**6. Molecular engineering of AAV capsid**

There are several challenges for AAV serotypes to exert their therapeutic potential in target organs including the need for high vector doses for efficient delivery, pre-existing antiviral immunity in the host, and the lack of cell type-specific tropism leading to off-target transduction [6]. One way to overcome these limitations is to randomly generate capsid mutants from a library to extend the capability of the traditional AAV vector by increasing its cell transduction efficiency for specific cell types and its ability to escape from antibody neutralization.

One approach used to create a mutant library is DNA shuffling, a strategy in which the open reading frame of capsid genes of different AAV serotypes is fragmented by nucleases. This is followed by random ligation, resulting in new and random combinations of capsid sequences. These new molecular-enhanced AAV vectors exhibit a broad range of cell tropism with numerous functional differences between chimeras and their parent serotypes. Consequently, there is potential to produce unlimited numbers of new AAV variants with novel gene delivery properties. This method of AAV capsid engineering was first described in 2008 by Grimm and colleagues [54] and has become a commonly used technique over the years. More recently, Lisowski and colleagues utilized a humanized mouse model to perform serial selection using a human-specific replication competent viral library composed of DNA-shuffled AAV capsids. After four rounds of selection, they identified a novel chimeric capsid variant composed of five different parental AAV capsids [55]. Of these, LK-03, which efficiently transduced human primary hepatocytes both *in vitro* and *in vivo*, was found to be a human liver cell-specific AAV serotype [55]. This study has opened up a new avenue to validate therapeutic potential of an AAV capsid variant in preclinical studies using human primary cell xenotransplanted models prior to commencing clinical studies.

In addition, a study using *in silico* ancestral sequence reconstruction (ASR) of AAV capsid protein generated nine functional putative ancestral AAVs. In this study, Zinn and colleagues also identified Anc80, the predicted ancestral sequence of the widely used AAV serotypes 1, 2, 8, and 9 and showed that Anc80 is a highly potent *in vivo* gene therapy vector compared to AAV2 and AAV8 for targeting liver, muscle, and retina in mice [56]. Nevertheless, Anc80 demonstrated a high stability and no toxicity in several safety studies carried out in mice. This synthetic viral vector has been evaluated in non-human primates (rhesus macaques), which demonstrated a superior expression of Anc80 in monkey liver following Anc80 administration compared to control monkeys injected with AAV8. Hence, future studies may also rely on the use of Anc80, in particular for liver-directed gene therapy studies.
