only to astrocytes in healthy tissue implies that reactive astrocytes are by definition ‘nonhomeostatic’, which can be misleading because disease-associated changes undergone by certain reactive astrocytes can exert homeostatic activities that preserve BBB function [64], tissue integrity [32], and neurological functions [65]. Third, there is already substantial evidence that there are multiple molecular expression patterns associated with reactive astrocytes in different disorder contexts rather than a single ‘disease-associated’ pattern [17, 23, 34, 50, 55–61, 63]. Caution is also urged with regards to assigning functional attributes to molecular ‘signatures’ purely on the basis of correlative association. For example, early transcriptome profiling studies identified an ‘A1’ molecular signature of 12 genes associated with a neurotoxic astrocyte subtype that emerged after exposure to specific cytokines secreted by microglia exposed to lipopolysaccharide (LPS), and an ‘A2’ molecular signature of 12 genes associated with a neuroprotective subtype after ischemic stroke [50, 66]. Notably, the assignment of these two proposed molecular signatures to two potential reactive astrocyte subtypes with different functions was based entirely correlative based on evidence, and the functions of all of these genes are not known; in addition, to date, no causation testing loss- or gain-of-function experiments have directly linked any of these
