Severe coronavirus infections, including SARS-CoV-2, can cause neurological symptoms, but the underlying neurotropic mechanisms are unclear. Experiments with SARS-CoV-2 variants B.1.258.17, B.1.1.7, and BA.5.3.2 (termed wild-type, alpha and omicron, respectively) revealed that human astrocytes, not neurons, support viral proliferation. During the COVID-19 pandemic, new virus variants exhibited milder disease progression. A retrospective study of patients with COVID-19 infected by wild-type or alpha variants was conducted to test whether ketamine, an anaesthetic that inhibits endocytosis, affects COVID-19. At admission, patients infected with the wild-type showed greater disease severity than alpha variant patients, but the disease course was similar. This may be due to distinct ketamine-mediated SARS-CoV-2 variant-dependent effects, revealing stronger ketamine inhibition of the wild-type variant than the alpha variant mediating astrocyte responses involving the expression of ACE2, a viral cell entry site, viral proteins RNA-dependent RNA polymerase and envelope protein-E in infected cells. Overexpression of SARS-CoV-2 protein 3a attenuated astroglial lysosomal traffic, and 3a and nsp6 differentially modulated lipid droplet accumulation and initiation of autophagy, where ketamine predominantly affected vesicle dynamics. In summary, human astrocytes, but not neurons, contribute to SARS-CoV-2 neurotropism, highlighting the potential benefits of ketamine treatment in coronavirus infections.