It’s All in the Delivery: Improving AAV Transfection Efficiency with Exosomes

Main Text

The efficacy of gene therapy treatments for inner-ear disease depends on the specificity and efficiency with which reagents can be introduced into the relevant cells. Several viral vectors, including adenovectors and adeno-associated vectors (AAVs), have been used to shuttle genes into cells of the inner ear,1, 2 but relatively low transfection efficiency continues to be an obstacle.3, 4 In this issue of Molecular Therapy, György et al.⁵ now show that AAV1 vectors linked to exosomes (exo-AAV1) achieve highly efficient transfection of all types of sensory cells (hair cells) within both the auditory and vestibular sensory epithelia of the inner ear.

Inner-ear diseases disrupting hearing and balance affect millions of people worldwide. Many years of research on the inner ear have contributed to our understanding of normal development and function as well as the causes and progression of disease. Researchers have recently begun to design therapeutic interventions aimed at hearing restoration and protection of the inner ear against ongoing damage. Techniques to manipulate gene expression in cells within the auditory system through surgical delivery of gene vectors and gene-editing agents present unique opportunities to address severe or progressive genetic forms of inner-ear disease in the clinic.

In the new study, György and co-authors⁵ show that AAV1 capsids bind to the surface and interior of exosomes and can be used to express transgenes in inner and outer hair cells (OHCs) of the cochlea (Figure 1). They further show that the overall level of transgene expression was significantly higher when using exo-AAV1 than free AAV1. The most impressive outcome of the transfection with exo-AAV was enhancement in the transfection efficiency of OHCs relative to conventional AAV1. This enhanced transduction of both inner and OHCs is promising, because many types of hearing loss are caused by damage to both cell populations.⁶

Figure 1.

Schematic Legend

The outer ear canal (A) ends medially in the tympanic membrane (ear drum, B), connecting to the middle ear ossicles (C). The inner ear consists of the vestibular portion (D) and the cochlea (E). A cross-sectional view of the cochlear spiral is shown in higher magnification in the middle portion of the schematic, and the sensory region, the organ of Corti, is shown on the right. Inner hair cells (F) and outer hair cells (G) are transduced by the exo-AAV vector with high efficiency (brown), although some remain without transgene expression (magenta). Some supporting cells are also GFP positive (H, green).

In addition to target-cell specificity, efficient vector delivery and gene expression depend on the mode of viral vector injection. To assess delivery route options, the authors compared two different approaches for injecting the viral vector into the perilymph (the fluid that bathes the sensory epithelium). They found that viral vector delivery via the round window (a membrane-covered fenestra at the base of the cochlea) is preferable to cochleostomy (a perforation created in the otic bone leading to the cochlear fluid), with better distribution of the virus throughout the cochlea and reduced variability in the efficiency of transgene expression. For clinical application in the future, it is notable that both delivery approaches (round window or cochleostomy) are feasible surgical routes for delivery of reagents into the human cochlea, and both currently are used for inserting cochlear implants in recipients. In addition to exo-AAV1 delivery into the cochlea, the authors show that the exo-AAV1 virus can induce transgene expression in the neighboring vestibular system of the inner ear. This outcome is not surprising, since the cochlea and the vestibular end organs are connected via a shared fluid space and the ability of viral vectors to flow from one compartment to the other in the ear has been shown previously.⁷

As proof of principal for its ability to improve inner-ear function, György et al.⁵ tested exo-AAV1 in a mouse model of human deafness caused by mutations in lipoma HMGIC fusion partner-like 5 (LHFPL5). They tested whether exo-AAV1 or exo-AAV9 efficiently targeted hair cells in mice engineered to lack the LHGPL5 ortholog, Lhfpl5.⁸ Lhfpl5^−/− mice exhibited no detectable LHFPL5 protein due to the insertion of a lacZ cassette that disrupted protein translation.⁸ Similar to Lhfpl5 mutant mice with a single bp missense mutation, Lhfpl5^−/− mice exhibited hearing loss and vestibular dysfunction, likely due to defects in hair cell structure and protein localization.⁸ Importantly, György et al.⁵ show that transduction of wild-type LHFPL5 into cochlear hair cells improves hearing thresholds (determined by recording auditory brainstem responses [ABRs]) and vestibular behaviors, suggesting potential for lasting recovery. The improvement in ABR thresholds achieved in Lhgpl5 mutant mice after exo-AAV viral infection was moderate, but significant, and could translate to important enhancement in quality of life for patients. For now, this study serves as an important proof of the principle that gene therapy can improve hearing and reduce balance deficits in affected mice.

The results also raise some important questions that should be addressed in future studies. First, it is unclear why the high efficiency of hair cell transfection did not yield better functional improvements. More detailed assessment of hair cell structure and function may reveal subtle defects that remain in the treated ear. Another potential problem needing consideration is cross-infection of exo-AAV1 to the vestibular organs. While potentially beneficial when a transgene is needed for therapy in both hearing and balance systems, as shown in the current paper, cross-transfection between the cochlea and the vestibular organs may also pose unwanted side effects when balance problems are not a component of the disease. Similarly, transduction of both hair cells and supporting cells in areas flanking the auditory epithelium (see Figure 1B in György et al.⁵ and Figure 1) could lead to undesired effects if genes that are critical for proper hair cell function are toxic to supporting cells.

Conversely, exo-AAVs might be useful for expressing genes to address dysfunction in supporting cells, as previously shown.⁹ Thus, accomplishing specific delivery to a sub-compartment of the inner ear (cochlea versus vestibular) or expression within a particular cell type is an important next step toward clinical application of gene therapy for treating inner-ear disease and may be influenced by specific regulatory elements that guide expression. Finally, the data reported in the paper were obtained in neonatal mice in which inner ears are immature, and it is unclear how relevant the results will be for more mature ears. This is important, because humans are born with mature inner ears and therapeutic access to developing inner ears would require complex embryonic interventions with low feasibility. Interestingly, other studies that have attempted viral-vector-mediated rescue of genetic deafness in mouse mutants have also reported positive outcomes only in developing ears.3, 4, 9

Exo-AAV1 as a delivery vehicle for gene replacement is relevant for treating loss-of-function alleles; however, it may be less useful for gain of function or missense sequence variants that result in novel protein activities, because these altered proteins may not respond to gene replacement strategies. To date, ClinVar and OMIM list the following four disease-causing homozygous pathogenic variants in LHFPL5: a 1-bp deletion (649delG), two missense variants (c.494C > T; p.Thr165Met and c.380A > G;p.Tyr127Cys), and a single nucleotide deletion (246delC). It is likely that additional pathogenic variants in LHFPL5 have yet to be uncovered. In summary, although much remains to be done, it is clear that the method and progress presented here are exciting and comprise important advances toward treatment of human hereditary deafness with gene-based therapies.