with statements like these…

‘The triantennary GalNAc ligand was subsequently used for hepatocyte-specific delivery of antisense oligonucleotides and short interfering ribonucleic neutrals (siRNNs) in mice, and anti-microRNA therapeutics in humans, confirming the value of the parent trivalent design.’

…they failed to even cite the Rozema paper and went on to say that now (i.e. for the first time) they were going to test the hypothesis that sequential monovalent GalNAcs could do the same job.  This obviously is a clear case of willful scientific dishonesty in their campaign to re-write GalNAc history.

Silence Therapeutics not even a pimple

While the Matsuda paper is geared towards claiming the sequential GalNAc assembly idea and is an affront to Arrowhead Research, it is also a reminder that Silence Therapeutics has been similarly treated with disregard in Alnylam’s ‘invention’ of ‘enhanced’ GalNAc-siRNAs. 

This is because (like Arrowhead Research actually), Alnylam, at least in essentially all RNAi trigger examples in the Matsuda paper, uses the AtuRNAi trigger design, US patents of which claim 2’-O-methylation every other base with a staggered pattern as it regards the annealed guide and passenger strands.

It therefore looks more and more like Alnylam will have to approach Silence Therapeutics for a license sometime before enhanced GalNAc-siRNAs hit the market (at least 2 by 2020 according to Alnylam’s 2020 guidance), if not ALN-TTRsc already (~2017-8).  If Alnylam will have to approach Arrowhead Research for a license regarding GalNAcs, I do not know, but given Alnylam’s noise, worth paying attention to the intricacies of the various IP estates. 

Merck Paper Reveals Interest in GalNAc-targeted RNAi Therapeutics

Merck’s efforts have to be considered to be the strongest in RNAi Therapeutics among Big Pharma.  Its RNAi Therapeutics strategy so far, however, has consisted largely of trying to replicate the most promising technologies in-house. 

The latest publication by Merck RNAi scientists on the expression pattern of the asialoglycoprotein receptor (ASGPR1; Shi et al. 2013) confirms this as ASGPR is the target receptor of the two most advanced SNALP alternatives for gene silencing in the liver: the trail-blazing DPCs by Arrowhead Research (first use of GalNAc-targeted RNAi Therapeutics in Rozema et al.2007) and more recently the GalNAc-targeted siRNA conjugates by Alnylam (what these really seem to be will be covered in an upcoming blog...so stay tuned!).

Surveying ASGPR expression levels

When developing ligand-targeted therapeutics, it is important that the corresponding receptor is present on the target cell population.  Especially when targeting cancer, it can be difficult to find receptors that are not only present in large quantities, but also throughout the cell population.  

In the case of liver cancer (hepatocellular carcinoma/HCC), this question has occupied the RNAi field ever since the finding that SNALPs work really well for knockdown in normal liver due to uptake mediated by the LDL-receptor.  But is this mechanism also present on cells in liver cancer, a cancer of high unmet need where RNAi Therapeutics could have the biggest impact in oncology near- to midterm?

 

The Merck scientists set out to answer essentially the same question, but instead of interrogating LDL-receptor expression, they wanted to know about ASGPR expression on HCC cells.  Using tissue microarrays, they were able to test an impressively large set of 100s of tissues, including healthy human livers, liver cancer biopsies, and biopsies for other hepatic diseases such as viral hepatitis, chronic active hepatitis and cirrhosis.

Despite some apparent limitations with the tissue microarrays (e.g. normal liver samples were often marked as false negatives despite the known very high ASGPR expression level), the results seem to confirm that liver cancers in general have a tendency towards lower ASGPR expression and that inter-sample heterogeneity is comparatively large (some liver cancer samples had much more ASGPR expression than normal liver).  

This suggests that just as in the case in breast cancer where treatment decisions are often based on receptor expression levels, ASGPR-targeted liver cancer RNAi Therapeutics should also be combined with a companion diagnostics for ASGPR. 

For those curious about the status of ASGPR expression in chronic HBV since Arrowhead’s exciting ARC520 program for this indication involves a GalNAc-targeted melittin-like peptide, relax: livers infected with HBV express ASGPR just as well as normal livers.

Home-brew versus licensing

Based on the literature and conference presentations, it is reasonable to assume that Merck is well behind Arrowhead, and even Alnylam in developing GalNAc-targeted RNAi Therapeutics.  This begs the question, as it has in the case of SNALP, why does Merck not take a license or even acquire the original?

Of course, it always takes two for a deal, but given the financial capabilities of a Merck, if it wanted access, it could get it.

The likely explanation is that Merck's strategy in replicating technologies in-house is to first identify target technologies by validating them and then to develop viable IP workaround solutions.  In fact, in the apparent absence of a broad gate-keeping patent estate around GalNAc-targeted therapeutics, ASGPR has been a recognized drug target receptors well back into the 90's, there should be relatively little restrictions on the use of GalNAc per se.

Instead, a competitive advantage is largely gained through specific know-how like how to best link the GalNAc ligand to the oligonucleotide payload and synthesize the molecules cost effectively. The latter issue came to my attention recently when access to 'proprietary process for manufacturing GalNAc conjugates' from Alnylam was mentioned as the top corporate highlight in the quarterly update provided by Regulus Therapeutics.

For the aficionados, the triantennary GalNAcs as practiced by Alnylam are much more costly to synthesize than single GalNAcs as in the case of the liver-targeted DPC versions (where high affinity through multivalency is achieved by having multiple single GalNAcs along the DPC).

This situation is not unlike other areas in the drug development industry.  Take for example antibodies where, despite the various patent battles, there have been a number of commercially viable platforms based on specific libraries or optimization methods.  Nevertheless, despite this apparent freedom-to-operate, the way by which Big Pharma ended up gaining access to monoclonal antibodies was not by way of successfully developing them in-house, but by acquiring them.

I don't expect this to be any different in RNAi Therapeutics.