Gene therapy, using Adeno-Associated Viral vectors (AAVs), shows immense promise for treating various diseases. However, delivering genes to the brain poses a significant challenge. Direct brain injections are invasive, while systemic or intrathecal injections lack the spatial precision needed to target specific neural circuits. Focused ultrasound blood-brain barrier opening (FUS-BBBO) presents a noninvasive, site-specific solution. FUS-BBBO transiently loosens tight junctions in the blood-brain barrier (BBB), allowing AAVs from the bloodstream to enter the targeted brain region. This method offers precise targeting, unlike intraparenchymal injections (invasive and limited to small regions) or spontaneously brain-penetrating AAVs (lacking spatial specificity). Proof-of-concept studies have demonstrated FUS-BBBO's potential for delivering various genetic payloads to the brain. Despite its promise, FUS-BBBO has three major limitations: (1) inefficient AAV entry at the FUS-BBBO site requiring high doses, (2) substantial off-target transduction of peripheral organs due to BBB limitations, and (3) the use of wild-type AAV serotypes which haven't evolved to cross physically loosened barriers. This study hypothesizes that engineering new AAV serotypes specifically optimized for FUS-BBBO delivery will address these limitations.

Previous research has established FUS-BBBO as a viable method for noninvasive gene delivery to the brain. Studies have successfully used FUS-BBBO in rodents to deliver AAVs carrying reporter genes (GFP), growth factors, and optogenetic receptors. The delivery of chemogenetic receptors to the hippocampus demonstrated the ability to modulate memory formation. However, these studies typically utilized high doses of AAVs and observed significant off-target transduction in peripheral organs. Existing literature also describes methods for AAV capsid engineering to improve gene delivery properties such as tissue specificity, immune evasion, or axonal tracing. However, these techniques have not yet been applied to optimize viral vectors for specific physical delivery mechanisms like FUS-BBBO. The inefficiency of wild-type AAVs in crossing the physically loosened BBB underpins the need for vector engineering to enhance targeted delivery.

To engineer AAVs optimized for FUS-BBBO, the researchers employed a high-throughput in vivo selection approach. A library of AAV9 capsid variants containing 7-mer randomized amino acid insertions between residues 588 and 589 was generated. AAV9 was chosen for its superior transduction compared to other serotypes in previous FUS-BBBO studies. A Cre-recombinase-based selection method was used to enrich for AAVs that transduced neurons. Transgenic mice expressing Cre in neurons were used, and the AAV library was delivered intravenously to one hemisphere after FUS-BBBO. Viral DNA was extracted, and Cre-dependent PCR was used to selectively amplify DNA from transduced neurons. The process was repeated in a second round of selection using a focused library of the most abundant sequences. Next-generation sequencing (NGS) identified AAV variants with enhanced transduction at the FUS-BBBO site and reduced transduction in untargeted brain regions. Five promising candidates (AAV.FUS.1-5) were chosen for detailed evaluation. In validation experiments, AAV9 and each AAV.FUS candidate were co-administered intravenously during FUS-BBBO in individual mice. Transduction efficiency was quantified by counting mCherry (AAV9) and EGFP (AAV.FUS)-expressing cells in the brain and liver. Immunostaining was used to assess neuronal tropism by quantifying the percentage of transduced cells that were also NeuN-positive (neurons). The efficiency of delivery to different brain regions (cortex, striatum, thalamus, hippocampus, midbrain) was also assessed separately for AAV.FUS.3, and AAV9 was tested independently, using a new batch of AAVs in C57BL/6J and BALB/cJ mice. To confirm that the improvements are not limited to FUS-BBBO, intraparenchymal co-injections were also performed, comparing AAV9 and AAV.FUS.3 in the hippocampus. Finally, low dose testing at 10⁶ vg/g was conducted for AAV.FUS.3 and AAV9, and the brain to liver ratio was analyzed.

The in vivo selection yielded several AAV capsid mutants with improved FUS-BBBO-mediated gene delivery. Four out of five AAV.FUS candidates (AAV.FUS.1, 2, 3, and 5) demonstrated significantly higher transduction efficiency in targeted brain regions compared to AAV9 (p<0.01). AAV.FUS.3 showed the most substantial improvement, exhibiting a 12.1-fold increase in brain-to-liver transduction ratio compared to AAV9 at a high dose (10¹⁰ vg/g). Importantly, AAV.FUS candidates showed significantly reduced off-target transduction in peripheral organs, notably the liver. The improved neuronal tropism of AAV.FUS variants was confirmed by immunostaining, with a higher percentage of transduced cells being neurons compared to AAV9 (p<0.0001). AAV.FUS.3 showed improved transduction efficiency across five different brain regions, with the hippocampus exhibiting the highest relative improvement (4.3-fold). AAV.FUS.3's superior performance was also observed at a lower dose (10⁶ vg/g), maintaining a similar brain-to-liver transduction ratio and showing a 4.6-fold increase in neuronal transduction compared to AAV9 across three brain regions. This improvement is retained in the BALB/cJ mice (a strain different from the one used for selection), showing the versatility of AAV.FUS.3. Finally, even without FUS-BBBO, direct intraparenchymal injection of AAV.FUS.3 showed a 2.29-fold improvement in hippocampal transduction compared to AAV9.

This study successfully demonstrates the feasibility of engineering viral vectors to enhance noninvasive, site-specific gene delivery to the brain using FUS-BBBO. The engineered AAV.FUS vectors show substantial improvements over existing AAV serotypes. These improvements include increased transduction efficiency in the brain, reduced peripheral transduction, and enhanced neuronal tropism, all crucial aspects for safe and effective gene therapy. The superior brain-to-liver transduction ratio of AAV.FUS.3 (12.1-fold at the high dose, and 11.6 at the low dose) significantly reduces the risk of dose-limiting toxicity associated with off-target transduction in peripheral organs. The efficacy of AAV.FUS.3 at a lower dose suggests that it may be possible to use smaller doses and reduce the cost of gene therapy. The mechanism of enhanced transduction likely involves increased efficiency of AAV entry through FUS-opened BBB, possibly due to reduced binding to the extracellular matrix or altered interactions with the endothelium. Improved cellular tropism toward neurons further contributes to higher targeting specificity. Future studies should investigate these mechanisms thoroughly.

This study successfully engineered AAV vectors for improved ultrasound-mediated brain gene delivery. AAV.FUS.3 stands out with significantly enhanced targeting specificity, reduced peripheral toxicity, and efficient transduction at both high and low doses. This work demonstrates the potential of viral vector engineering to optimize gene delivery for specific physical delivery methods. Future research should focus on deciphering the underlying mechanisms of enhanced transduction and translating AAV.FUS.3 into clinical applications, while also testing its efficacy across various peripheral organs and nervous system components.

While the study demonstrates significant improvements in AAV-mediated gene delivery, it primarily focuses on two mouse strains. Further studies are needed to validate the findings in larger animal models and ultimately in humans. The study focuses on a limited set of brain regions; a more comprehensive investigation across various brain areas is warranted. The long-term effects of AAV.FUS.3 administration in the brain and other organs require further evaluation. Finally, cost-effectiveness and scalability of producing this modified AAV needs to be assessed before clinical implementation.