Transcribed by: Tia Byer
Adeno-associated viruses (AAVs) are the leading vehicle for in vivo gene therapy delivery. AAV vectors are preferable over non-viral vectors due to their non-pathogenicity and ability to target numerous cell and tissue types. Currently, AAV vectors have shown clinical efficacy in the gene therapy for 8 human diseases, including lipoprotein lipase deficiency, Leber congenital amaurosis, Leber hereditary optic neuropathy, aromatic L-amino decarboxylase deficiency, chlorideremia, haemophilia A and B, and spinal muscular atrophy. The AAV vector-based gene therapy market is expected to grow exponentially from USD 3.8 billion in 2019 to USD 13.0 billion by 2024, at a CAGR of 27.8%. Key players in the field include Abeona Therapeutics, Adverum Biotechnologies, AGTC, Asklepios Biopharmaceutical (Bayer), Audentes Therapeutics (Astellas Pharma), Avexis (Novartis), Biogen, Biomarin Pharmaceuticals, CSL Behring, Freeline Therapeutics, Pfizer, ReGenx Biosciences, Sanofi, Sarepta Therapeutics, Spark Therapeutics (Roche), Takeda Pharmaceutical, Ultragenyx, and many others.
AAV vectors also represent a promising drug therapeutic class, with the first of its kind being Glybera (AAV1) which was approved in 2012, but only in the European Union. Glybera was the most expensive drug for its time, costing a hefty $1.11 million per injection, and was subsequently withdrawn, being deemed a commercial failure with only a 2 patient uptake. The next was Luxturna (AAV2) in 2017, and in 2019 the FDA approved Zolgensma (AAV9) with a price tag of $2.125 million per patient, and Zolgensma surpassed Glybera as the world’s most expensive drug.
Friend or Foe: The Long Road to Optimisation
AAV vectors are the safest and most effective vehicle available and are prized for their ability to maintain long-term gene and protein expression with just a single injection. Since their advent, the gene therapy world has been posed with the question of how best to optimise AAV. Despite their initial promise, the first generation of AAV vectors, although only mildly immunogenic, raised concerns about the use of extremely doses.
Unfortunately, these fears were realised when in 2020, Audentes Therapeutics reported the deaths of 3 patients who had been treated with a high dose of the first generation of AAV8 vectors, and more recently in September 2021, the death of a fourth patient, even at a lower dose, in a gene therapy trial of X- linked myotubular myopathy. The causes of death were determined as either sepsis or gastrointestinal bleeding in the first 3 patients. The cause of death of the fourth patient is currently pending. Also in 2020, Lysogene reported the death of a patient treated with a high dose of the first generation of AAVrh10 vectors in a gene therapy trial of mucopolysaccharidosis type III (also known as Sanfilippo syndrome). The official cause of death remains unknown. In April 2021, Adverum reported that one patient lost vision in the treated eye that had been administered with AAV7m8, and in July, 2021, “similarly clinically-relevant events” in 5 of 12 patients who received a high dose of this vector. The moral of this story? The first generation of AAV vectors were not optimal. As such, there was a need to develop the next generation of AAV vectors.
The First Limitation of First Generation AAV Vectors
A major limitation of conventional first-generation of recombinant AAV vectors is that AAV evolved as a virus, and not as a vector. For example, if you look at the naturally occurring virus and the first generation of recombinant AAV vector, they look identical. This is because they are, which means the host immune system cannot distinguish between the naturally occurring AAV and the first generation of recombinant AAV vector, targeting both equally well. It can be argued that the two are not identical because the naturally occurring AAV contains its own genome, whereas the recombinant vector contains a therapeutic gene. However, the immune system cannot see what is inside the virus or the vector and targets and attacks as if they are both harmful infections. Therefore, the next generation of vectors must allow the immune system to distinguish between the two and respond accordingly. In other words, the vector must be different from the virus.
The Solution: The Next Generation of AAV Vectors
The first generation of AAV vectors contain 7 tyrosine residues that are surface-exposed and cellular tyrosine kinases target these tyrosine residues for phosphorylation. This means that a large fraction of incoming AAV vectors become phosphorylated which serves as a signal for ubiquitination. Proteasome-mediated degradation follows, which negatively impacts not only the transduction efficiency of these vectors, but the broken-down peptides also trigger a cytotoxic T cell response.
In 2008, the University of Florida College of Medicine’s Li Zhong and colleagues mutagenised each of these 7 tyrosine (Y) residues and replaced them with phenylalanine (F) residues, one at a time to generate 7 different Y-F mutant AAV vectors. They called these the next generation of AAV vectors and found that by changing just one amino acid on the capsid, the transduction efficiency of these vectors was increased between 2 to 30-fold in mouse liver following tail vein injection. When the 3 most efficient single Y-F mutants were combined into one capsid, Zhong and colleagues found that the resulting triple mutant Y-F AAV vector was up to 80-fold more efficient than the conventional first generation AAV vector in the mouse liver. It was therefore concluded that the NextGen AAV vectors are more efficient than the conventional first generation AAV vectors at significantly lower doses.
The Second Limitation of First Generation AAV Vectors
Importantly, an additional pressing limitation is that wild-type AAV and most recombinant AAV vectors contain single-stranded DNA. This is problematic since there is no host cell RNA polymerase that can transcribe a single-stranded DNA genome. Again, AAV as a virus does not express its own genes efficiently, as viral second-strand DNA synthesis is needed before gene expression can occur. Thus, to expect AAV as a vector to express therapeutic genes to high levels, is unrealistic. Therefore, in a vector, the single-stranded DNA genome requires modification to allow more efficient viral second-strand DNA synthesis to become transcriptionally active.
The Solution: Generation X Vectors
The University of Florida had known about this problem since 1997 when Keyun Qing and colleagues discovered that a tyrosine-phosphorylated cellular protein, FKBP52, binds to the single-stranded D-sequence at the 3’-end of the AAV DNA, and strongly inhibits AAV second-strand DNA synthesis. The idea at the time was to simply delete the D-sequence at the 3’-end so that FKBP52-binding would not interfere with AAV second-strand DNA synthesis, and thus, would lead to the development of transcriptionally active AAV vectors. However, when the D-sequence at the 3’-end was deleted, the AAV genome failed to get packaged. So, it was learnt that the D-sequence at the 3’-end is the “packaging signal” for AAV. However, in 2015, Chen Ling and his colleagues at University of Florida reported that the removal of the D-sequence at the 5’-end allowed AAV DNA to undergo successful packaging, which led to the development of what they called GenX vectors. Ling and colleagues showed that the extent of the transgene expression from GenX vectors was up to 8-fold higher than that from the conventional AAV vectors. GenX vectors therefore successfully overcame the second major limitation of the first generation of single-stranded AAV vectors.
The next obvious question was: Can GenX genomes be packaged into NextGen capsids to generate optimized (Opt) AAV vectors to achieve even higher-efficiency of transduction? Indeed, in 2016, Ling and colleagues described the development of Opt AAV serotype vectors, and showed that these vectors were 20-30-fold more efficient than the corresponding NextGen AAV serotype vectors. Thus, Opt AAV serotype vectors overcome both major limitations of the first generation of AAV serotype vectors.
If the history of the development of AAV vectors has taught us anything, it is that, unsurprisingly, AAVs did not evolve to be used as a vector for the delivery of therapeutic genes. The use of the first generation AAV serotype vectors composed of naturally occurring capsids proved less than ideal. Furthermore, the use of single-stranded AAV genomes with wild-type genome ends also resulted in sub-optimal transgene expression. Looking ahead, the University of Florida predicts that the full potential of AAV as a vector will be realized after its capsid is modified to evade the host immune response and its genome modified to express optimal levels of the therapeutic transgene. This optimism is based on the fact that one of these NextGen AAV vectors showed clinical efficacy in a Phase I gene therapy trial of Leber hereditary optic neuropathy in 2017. Further advancements in the development of AAV vectors are ongoing, bringing the field ever closer to reaching the safe and optimal gene therapy of a wide variety of human diseases. The NextGen, GenX, and Opt AAV vectors for human gene therapeutics are set to further change the healthcare industry for the better, and at Oxford Global, we can hardly wait to see how.