Gene therapy using recombinant adeno‐associated virus (rAAV) vectors to treat blinding retinal dystrophies has become clinical reality. Therapeutically impactful targeting of photoreceptors still relies on subretinal vector delivery, which detaches the retina and harbours substantial risks of collateral damage, often without achieving widespread photoreceptor transduction. Herein, we report the development of novel engineered rAAV vectors that enable efficient targeting of photoreceptors via less invasive intravitreal administration. A unique in vivo selection procedure was performed, where an AAV2‐based peptide‐display library was intravenously administered in mice, followed by isolation of vector DNA from target cells after only 24 h. This stringent selection yielded novel vectors, termed AAV2.GL and AAV2.NN, which mediate widespread and high‐level retinal transduction after intravitreal injection in mice, dogs and non‐human primates. Importantly, both vectors efficiently transduce photoreceptors in human retinal explant cultures. As proof‐of‐concept, intravitreal Cnga3 delivery using AAV2.GL lead to cone‐specific expression of Cnga3 protein and rescued photopic cone responses in the Cnga3−/− mouse model of achromatopsia. These novel rAAV vectors expand the clinical applicability of gene therapy for blinding human retinal dystrophies.

The emergence of efficient viral vectors derived from adeno-associated viruses (AAV) has led many groups to develop gene therapies for inherited monogenic diseases, such as retinal dystrophies. To evaluate the potency of new gene therapy vectors in a preclinical context, it is common to use animal models, such as gene-deficient or mutant animal models of a given human disease, and then assess vision restoration with functional or behavioral assays. While such animal models are invaluable to the preclinical testing process, they cannot be readily used as batch release tests during manufacturing or to validate biological activity at later stages of development. There is therefore a need for rapid and reliable in vitro models that can determine whether therapeutic vectors have delivered their cargo gene, and more importantly, whether this has resulted in the intended biological activity. Given our previous experience, we chose CNGA3-linked achromatopsia to develop a cell-based system to verify biological activity of AAV vectors designed to deliver a healthy CNGA3 gene copy into human cone photoreceptors. Our system is based on an immortalized cell line with high susceptibility to AAV transduction, i.e., HeLa cells, which we engineered to express a fungal rhodopsin guanylyl cyclase (RhGC) from Blastocladiella emersonii and a sensitive genetically encoded calcium indicator (GECI) under the control of a tetracycline operator. Using this system, we were able to confirm and quantify the function of the ion channel encoded by AAV/CNGA3 and differentiate between AAV vector potencies with a simple fluorometric assay. Finally, we show that this approach can be readily adapted for the assessment of phosphodiesterase function.