AAV Capsid Engineering: Zooming in on the Target

Clinical evidence indicates that sustained expression of various therapeutic genes could be achieved after in vivo gene therapy using adeno-associated viral vectors (AAV). ^{1 –3} Though this is encouraging, the therapeutic gene should ideally be delivered only to the desired target cells while preventing unwanted gene delivery to so-called “off-target” cells. Unfortunately, the current gene therapy vectors that have so far been used in clinical trials result in significant off-target transduction. Consequently, this diminishes their overall efficacy. Indeed, for many diseases, therapeutic gene expression should be restricted to the target tissue that is directly afflicted by the underlying disease, since its inadvertent expression in off-target cells is not expected to yield any beneficial effects. Moreover, ectopic expression in off-target cells may also increase the risk of undesirable side-effects due to unexpected cellular toxicity. This could be compounded by untoward inflammatory reactions provoked by the vector itself. This is consistent with clinical evidence revealing dose-limiting toxicity of AAV vectors due to hepatic transduction. ⁴ It has been proposed that this was due, at least in part, to AAV capsid-specific T-cell responses that resulted in an acute inflammatory reaction and concomitant clearance of the transduced hepatocytes. ⁵ Consequently, it would be important to maximize AAV-mediated gene delivery in the desired target tissue, while minimizing off-target transduction in the liver. For example, in the case of Duchenne muscular dystrophy (DMD), the therapeutic gene should ideally be expressed in the skeletal muscle, heart, and diaphragm, since the function of these tissues is compromised in DMD, while gene transfer and transgene expression in the liver should preferably be avoided.

Though careful selection of the appropriate AAV capsid serotype may result in a more favorable biodistribution, off-target transduction cannot be excluded. Consequently, further refinements of the AAV vector by capsid engineering is required to obtain a more optimal transduction profile and minimize the risk of side-effects. In this issue, Tarantal et al. demonstrated that AAV capsid engineering resulted in reduced off-target transduction in the liver in newborn nonhuman primates. ⁶ This was consistent with the low vector copy number in the liver and the barely detectable reporter gene expression. To achieve this, they used an AAV capsid variant, designated as AAV2i8, that was initially developed by Samulski et al. ⁷ The AAV2i8 capsid contained six amino acids from AAV8 in the altered capsid surface loop of AAV2. Though it is known that that vector biodistribution profiles do not necessarily translate from mouse to large animal models, it is particularly encouraging that AAV2i8 resulted in significant liver detargeting in nonhuman primates compared with AAV9. This was consistent with the reduction of hepatocyte gene transfer with AAV2i8 in mouse models. ⁷

These preclinical studies in nonhuman primates suggest that AAV2i8 may also result in efficient liver detargeting in human subjects. However, this can only be addressed in future clinical trials. Diminishing inadvertent gene transfer in hepatocytes with AAV2i8 constitutes an important step toward optimizing the target cell specificity of AAV vectors and may reduce the risk of hepatotoxicity associated with AAV-specific inflammatory immune responses. ^4,5 Compared with AAV9, the AAV2i8 vector unexpectedly resulted in relatively robust transduction of intercostal muscles in the primate model. This is another added advantage of AAV2i8, particularly since these types of muscles are dysfunctional in many neuromuscular disorders. This eventually results in high morbidity and lethal complications due to pulmonary failure. In addition, AAV2i8 also resulted in significant luciferase reporter gene expression in the heart and skeletal muscle in the primate model. ⁶ Hence, the current study has important implications for the development of gene therapy of neuromuscular disorders, in particular DMD. The use of AAV2i8 hereby overcomes some of the limitations of early-generation AAV vectors based on naturally occurring serotypes. This sets the stage to assess further the efficacy and safety of AAV2i8 for gene therapy of neuromuscular disease in the appropriate preclinical disease models that mimic the cognate human disorders.

Nevertheless, AAV2i8 is no “magic bullet,” and low-level off-target gene transfer was still apparent. Moreover, vector copy number after AAV2i8 gene transfer was at least one order of magnitude lower in the heart and certain skeletal muscle groups (e.g., diaphragm) compared with AAV9. It is noteworthy that this superior cardiac transduction of AAV9 in the primate model was consistent with its unprecedented cardiac tropism in mice compared with other AAV serotypes. ^{8 –10} This implies that AAV9 still serves as the gold standard for cardiac transduction, with implications for future clinical studies to treat heart failure. Consequently, additional capsid engineering may be warranted to combine the favorable attributes of AAV2i8 with that of other serotypes in the hope of ultimately identifying the ideal “molecular key” to deliver the therapeutic cargo efficiently and safely to its correct cellular address.