Suppressing Nonsense for Gene Therapy

Genetic diseases may result from mutations that generate a nonsense codon in an open reading frame thus leading to premature termination of a critical protein. Proof of concept of a novel gene therapy approach to such diseases is provided in a recent report from Wang et al. ¹ that describes delivery of a suppressor tRNA (sup-tRNA^tyr) using an adeno-associated virus (AAV) vector to rescue a genetic disease in a mouse model carrying a nonsense mutation.

Overall, about 10–15% of human mutations that lead to pathogenesis in genetic diseases are due to generation of a nonsense or premature termination codon (PTC). ² Examples of diseases that may be caused by nonsense mutations include many lysosomal storage diseases (LSDs), hemophilia A, cystic fibrosis, Duschenne muscular dystrophy (DMD), Dravet disease (a severe childhood epilepsy), and β-thalassemia.

PTC may cause pathogenesis by at least two mechanisms; first by direct production of a truncated polypeptide with decreased functionality and second by leading to decreased levels, or complete absence, of the mRNA in a process known as nonsense-mediated decay (NMD). ³

Several approaches to treat nonsense-mediated diseases have been studied for clinical application. ⁴ PTC read-through can be mediated by aminoglycosides and other small molecules that induce near-cognate tRNAs to insert various amino acid residues. One such molecule, Ataluren, is approved in Europe for DMD. Another approach uses oligonucleotides that alter RNA splicing to skip the specific exon containing the mutation, and in the United States three such products, Eteplirsen, Golodirsen, and Casimersin, are approved for DMD.

Direct delivery of a sup-tRNA^tyr designed to ensure insertion of the normal (wild-type) amino acid residue was suggested many years ago ^5,6 but was not yet used in vivo mainly because of the absence of a practical delivery system.

A sup-tRNA^tyr can be generated from a naturally occurring tRNA by changing the anticodon triplet to one that recognizes a termination (nonsense) codon. ⁷ For instance, a tyrosine tRNA can be converted to a suppressor by altering the anticodon sequence to recognize the termination codon UAG and will thus insert a tyrosine in a protein instead of premature termination.

Wang et al. have now used an AAV vector packaged in a serotype 9 AAV capsid (rAAV9) as an in vivo delivery system in a mouse disease model to express a tyrosine sup-tRNA^tyr. The mouse model (Idua-W401X,TCG→TAG) has a homozygous mutation in the α-l-iduronidase gene that converts a tyrosine codon the stop codon UAG. This model recapitulates a human LSD, mucopolysaccharidosis disease type I (or Hurler Syndrome), caused by absence of the enzyme α-l-iduronidase (IDUA) leading to accumulation of glycosaminoglycans (GAG) and resulting pathogenesis.

In IDUA deficiency, relatively low levels of the wild-type enzyme can provide substantial correction of the pathogenic defects. Systemic delivery by injection of a rAAV9-suptRNA^tyr into the tail vein of adult male Idua^W401X/W401X knock-in (KI/KI) mice showed, after 10 weeks, restoration of enzyme activity to levels of 2.5%, 9.5%, and 27%, respectively, in serum, liver, and heart. ¹ Several hallmarks of the disease were also improved, including normalization of GAG accumulation in the liver and significant reduction in urine GAG concentration. Also, lysosome abundance in the liver was normalized.

These effects of the gene therapy were essentially stable for up to 6 months. Similar outcomes were seen in female KI/KI mice but were less effective probably because of the well-known decreased efficiency of rAAV in female mouse liver. It is worth noting that in cultured human fibroblasts with the analogous mutation (W402X/W402X) there was correction of the enzyme level to about 6% of normal by the sup-tRNA^tyr.

In the systemic delivery studies of the rAAV9-suptRNA^tyr there was only very low enzyme expression in muscle at about 0.5% of normal and brain was poorly responsive. However, muscle expression was improved by direct injection or with a modified muscle-specific capsid. Expression in brain after systemic delivery was improved by using a modified AAV capsid (AAV9.PHP) that may more readily cross the blood–brain barrier. These observations are consistent with our current knowledge about the biodistribution of various rAAV mediated by different capsids, and reflect extensive ongoing efforts in many laboratories to modify AAV capsids to improve and refine targeting of rAAV to specific organs.

The systemically delivered sup-tRNA appears to correct the defect both by insertion of the normal amino acid residue to generate functional Idua and by inhibiting NMD since the level of Idua mRNA was raised about 5- to 10-fold in liver and heart to reach normal levels.

This study provides clear scientific proof-of-principle for use of rAAV-mediated direct delivery of a sup-tRNA^tyr for treatment of genetic disease caused by nonsense mutations. What are the challenges involved in developing such an approach for use in humans? Primary considerations are the likely dose required and the safety thereof as well as possible off-target effects of nonsense suppression.

With regard to general safety of the rAAV administration, in the mouse studies there was no gross toxicity observed by histopathology or serum biochemistry. However, rAAV are usually we tolerated in mice, which are not necessarily predictive of responses in humans. The systemic delivery used a dose of 1 to 2 × 10¹² rAAV per mouse and this probably scales to about 5 to 10 × 10¹³ per kg (or a total dose of 10¹⁵–10¹⁶) for an adult 70 kg human. The dose required for humans might, therefore, be high but doses of this order of magnitude are being used in an approved rAAV9 product, Zolgensma, for spinal muscular atrophy, and for a rAAV5 in phase III clinical trials for hemophilia A. ⁸

A second important question is the likelihood of off-target effects of the sup-tRNA^tyr. Wang et al. provide an initial analysis of such off-target effects using a variety of sequence methods, including Ribo-seq, tRNA-seq, and mRNA sequencing, which suggest that any global read-through was modest and restricted to UAG (and not UAA or UGA) stop codons, that there was no substantial impact on endogenous tRNAs or amino-acyl charging efficiencies, and only a very low level of upregulated mRNA transcripts. These observations provide some assurance, but clearly off-target effects will require more substantial investigation for any specific therapeutic development.

The delivery of a sup-tRNA^tyr has some potential advantages. It does not require delivery of a cDNA for a gene that may exceed the packaging capacity of AAV and it may result in a more appropriate level of physiological expression because it operates on endogenous mRNA. Compared with other methods such as gene editing it may avoid potential immune responses to components such as CRISPR-associated proteins.

Wang et al. have provided an elegant proof of concept for a novel addition to the tool box of potential gene therapies. It will require more development to determine where this approach may be best applied in terms of specific diseases. However, one intriguing possibility is that delivery of a nonsense suppressor might be particularly useful for diseases such as DMD or Dravet syndrome (due to mutations in the gene, SCN1A, for a sodium channel) in which the size of the coding sequence required for the relevant protein substantially exceeds the rAAV capacity and a substantial proportion of the disease incidence is due to nonsense mutations.