Astrocytes and neurons in the central nervous system (CNS) interact functionally to mediate processes as diverse as neuroprotection, neurogenesis and synaptogenesis. Moreover, the interaction can be homotypic, implying that astrocyte-derived secreted molecules affect their adjacent neurons optimally vs remote neurons. Astrocytes produce neurotrophic and extracellular matrix molecules that affect neuronal growth, development and survival, synaptic development, stabilization and functioning, and neurogenesis. This new knowledge offers the opportunity of developing astrocyte-derived, secreted proteins as a new class of therapeutics specifically to treat diseases of the CNS. However, primary astrocytes proliferate slowly in vitro, and when induced to immortalize by genetic manipulation, tend to lose their phenotype. These problems have limited the development of astrocytes as sources of potential drug candidates. We have successfully developed a method to induce spontaneous immortalization of astrocytes. Gene expression analysis, karyotyping and activity profiling data show that these spontaneously immortalized type-1 astrocyte cell lines retain the properties of their primary parents. The method is generic, such that cell lines can be prepared from any region of the CNS. To date, a library of 70 cell lines from four regions of the CNS: ventral mesencephalon, striatum, cerebral cortex and hippocampus, has been created. A phenotype-selective neurotrophic factor for dopaminergic neurons has been discovered from one of the cell lines (VMCL1). This mesencephalic astrocyte-derived neurotrophic factor (MANF) is a 20 kD, glycosylated, human secreted protein. Homologs of this protein have been identified in 16 other species including C. elegans. These new developments offer the opportunity of creating a library of astrocyte-derived molecules, and developing the ones with the best therapeutic indices for clinical use.