The role played by certain domestic species such as dogs as a translational model in comparative oncology shows great interest to develop new therapeutic strategies in brain tumors. Gliomas are a therapeutic challenge that represents the most common form of malignant primary brain tumors in humans and the second most common form in dogs. Gene-directed enzyme/prodrug therapy using adipose mesenchymal stem cells (Ad-MSCs) expressing the herpes simplex virus thymidine kinase (TK) has proven to be a promising alternative in glioblastoma therapy, through its capacity to migrate and home to the tumor and delivering local cytotoxicity avoiding other systemic administration. In this study, we demonstrate the possibility for canine Ad-MSCs (cAd-MSCs) to be genetically engineered efficiently with a lentiviral vector to express TK (TK-cAd-MSCs) and in combination with ganciclovir (GCV) prodrug demonstrated its potential antitumor efficacy in vitro and in vivo in a mice model with the human glioblastoma cell line U87. TK-cAd-MSCs maintained cell proliferation, karyotype stability, and MSCs phenotype. Genetic modification significantly affects its secretory profile, both the analyzed soluble factors and exosomes. TK-cAd-MSCs showed a high secretory profile of some active antitumor immune response cytokines and a threefold increase in the amount of secreted exosomes, with changes in their protein cargo. We also found that the prodrug protein is not released directly into the culture medium by TK-cAd-MSCs. We believe that our work provides new perspectives for glioblastoma gene therapy in dogs and a better understanding of this therapy in view of its possible implantation in humans.