Perivascular endfeet of astrocytes are enriched with aquaporin-4 (AQP4)—a water channel that is critically involved in water transport at the brain–blood interface and that recently was identified as a key molecule in a system for waste clearance. The factors that determine the size of the perivascular AQP4 pool remain to be identified. Here we show that the size of this pool differs considerably between brain regions, roughly mirroring regional differences in Aqp4 mRNA copy numbers. We demonstrate that a targeted deletion of α-syntrophin—a member of the dystrophin complex responsible for AQP4 anchoring—removes a substantial and fairly constant proportion (79–94 %) of the perivascular AQP4 pool across the central nervous system (CNS). Quantitative immunogold analyses of AQP4 and α-syntrophin in perivascular membranes indicate that there is a fixed stoichiometry between these two molecules. Both molecules occur at higher densities in endfoot membrane domains facing pericytes than in endfoot membrane domains facing endothelial cells. Our data suggest that irrespective of region, endfoot targeting of α-syntrophin is the single most important factor determining the size of the perivascular AQP4 pool and hence the capacity for water transport at the brain–blood interface.

More than 90% of the cases of Parkinson’s disease have unknown etiology. Gradual loss of dopaminergic neurons of substantia nigra is the main cause of morbidity in this disease. External factors such as environmental toxins are believed to play a role in the cell loss, although the cause of the selective vulnerability of dopaminergic neurons remains unknown. We have previously shown that aquaglyceroporin AQP9 is expressed in dopaminergic neurons and astrocytes of rodent brain. AQP9 is permeable to a broad spectrum of substrates including purines, pyrimidines, and lactate, in addition to water and glycerol. Here we test our hypothesis that AQP9 serves as an influx route for exogenous toxins and, hence, may contribute to the selective vulnerability of nigral dopaminergic (tyrosine hydroxylase-positive) neurons. Using Xenopus oocytes injected with Aqp9 cRNA, we show that AQP9 is permeable to the parkinsonogenic toxin 1-methyl-4-phenylpyridinium (MPP+). Stable expression of AQP9 in HEK cells increases their vulnerability to MPP+ and to arsenite—another parkinsonogenic toxin. Conversely, targeted deletion of Aqp9 in mice protects nigral dopaminergic neurons against MPP+ toxicity. A protective effect of Aqp9 deletion was demonstrated in organotypic slice cultures of mouse midbrain exposed to MPP+ in vitro and in mice subjected to intrastriatal injections of MPP+ in vivo. Seven days after intrastriatal MPP+ injections, the population of tyrosine hydroxylase-positive cells in substantia nigra is reduced by 48% in Aqp9 knockout mice compared with 67% in WT littermates. Our results show that AQP9 –selectively expressed in catecholaminergic neurons—is permeable to MPP+ and suggest that this aquaglyceroporin contributes to the selective vulnerability of nigral dopaminergic neurons by providing an entry route for parkinsonogenic toxins. To our knowledge this is the first evidence implicating a toxin permeable membrane channel in the pathophysiology of Parkinson’s disease.

Aquaporin-4 (AQP4) is the predominant water channel in the brain and is expressed in high density in astrocytes. By fluxing water along osmotic gradients, AQP4 contributes to brain volume and ion homeostasis. Here we ask whether deletion of Aqp4 leads to upregulation of the gap junctional proteins connexin-43 (Cx43) and connexin-30 (Cx30). These molecules couple adjacent astrocytes to each other and allow water and ions to redistribute within the astrocyte syncytium. Immunogold analysis of parietal cortex and hippocampus showed that the number of gap junctions per capillary profile is increased in AQP4 knockout (AQP4 KO) mice. The most pronounced changes were observed for Cx43 in hippocampus where the number of connexin labeled gap junctions increased by 100% following AQP4 KO. Western blot analysis of whole tissue homogenates showed no change in the amount of Cx43 or Cx30 protein after AQP4 KO. However, AQP4 KO led to a significant increase in the amount of Cx43 in a Triton X-100 insoluble fraction. This fraction is associated with connexin assembly into gap junctional plaques in the plasma membrane. In line with our immunoblot data, RT-qPCR showed no significant increase in Cx43 and Cx30 mRNA levels after AQP4 KO. Our findings suggest that AQP4 KO leads to increased aggregation of Cx43 into gap junctions and provide a putative mechanistic basis for the enhanced tracer coupling in hippocampi of AQP4 KO mice. The increased number of gap junctions in AQP4 deficient mice may explain why Aqp4 deletion has rather modest effects on brain volume and K homeostasis.