Of the seven thousand diseases that are described as rare, 80% of them have an identified genetic cause. With this in the mind, the development of technologies, such as gene therapy, to address the genetic factors involved in these pathologies, might be a game-changing therapeutic approach. Viral delivery of transgenes for RNA interference, DNA editing, or protein overexpression may offer opportunities for novel treatments. In this thesis, we developed adeno-associated virus (AAV) vector systems to tackle two rare diseases: Amyotrophic lateral sclerosis (ALS) caused by gain-of-function mutations in the superoxide dismutase 1 (SOD1) gene and inherited hearing loss disorder caused by mutation in the transmembrane channel-like 1 (TMC1) gene. ALS is a fatal neurodegenerative disease caused by the progressive degeneration of motoneurons (MN). Glial cells have been shown to importantly contribute to the disease. Silencing of the human pathogenic mutated SOD1 protein (SOD1G93A) in MN and/or astrocytes by AAV-mediated expression of microRNA targeting SOD1 has been shown to provide therapeutic benefits in a mouse model which is overexpressing SOD1G93A. However, in order to have an effective treatment, it is important to understand the influence of the therapy on several relevant ALS cell types. In this work, we determined the effects of lowering SOD1G93A expression in astrocytes using an AAV serotype 9, in combination with an astrocyte-specific promoter, in order to express a microRNA targeting SOD1. In the treated mice, we observed a significant improvement of the neuromuscular function and a partial protection of the vulnerable fast-fatigable MN. SOD1 silencing also induced a significant re-innervation of the neuromuscular junctions (NMJ) in the gastrocnemius muscle, observed after the initial phase of denervation, occurring around the age of 60 days. Re-innervation was associated with a grouping of the muscle fibers from the same type, consistent with enhanced axonal plasticity in the treated SOD1G93A mice. Overall, we demonstrated that silencing of SOD1 in astrocytes has an impact on MN plasticity at the level of NMJ, which translates into improved neuromuscular function towards the end stage of the disease. Mutations in the TMC1 gene lead to either complete deafness at birth, or to progressive hearing loss beginning in childhood, depending on the mode of inheritance (autosomal recessive or dominant, respectively). TMC1 is likely a component of the mechanotransduction channel, which is responsible for the transformation of mechanical sound-induced stimuli into electrical signals. This protein has been found to be expressed in the stereociliae of hair cells. In the second part of the thesis, we report on the development of viral tools to deliver a functional copy of the TMC1 gene, or its related paralog TMC2, into hair cells. We engineered several AAV vectors, in combination with different promoters, to identify a suitable combination able to restore TMC expression in the mouse cochlear hair cells, both in vitro and in vivo. We demonstrated, in collaboration with the lab of J. Holt, that when injected in TMC-null mice at birth, these vectors could partially restore loss of auditory function in this mouse model otherwise completely deaf. Overall we showed that AAV-based gene therapy is a promising approach to treat rare genetic disease.