The tremendous development of new high-tech treatments for a wide range of human diseases has been made possible today by rapid advances in biomedical technology. However, despite many scientific discoveries in the field of gene and cell therapy, some diseases still have no effective treatments. Advances in genetic engineering methods have enabled the development of effective gene therapy approaches for a variety of diseases based on adeno-associated viruses (AAVs). AAVs can be effectively used to treat a wide range of inherited and acquired human diseases, more specifically monogenic inherited diseases, i.e. diseases caused by mutations in one gene. Today, many AAV-based gene therapy drugs are being studied in preclinical and clinical trials, and new drugs are available on the market for the treatment of a wide range of human diseases, including those that were previously considered untreatable.

To date, only a few AAV-based gene therapy drugs have been approved. For example, the U.S. Food and Drug Administration (FDA) recently approved "Hemgenix", a drug for the treatment of hemophilia B caused by congenital factor IX deficiency. The drug uses AAV serotype 5 (AAV5), which carries the gene for the lack of factor IX. As for hemophilia A, Roctavian was also recently approved by the FDA as a gene therapy product based on an AAV5 vector, carrying the defective factor VIII gene, driven by a liver-selective promoter. Another breakthrough gene therapy drug is "Zolgensma", which is designed to treat spinal muscular atrophy (SMA), a disease characterized by degeneration of motor neurons in the anterior horns of the spinal cord, leading to their loss of function. Zolgensma is based on an AAV serotype 9 (AAV9) vector, encoding the complementary DNA (cDNA) of the survival motor neuron 1 gene (SMN1). The drug was approved by the FDA in May 2019 for the treatment of type 1 SMA.

A Brief History of The Discovery of AAV

AAV is a small, non-enveloped DNA virus belonging to the family Parvoviridae, first isolated in 1965 as a contaminant in a preparation of simian adenovirus (Ad). These viruses were found to be unable to efficiently infect cells unless co-infected with a helper virus (usually Ad or any type of herpes virus), so they were named "adeno-associated viruses" and classified in the genus Dependovirus. Long considered to be defective viruses due to their co-dependency, subsequent studies of AAVs disproved this theory and showed that they initiate latent infections in host cells that can transform into productive infections under stress. Despite the high seroprevalence of AAV in humans (estimated to be 50% to 96% of the population seropositive for AAV serotype 2 (AAV2), depending on age and ethnic group), however, it has not been associated with any disease in either humans or any other species. Different AAVs have been isolated and detected not only in primates, but also in avian, goat, cattle, and equine species.

Properties, Structure, and Genome of AAV

With the exception of AAV5, which is the most divergent, all AAVs have similar structures and properties. AAVs are easy to manipulate because their particles can remain biologically stable under extreme pH and temperature conditions. They share a genome of approximately 4.7 kb single-stranded DNA, packaged in an icosahedral, non-enveloped capsid of 20-25 nm in diameter. The AAV genome consists primarily of two viral genes: rep (replication) and cap (capsid), flanked by inverted terminal repeats (ITRs). Since the ITRs have a palindromic nucleotide sequence, they generate a characteristic T-shaped hairpin structure that provides the necessary structural elements for viral genome replication and packaging. ITRs also play a regulatory role in viral gene expression and host genome integration. The open reading frame (ORF) of rep encodes several nonstructural proteins required for gene regulation, replication, transcription, and encapsidation, while the ORF of cap encodes three structural proteins: virion protein 1 (VP1), VP2, and VP3, which are present in a molar ratio of 1:1:10 in AAV particles. The different tissue tropisms of different AAV serotypes are due to differences in the processing of this cap ORF, resulting in different immune and transduction profiles.

AAV Serotypes and Tropism

Depending on its serotype, AAV has a specific tropism for specific organs and tissues of the body. There are different serotypes of AAV, and they differ in many aspects. Next, this article will discuss each serotype separately. Figure 1 shows their different tropisms.

Figure 1. Variant tropisms of AAV serotypes. (Issa S S, et al., 2023)

AAV1

The exact origin of AAV serotype 1 (AAV1) remains unknown, as it was not originally isolated from tissue but as a contaminant of Ad preparations, and antibodies to it have been found in both humans and non-human primates (NHPs). This serotype uses sialic acid as its primary cell surface receptor and the AAV receptor (AAVR) as a coreceptor. According to a systematic analysis of 10 AAV serotypes by Mary B. et al., recombinant AAV1 (rAAV1) showed no detectable post-translational modifications (PTMs). It was the first viral vector approved for gene therapy. In 1999, Xiao W. et al. conducted a study investigating viral vectors for gene therapy and found that AAV1 was the most efficient serotype for skeletal muscle transduction. Since then, many studies have confirmed that AAV1 has a high tropism for skeletal muscle of murine, canine, and NHP origin compared to other serotypes. AAV1 has also been found to achieve efficient transduction of neurons, glial cells, and ependymal cells in the mouse brain. Furthermore, it was found to be able to effectively transduce the heart, endothelial and vascular smooth muscle, and retina.

AAV2

AAV2 is considered to be the most studied serotype of all AAVs. It was first discovered in 1965 as a contaminant in simian Ad preparations. Later, in 1998, Summerford C. and Samulski R.J. discovered its major cellular receptor, heparan sulfate proteoglycan (HSPG), and subsequently proposed that the amino acid residues that provide its affinity for HSPG are R585 and R588. Binding of AAV2 to its major receptor was found to be insufficient for cell entry, and thus several of its co-receptors were later identified, including human fibroblast growth factor receptor 1 (FGFR1), αVβ5 and α5β1 integrins, hepatocyte growth factor receptor (HGFR), laminin receptor (LR), and CD9. The capsid of rAAV2 has been reported to acquire multiple PTMs, including ubiquitination, phosphorylation, SUMOylation, and multisite glycosylation. As the most studied serotype, AAV2 in fact exhibits a wide range of tropisms for a variety of tissues in NHP, murine, canine, avian, and human cell types, including renal tissue, hepatocytes, retina, non-mitotic cells of the central nervous system (CNS), and skeletal muscle. Nonetheless, further innovations in AAV chimeras and cross-packaging (or cross-typing), where the viral genome of one serotype can be packaged into a capsid of another type, have given AAV2 a broader tissue tropism.

AAV3

AAV serotype 3 (AAV3) was originally isolated from humans. Similar to AAV2, this serotype uses HSPG, FGFR1, and LR receptors, as well as the human HGFR (hHGFR) receptor. AAV3 was mostly overlooked as an option for gene therapy due to its insufficient transduction efficiency in vitro and in mouse cell lines. However, due to its later discovery of using hHGFR as a coreceptor, it exhibited extremely efficient transduction of human hepatoma cells as well as human and NHP hepatocytes. Since the discovery of this selective tropism of AAV3, various studies have aimed to optimize strategies to generate rAAV3 vectors with higher transduction efficiency. The developed strategies proposed different approaches, mainly the capsid modification of AAV3 vectors and the modification of hHGFR expression levels, as well as the tyrosine kinase activity associated with it. AAV3 was also found to have a specific tropism for cochlear inner hair cells, showing high in vivo transduction efficiency in a mouse model.

AAV4

AAV serotype 4 (AAV4) is considered to be one of the most antigenically distinct serotypes. It is reported to originate from NHPs, mainly African green monkeys, as antibodies to its viral particles have been detected in their sera. Studies on the structure of AAV4 have shown that its capsid surface topology has significant similarities with that of human parvovirus B19 and Aleutian mink disease virus. The only reported PTM of rAAV4 is the ubiquitination of its capsid protein. AAV4 is considered to be able to transduce human/NHP cells as well as cells of mouse and canine origin. The specific tropism of AAV4 results in transduction efficacy of specific cell types (primarily ependymal cells) in the mammalian central nervous system (CNS). In addition, after subretinal delivery, AAV4 exhibits stable transduction of retinal pigment epithelial (RPE) cells in rodent, canine, and non-human primate models, a unique feature enabled by the specificity of its capsid. In mouse models, AAV4 also demonstrated significant transduction rates of kidney, lung, and heart cells.

AAV5

Since AAV5 was first isolated from male genital lesions in 1983, it has become the only AAV serotype isolated directly from human tissue. This serotype is considered to be the most genetically divergent of all AAVs, with various unique features, such as the different size and function of its ITR region, the use of herpes simplex virus (HSV) as a helper virus for human infection, and the use of an atypical endocytic pathway as a route of viral entry. Another notable feature of AAV5 is its ability to transduce cells that AAV2 cannot, a unique advantage for gene therapy use. AAV5 has also been found to use sialic acid as its primary receptor, as well as platelet-derived growth factor receptor (PDGFR) α and β as co-receptors. AAV5 has been shown to have a remarkable transduction efficiency of mouse retinal cells, primarily photoreceptor cells. In addition, the tropism of AAV5 has been studied in the mouse brain, demonstrating its ability to transduce multiple neural cell types, including Purkinje cells, stellate, basket, and Golgi neurons, and its ability to reach the underlying neuropil and ventricular epithelium. AAV5 is also known for its efficient transduction of mouse airway epithelium, vascular endothelial cells, and smooth muscle via apical infection. It has also been reported to have a tropism for mouse hepatocytes.

AAV6

The classification of AAV serotype 6 (AAV6) remains controversial, as it shares a high degree of genomic similarity with both AAV1 and AAV2 serotypes, however, it still has its own serotype number. The serological characteristics of AAV6 are almost identical to AAV1, with up to 99% sequence homology in its coding region and multiple regions identical to AAV2. Therefore, it is considered to be a natural hybrid resulting from homologous RECOMation between AAV1 and AAV2. The only reported PTM of rAAV6 is acetylation of its capsid protein. Similar to the previously described serotypes, AAV6 can be purified by affinity chromatography on either heparin or mucin columns, as it binds both. AAV6 has been reported to have tropism for a variety of tissues, including airway epithelium in mouse and canine models, mouse hepatocytes, and skeletal muscle in mouse and canine models, with transduction efficiencies even higher than AAV2. In addition, cardiomyocytes in mouse, porcine, canine, and ovine models have been transduced.

AAV7

AAV serotype 7 (AAV7) was first isolated in 2002 from NHP tissues, specifically from rhesus macaques. The mechanisms of its cell binding and cell entry remain unknown, but it is well established that this serotype does not generally bind heparin or any other glycans. The capsid protein of rAAV7 undergoes multiple PTMs, primarily glycosylation, as well as phosphorylation, SUMOylation, and acetylation. AAV7-based viral vectors have demonstrated high transduction efficiencies for skeletal muscle cells in mouse models, similar to those achieved with AAV1 and higher than AAV2. The serotype has also been shown to have a strong tropism for hepatocytes in both murine and human tissues. In the CNS of NHPs, AAV7 viral vectors were found to achieve robust transduction primarily in cortical and spinal cord tissues. In addition, AAV7-based viral vectors have been reported to achieve significantly high transduction rates of mouse neurons and retinal photoreceptor cells in vitro and in vivo. AAV7 vectors also appear to have a limited tropism for vascular endothelial cells, which can be relatively enhanced by proteasome inhibition, and have an in vivo transduction preference for epicardial cells in mouse cardiac tissue.

AAV8

Similar to AAV7, AAV serotype 8 (AAV8) was first isolated from rhesus macaques in 2002. As the primary receptor, AAV8 binds to the LR, the same receptor used by AAV2 and AAV3. Phosphorylation, glycosylation, and acetylation are three PTMs reported for the rAAV8 capsid protein. AAV8 is known for its strong tropism for hepatocytes, resulting in a much higher transduction efficiency of hepatocytes than all other AAV serotypes in different models, including mouse, canine, and NHP. After systemic delivery in mouse models, AAV8 was shown to be the most efficient serotype for skeletal and cardiac muscle transduction due to its ability to cross the vascular barrier, a feature that both AAV1 and AAV6 lack, limiting their muscle delivery efficiency to local transduction. AAV8 was also able to successfully achieve in vivo transduction of mouse pancreatic cells after local delivery. Furthermore, high rates of transduction of mouse kidney cells were achieved by local delivery of AAV8 viral vectors directly into renal tissue. AAV8 was also found to achieve efficient transduction of different cells in the mouse retina, including amacrine cells, Müller cells, and presumptive bipolar cells, as well as some horizontal cells and cells in the ganglion cell layer (GCL).

AAV9

AAV serotype 9 (AAV9) was first identified in human isolates in 2004 and was named a novel serotype because it has serological characteristics that differ from previously known AAVs, but it is thought to be closely related to the clade containing AAV7 and AAV8. AAV9 appears to be able to achieve cell transduction with superior efficiency to other AAVs in most tissues. For example, in a study designed to investigate the distribution of AAV1-9 after systemic delivery in a mouse model, AAV9 demonstrated rapid onset of efficacy, optimal genome distribution, and the highest protein levels compared to all other AAVs. In addition, AAV9 can cross the blood-brain barrier and can transduce not only neurons but also non-neuronal cells that other AAVs cannot normally transduce, including astrocytes, and also shows tropism for retinal photoreceptor cells. AAV9-based viral vectors have also been shown to be more efficient than AAV1 and AAV8 viral vectors for transduction of mouse, NHP, and porcine myocardium. Another example of the superiority of AAV9 over other AAVs was presented in a study by Inagaki K et al., where this serotype achieved robust transduction of mouse hepatocytes, skeletal muscle, and pancreatic cells. AAV9-based viral vectors also appear to have affinity for mouse photoreceptor cells, renal tubular epithelial cells, Leydig cells in testicular interstitial tissue, and alveolar and nasal epithelial cells.

AV10 and AAV11

AAV serotypes 10 and 11 (AAV10 and AAV11) were first discovered and described in 2004 in NHP isolates (i.e., cynomolgus monkeys) and have capsid proteins that are very similar to AAV8 and AAV4, respectively, resulting in serological cross-reactivity with both serotypes. However, antisera raised against AAV10 and AAV11 were not found to have any cross-reactivity with sera raised against AAV2, suggesting that they are good viral vector candidates for gene therapy in individuals with antibodies against AAV2. A study investigating the biodistribution of AAV10 and AAV11 in monkeys showed a tropism for NHP enterocytes, hepatocytes, lymph nodes, and less so for renal cells and adrenal glands. AAV10 is also known to have a tropism for mouse small intestinal and colonic cells. AAV10 appears to have the greatest range of tropism for murine retinal cells compared to AAV8 and AAV9, as it has been reported to transduce multiple cell types, including RPE, cells in the ganglion cell layer, several cell types in the inner nuclear layer, photoreceptors, as well as horizontal cells. Following intravenous injection, AAV10 was found to target mouse hepatocytes and lung cells, however, following local delivery, it transduced mouse kidney and pancreatic cells. As for AAV11, it was found to have a mild tropism for the NHP CNS, primarily the brain and spinal cord. A recent study using rAAV11 for neurogene therapy found that this serotype could also target mouse projection neurons and astrocytes.

AAV12

AAV serotype 12 (AAV12) was first isolated from a simian Ad population and then characterized as a novel serotype because it exhibits unique biological and serological properties. Although it has been demonstrated that AAV12 does not use heparan sulfate proteoglycans or sialic acid for cell attachment and entry, how exactly it binds to target cells remains unclear. However, according to a study investigating the components of a potential receptor complex for AAV12, mannose and mannosamine are believed to be components of this complex as they inhibit AAV12 cell transduction. In addition, AAV12 is resistant to neutralization by human antibodies and is a good candidate for human gene therapy applications. In mouse models, it shows tropism for salivary glands and muscles. It also shows a strong, localized in vivo tropism for the mouse nasal epithelium after intranasal administration.

AAV13

AAV serotype 13 (AAV13) is another simian Ad that appears to bind to HSPGs, although its primary cellular receptor remains unknown. It has also been found to have structural similarities to AAV2 and AAV3, making it the AAV most related to these two serotypes. Its capsid retains the structural features of all AAV capsids, but data on the tropism and transduction efficiency of this serotype are limited.

Novel Hybrid AAV Vectors

In addition to the above-mentioned natural AAV serotypes, novel AAV vectors have been developed over the past two decades and are still being developed. Using different engineering strategies, novel hybrid vectors have been generated to enhance their transduction, modulate their immunogenicity, or restrict their tropism to specific cells. There are different types of such engineered novel vectors, including mosaic, chimeric, and combinatorial vector libraries. Chimeric vectors have subunits of multiple different serotypes in their capsids, selected for their receptor binding and intracellular trafficking properties. In chimeric virions, the capsids often have modified proteins generated by domain swapping and DNA shuffling strategies to change specific amino acids. Combinatorial vector libraries also use DNA shuffling and error-prone PCR methods to generate novel serotype AAV libraries with mixed genomes.