In a previous post I introduced the biology of gene therapy and reviewed the major efforts in prion gene therapy to date.  The present post will explore a few aspects of prion gene therapy in greater detail.

antisense oligonucleotides: probably precluded by adverse reactions

A colleague recently pointed me to a recent paper on ASOs in prion disease that I had missed on my first scan of the literature [Nazor Friberg 2012].  Earlier studies had shown that chemically modified ASOs (phosphorothioate oligonucleotides, which are nuclease resistant and so more stable in cells [Bennett & Swayze 2010]) reduce PrPSc formation regardless of sequence in cell culture [Karpuj 2007] and in mice [Kocisko 2006].  That’s right – even if not designed against PrP mRNA, their chemical structure itself is somehow inhibitory of prion conversion.  Kocisko provided some evidence that this may be because these ASOs bind PrPC and induce its endocytosis.  Kocisko also did therapeutic trials using random ASOs and found significant effects for prophylactic treatment or co-incubation of inoculum with the ASOs.

Nazor Friberg’s study appears to be the first to look at ASOs designed specifically to target the PrP mRNA sequence.  The authors screened 78 ASOs designed for complementarity to PrP mRNA’s 3′UTR; the most effective ASO they identified achieved a knockdown of 97% in cell culture.  Curiously, the maximum inhibition of protein (as opposed to mRNA) was 84% and was achieved only after two weeks of incubating cells with the ASOs – a timeframe much longer than the mRNA and protein half-life would lead us to expect – does this reflect slow uptake of ASOs?  In vivo, with intraventricular infusion, the best ASO was able to achieve a ~70% knockdown of PrP – both of mRNA and of protein – in brain regions immediately adjacent to the infusion site.

In a survival study, mice infused with ASO at 1 day post infection (dpi) appear to have lived to an average of something like 200 dpi (see Fig 6; exact average is never stated) compared to 136 dpi for controls, so a ~50% extension of survival.  The dose given was 100 times lower than used in the earlier studies of generic (non-sequence-specific) ASO interference with PrPSc formation, so the effect is presumably due to specific PrP mRNA degradation.  Troublingly, though, mice treated at 60 dpi (i.e. deep into prion infection but still before obvious symptoms) had a severe adverse reaction and had to be sacrificed.  The reasons for this reaction are not clear.

Intracerebral infusion of ASOs is invasive enough that it’s never going to be something that healthy volunteers – even PRNP mutation carriers – would sign up for.  As of now, it seems ASOs’ only realistic route to becoming a treatment is if they work in symptomatic prion disease patients, and the adverse reactions in prion-infected mice make it unlikely that ASOs could be used as treatments so late in the disease course.

Of course, since we don’t know exactly what triggered the reaction at the molecular level, it’s too early to say whether it’s an insurmountable problem.  Perhaps a different delivery method, or a different chemical modification to the ASO backbone, would avoid the adverse reactions.  The biggest news in gene therapy of late is that Roche and Isis have announced they’re investing $33M to bring anti-huntingtin ASOs to clinical trials for Huntington’s Disease [HDBuzz], following on last year’s landmark demonstration of feasibility and efficacy in animals [Kordasiewicz 2012].  Roche is talking up a ‘brain shuttle’ technology for ASO delivery across the blood-brain barrier without intrathecal or intraventricular infusion, though so far the web seems to be scant on details of what this ‘shuttle’ consists of.  In any event, there is sure to be a lot of exciting development on ASOs for the brain over the coming years, and it’s worth staying tuned.  But for now, at least, ASOs don’t seem to be the most promising path forward for prion disease.

maximizing the efficacy of RNAi

That leaves RNAi as the other method of antisense gene therapy.  The major study to date of RNAi in prion disease [White 2008] achieved an 18% extension of survival with one lentiviral shRNA injection late in disease course.  That’s pretty good – treatments are almost never effective late in prion disease – but 18% is also far from being a cure.  Does that mean RNAi is not so promising either?

Not so fast.  White’s study was an excellent demonstration of feasibility and a good piece of evidence for Mallucci’s ‘window for intervention’ hypothesis.  But that study wasn’t necessarily designed to all-out maximize therapeutic effect.  Here are a few reasons why:

  • siRNA screening.  The siRNA used in this study, dubbed MW1, only achieved about an 80% knockdown of PrP mRNA in cell culture (N2a cells, see Fig S1) and in the mouse hippocampus (Fig 1).  80% mRNA knockdown is pretty good for an siRNA, on the high end of what you’d expect from an off-the-shelf siRNA that you order online, but by no means the best of the best.  It’s not clear how much time White and Mallucci spent screening siRNAs to find the best sequence – all that the paper says is that “siRNA sequences targeting PrP were screened in vitro for efficacy of knockdown (data not shown)”.  With extensive screening to identify the most potent siRNA sequences, knockdown levels of well over 90% are possible.  Alnylam, a startup in Cambridge, MA focused on delivering synthetic siRNAs (i.e. direct siRNA delivery, no viral vector) has claimed 99% knockdown of transthyretin mRNA and protein in the mouse liver for ALN-TTR01.  Arrowhead Research has made similar claims of 99% knockdown of target genes in monkeys [Wooddell 2013 (ft)].
  • dosing. White’s study only examined one dose, delivered in a one-time injection into one brain region (the hippocampus).  Testing multiple doses would allow one to build out a dose-response curve and be able to test a dose with maximum efficacy.  Multiple timepoints and/or multiple injection sites could also help to achieve greater therapeutic effects, and none of that was explored in this study.  If intracerebral RNAi ever makes it to the clinic for prion diseases, it’s likely to involve multiple injection sites – see for instance the 12 injection sites used in a current Batten Disease clinical trial.
  • delivery.  By far the biggest problem in RNAi for the CNS is delivery – distribution across the brain.  Indeed, the 18% effect seen in White’s study is surely an overestimate of what a similar delivery technique would achieve in humans, for the simple reason that mouse brains are small and human brains are huge.  Indeed, this is a problem not just for RNAi but for any intracerebral delivery of therapeutics.  Intraventricular infusion, for instance, may penetrate 2mm into tissue from the ventricles, which would cover a large fraction of the mouse brain and negligible fraction of the human brain.  This is one factor that may have contributed to the failure of pentosan polysulfate in humans with CJD.  Luckily, RNAi delivery is under heavy investigation for a variety of diseases, and efforts in prion diseases will be able to benefit from those discoveries.  One simple (and by no means new) technique is ‘convection-enhanced delivery’, i.e. slow, pressurized injection, which can increase diffusion across brain regions and has demonstrated in monkeys and in human patients [Bankiewicz 2000 (ft), Voges 2003].  There are also efforts to develop ways to deliver RNAi (and drugs) to the brain from the periphery, for instance with lipid nanoparticles [reviewed in Bondi 2012].  So far none of these methods have proven very effective, but keep your eye on the Roche/Isis anti-huntingtin ASO developments.

outlook: how promising is RNAi?

We know from studies of Tet-off PrP mice that even 10% of wild-type PrP levels is sufficient to sustain a lethal prion infection, albeit with dramatically lengthened incubation times [Safar 2005 (ft)], while Cre-mediated knockout (reducing PrP probably very close to 0%) cures prion infection [Mallucci 2003].  There is presumably a threshold somewhere in between 90% and 100% knockdown where symptoms would be reversed and a patient would not experience prion toxicity within their lifetime.  As of now we don’t know if that threshold is something like 95% or something more like 99.999%.

As mentioned above, the best siRNA sequences can achieve as much as 99% knockdown of targeted transcripts and their protein products.  But even with such an optimized anti-PrP siRNA and optimistically assuming that the magic threshold to reverse prion infection is only 95% knockdown, RNAi is probably not poised to be an outright cure for prion disease given today’s technology.  The human brain is large and not super permeable to viruses, ASOs or synthetic siRNAs; no one has yet been able to demonstrate a uniform high distribution of gene therapy molecules across the whole brain.  Optimistically, 99% knockdown might be achieved adjacent to an injection site, but prion infection would still proceed elsewhere.

But that’s just using today’s technology.  The point of research is to make things better.  And in the scheme of neurodegenerative diseases, prion disease is actually a pretty good candidate in which to move RNAi forward, for a few reasons.  Prion diseases are entirely untreatable at present, and very rapidly fatal.  An 18% increase in survival may not seem like much, but it’s an entirely reasonable endpoint for a clinical trial.  For instance, earlier this year the FDA approved the use of Kadcyla (Herceptin with a toxin conjugated to it) for breast cancer based on an average survival of 30.9 months compared to 25.1 months for patients treated with regular Herceptin and other drugs – that’s a 23% extension of survival.  Or for an example from neurodegenerative disease, the FDA approved riluzole for ALS in 1995 based on a ~3 month increase in survival, against a background of about a 4 year life expectancy from diagnosis – that’s something like a 6% extension of survival.  The ASO and RNAi treatments currently under investigation for Huntington’s, Parkinson’s, and Alzheimer’s may well prove to have similar effects, but because those diseases progress slowly, it will be years before we know.  In comparison, average survival with sCJD is 7 months [Pocchiari 2004 (ft)].  The variance on that number is fairly high and depends on codon 129 genotype and a few other things, but still, if RNAi were to hypothetically increase survival by just a few months, then a clinical trial would probably be able to determine that result in a year or so.

The rapid mortality of prion diseases may also mean that the inherent risks of RNAi are more tolerable than in a disease like Huntington’s where people can live with symptoms for many years.  The experience with pentosan polysulfate has shown that families are not just willing to accept the risks of intracerebral infusion, but in fact will actively campaign to be allowed to take on risky treatments for their loved ones.

As of today, RNAi is not positioned to be a cure, though that could change if delivery systems improve and if the ‘reversal threshold’ proves to be low enough.  Cure or not, it does have potential to be a treatment for a class of diseases that have so far proved untreatable.  And we don’t know yet, but it’s possible that even a moderate knockdown of, say, 50% or 80%, could prove highly effective if combined with other treatments such as the newly introduced anle138b.  As of today, we are just at the beginning of having treatments that look promising in mice, and researchers have not even begun to explore what could be achieved by combining different treatments.  The road to the end of prion diseases may involve a few different strategies, and RNAi could yet prove to be one of those.