Engineering tools to specifically deliver genes in a restricted cell subpopulation in the central nervous system (CNS) is of great importance and is now the subject of active research. Such tools can be used for various applications, and the modelization of neurodegenerative diseases in a mouse model was discussed in this study.Neurofibrillary degeneration (NFD) result of the intraneuronal aggregation of insoluble fibrillar material composed of Tau proteins. This lesion is characteristic of a heterogeneous group of neurodegenerative diseases called tauopathies, some of which present a spatial and temporal progression of the DNF, affecting several neuronal populations connected by neuroanatomical connections in a hierarchical and stereotypical manner. Currently, the molecular and cellular mechanisms underlying the spatial and temporal progression of the DNF are not yet understood, partly due to the lack of relevant study models. In this context, my thesis work was to develop tools that can specifically and efficiently deliver genes to a neuronal subpopulation early affected by the DNF – the CA1 pyramidal neurons – with the ultimate goal of initiating a progressive Tau pathology and analyzing the mechanisms underlying the spatial and temporal evolution in a rat model. Different approaches exist to transfer a gene into a cell but viral technology still remains the most effective. In addition, it is now possible to change the properties of viral vectors in order to acquire them with a specificity of cell targeting. These new abilities, combined with the capacity to deliver locally these vectors into the CNS via stereotaxic injection, led us to develop two complementary approaches using pseudotyped lentiviral vectors or modified adenoviral vectors.Vesicular stomatitis virus G-pseudotyped Lentiviral vectors (LV), known to efficiently infect neurons in murine models, were injected locally into the CA1 layer of the pyramidal neurons of rats to mediate the expression of 4R Tau isoform. Overexpression of both wild-type or P301L mutated Tau protein initiated a progressive degenerative process in these neurons, and was characterized by the appearance of histopathological features and/or a memory deficit. The evolution of this neurodegeneration is slower with the wild-type form of Tau protein, consistent with the long prodromal phase observed in sporadic Tauopathies. Interestingly, a spatial evolution of histological lesions was observed over time, especially with the wild-type isoform of Tau protein, which propagated in distant brain areas but anatomically connected to the hippocampus.The use of a modified adenoviral vector HAdV-5-SRIF, targeting a membrane protein highly expressed in the somatodendritic compartment of the CA1 pyramidal neurons - the somatostatin receptor type 2 (sstr2) – led us to demonstrate that targeting neurons with adenoviral vectors was a feasible approach. Indeed, the parental adenoviral vectors used in several previous studies have been disappointing mainly because of the heterogeneous expression of primary receptor - the hCAR - in the brain. Our vector, containing the sstr2 sequence of recognition in the adenoviral fiber, is capable to induce the expression of a reporter gene in vitro and in primary neurons, with high specificity despite low viral doses used. Preliminary results following intra-hippocampal injection of this vector in a rat model indicate that this modified adenoviral vector is capable of delivering genes in the CA1 pyramidal neurons.The local injection of viral vectors that display neuronal tropism allows us to deliver genes in a specific subpopulation of interest, to initiate slow and progressive neurodegenerative process in rat model, especially with the wild-type Tau protein isoform.