Cx43 carboxyl terminal domain determines AQP4 and Cx30 endfoot organization and blood brain barrier permeability

Abstract

The neurovascular unit (NVU) consists of cells intrinsic to the vessel wall, the endothelial cells and pericytes, and astrocyte endfeet that surround the vessel but are separated from it by basement membrane. Endothelial cells are primarily responsible for creating and maintaining blood–brain-barrier (BBB) tightness, but astrocytes contribute to the barrier through paracrine signaling to the endothelial cells and by forming the glia limitans. Gap junctions (GJs) between astrocyte endfeet are composed of connexin 43 (Cx43) and Cx30, which form plaques between cells. GJ plaques formed of Cx43 do not diffuse laterally in the plasma membrane and thus potentially provide stable organizational features to the endfoot domain, whereas GJ plaques formed of other connexins and of Cx43 lacking a large portion of its cytoplasmic carboxyl terminus are quite mobile. In order to examine the organizational features that immobile GJs impose on the endfoot, we have used super-resolution confocal microscopy to map number and sizes of GJ plaques and aquaporin (AQP)-4 channel clusters in the perivascular endfeet of mice in which astrocyte GJs (Cx30, Cx43) were deleted or the carboxyl terminus of Cx43 was truncated. To determine if BBB integrity was compromised in these transgenic mice, we conducted perfusion studies under elevated hydrostatic pressure using horseradish peroxide as a molecular probe enabling detection of micro-hemorrhages in brain sections. These studies revealed that microhemorrhages were more numerous in mice lacking Cx43 or its carboxyl terminus. In perivascular domains of cerebral vessels, we found that density of Cx43 GJs was higher in the truncation mutant, while GJ size was smaller. Density of perivascular particles formed by AQP4 and its extended isoform AQP4ex was inversely related to the presence of full length Cx43, whereas the ratio of sizes of the particles of the AQP4ex isoform to total AQP4 was directly related to the presence of full length Cx43. Confocal analysis showed that Cx43 and Cx30 were substantially colocalized in astrocyte domains near vasculature of truncation mutant mice. These results showing altered distribution of some astrocyte nexus components (AQP4 and Cx30) in Cx43 null mice and in a truncation mutant, together with leakier cerebral vasculature, support the hypothesis that localization and mobility of gap junction proteins and their binding partners influences organization of astrocyte endfeet which in turn impacts BBB integrity of the NVU.