Focused ultrasound as a novel strategy for noninvasive gene delivery to retinal Muller glia

Muller glia are specialized retinal cells with stem cell properties in fish and frogs but not in mammals. Current efforts to develop gene therapies to activate mammalian Muller glia for retinal repair will require safe and effective delivery strategies for recombinant adeno-associated viruses (AAVs), vectors of choice for clinical translation. Intravitreal and subretinal injections are currently used for AAV gene delivery in the eye, but less invasive methods efficiently targeting Muller glia have yet to be developed. Methods: As gene delivery strategies have been more extensively studied in the brain, to validate our vectors, we initially compared the glial tropism of AAV-PHP.eB, an AAV9 that crosses the blood-brain and blood-retinal barriers, for its ability to drive fluorescent protein expression in glial cells in both the brain and retina. We then tested the glial transduction of AAV2/8-GFAP-mCherry, a virus that does not cross blood-brain and blood-retinal barriers, for its effectiveness in transducing Muller glia in murine retinal explants ex vivo. For in vivo assays we used larger rat eyes, performing invasive intravitreal injections, and non-invasive intravenous delivery using focused ultrasound (FUS) (pressure amplitude: 0.360 - 0.84 MPa) and microbubbles (Definity, 0.2 ml/kg). Results: We showed that AAV-PHP.eB carrying a ubiquitous promoter (CAG) and green fluorescent protein (GFP) reporter, readily crossed the blood-brain and blood-retinal barriers after intravenous delivery in mice. However, murine Muller glia did not express GFP, suggesting that they were not transduced by AAV-PHP.eB. We thus tested an AAV2/8 variant, which was selected based on its safety record in multiple clinical trials, adding a glial fibrillary acidic protein (GFAP) promoter and mCherry (red fluorescent protein) reporter. We confirmed the glial specificity of AAV2/8-GFAP-mCherry, showing effective expression of mCherry in astrocytes after intracranial injection in the mouse brain, and of Muller glia in murine retinal explants. For in vivo experiments we switched to rats because of their larger size, injecting AAV2/8-GFAP-mCherry intravitreally, an invasive procedure, demonstrating passage across the inner limiting membrane, leading to Muller glia transduction. We then tested an alternative non-invasive delivery approach targeting a different barrier - the inner blood-retinal-barrier, applying focused ultrasound (FUS) to the retina after intravenous injection of AAV2/8 and microbubbles in rats, using magnetic resonance imaging (MRI) for FUS targeting. FUS permeabilized the rat blood-retinal-barrier and allowed the passage of macromolecules to the retina (Evans blue, IgG,IgM), with minimal extravasation of platelets and red blood cells. Intravenous injection of microbubbles and AAV2/8-GFAP-mCherry followed by FUS resulted in mCherry expression in rat Muller glia. However, systemic delivery of AAV2/8 also had off-target effects, transducing several murine peripheral organs, particularly the liver. Conclusions: Retinal permeabilisation via FUS in the presence of microbubbles is effective for delivering AAV2/8 across the inner blood-retinal-barrier, targeting Muller glia, which is less invasive than intravitreal injections that bypass the inner limiting membrane. However, implementing FUS in the clinic will require a comprehensive consideration of any off-target tropism of the AAV in peripheral organs, combined ideally, with the development of Muller glia-specific promoters.