ECM adhesion and polarization to astrocyte endfeet requires M23–AQP4. (A) Wide-field and TIRF images of U87-MG cells stably transfected with M1– or M23–AQP4 and stained with antibody to AQP4 and for F-actin with phalloidin. Top panels shows untreated control cells; in bottom panels live cells were subject to detergent extraction by treatment with 0.5% Triton X-100 for 5 min before fixation. (B) Mean cellular AQP4 immunofluorescence staining intensity measured from TIRF images of M1- or M23-transfected cells under control conditions or with 0.5% Triton X-100 treatment before fixation. (n = 5 cells each, ±SE; n.s., P > 0.05; ***, P < 0.001 by t test). (C) TIRF images and ratio image of live U87-MG cells transfected with M1–AQP4–GFP and M23–AQP4–mCherry before and after incubation with 0.5% Triton X-100. (D) Quantification of M1/M23 ratio in transfected U87-MG cells before and after incubation with 0.5% Triton X-100; *, P < 0.05 by paired t test. (E) AQP4 distribution in astrocytes after intracerebral injection of M1– (top) or M23–AQP4 (bottom) adenoviruses into AQP4^−/− mice. Left panels show AQP4 distribution in GFAP-labeled astrocytes adjacent to cerebral blood vessels (dotted line), right panels show an enlarged view of AQP4 and GFAP staining in astrocyte end-feet (labeled “E”). (F) Quantification of the ratio of AQP4 staining density in GFAP-labeled processes adjacent to blood vessels versus nonadjacent processes for cells transfected with M1– or M23–AQP4; **, P < 0.01 by t test.