Background
It is generally accepted that cerebrospinal fluid (CSF) is produced in the ventricular system [
1‐
5] exclusively by the choroid plexus [
6‐
15]. Nonetheless, several studies have challenged the classical view of choroid plexus as the sole source of CSF. After surgical plexectomy of non-human primates, CSF production decreased by only 30%, and the composition of the remaining produced CSF did not differ from that in non-plexectomized rhesus monkeys [
16,
17]. Furthermore, the biochemical composition of CSF harvested from the exposed choroid plexus differs from bulk cisternal CSF, suggesting that extra-choroidal sources of CSF exist [
4]. Accordingly, studies have suggested that some fraction of CSF production must occur outside the ventricular system in mammals, specifically at the subarachnoid space (SAS) [
18‐
20].
Along with the pia mater, the SAS is a component of the leptomeninges that is filled with CSF, enclosing the brain and spinal cord (reviewed in [
21]). In an experimental paradigm in which dogs were perfused with artificial CSF at different pressures into the SAS, quantitative measures indicated that approximately 40% of the total CSF production occurred at this meningeal layer [
18‐
20]. However, that approach gives only a coarse indication of the cellular site and mechanisms underlying extra-choroidal CSF production.
Remarkably, the extensive capillary network present in the CNS has been postulated to contribute to the production of interstitial fluid (ISF), which ultimately mixes with the CSF [
4,
22‐
29]. Fluid filtration from the vascular compartment might occur in the brain capillary endothelia, which contain more mitochondria than the endothelial cells from non-neural tissues [
30]. The high content of mitochondria might support active transport of ions at the blood–brain barrier (BBB), which in turn triggers water fluxes, in a manner resembling the mechanism underlying CSF secretion by the choroid plexus. However, the hypothesis that the BBB is a source of CSF presently lacks substantiation from functional studies [
29].
Ion pumps, channels and co-transporters in the choroid plexus epithelia drive CSF secretion. Namely, Na
+ transport occurs via the Na
+/K
+ ATPase, and Cl
− transport occurs among others via NKCC1, establishing osmotic gradients that result in water movement across the blood-CSF barrier (BCSFB). This passive water movement is especially facilitated by water channels formed by the protein aquaporin 1 (AQP1), which is also expressed by epithelial cells in the choroid plexus (reviewed in [
13]). Thus, we asked whether the leptomeningeal vasculature express AQP1 and NKCC1, both proteins involved in CSF secretion by the choroid plexus [
11,
13,
31]. Indeed, we observed that both proteins are present in the vasculature distributed within the SAS in the adult rodent brain and spinal cord. The display of the molecular setup involved in CSF production suggest that leptomeningeal vessels may actively contribute to extra-choroidal CSF production throughout the adult rodent CNS.
Discussion
We applied the uDISCO technique [
43] to render the entire adult mouse head transparent, thus facilitating the brain-wide depiction of AQP1 immunoreactivity. This approach revealed the distribution of AQP1 in the vasculature of the intact brain, including the very fragile leptomeningeal vessels, which often become detached when the brain is removed from the skull (Fig.
1). AQP1 immunoreactive vascular profiles have been previously reported within the cerebral cortex of adult rats [
63]. The diameters of such AQP1
+ vessels corresponded mostly to arterioles, although some diameters were more consistent with capillaries. However, the AQP1 immunoreactivity was earlier described as occurring in discontinuous patches or zones along the vessels [
63]. Our use of tissue clearing coupled to light-sheet microscopy proved greatly advantageous for establishing that AQP1 has continuous distribution along the leptomeningeal vasculature. Besides, tissue processing for immunohistochemistry, which entails the use of fixatives and pre-treatments such as dehydration and freezing, can modify the structure of epitopes and the availability of antibody-binding sites. We thus speculate that a proper appreciation of the AQP1 distribution in the leptomeningeal vasculature of adult rodents was hindered due to disruptive effects of histological processing.
The water channels formed by AQP1 are distributed predominantly in the apical but also in the basal membranes of epithelial cells in the choroid plexus, which positions them for an important role in CSF production. AQP1 allow bi-directional water movement in response to osmotic gradients, which are generated by ion pumps, transporters and co-transporters also present in the choroid plexus epithelia ([
64]; reviewed in [
13]). Nevertheless, transgenic mice lacking AQP1 have shown only a 20% decrease in CSF production [
64] indicating that other proteins or mechanisms are involved in CSF production. Accordingly, the Na
+/K
+/2Cl
− cotransporter 1 (NKCC1) has been proposed as a main mediator of CSF production due to its capacity to couple water movement to ion translocation [
31]. Thus, we sought in this study to determine if the distribution of NKCC1 correlates with that of AQP1 in the leptomeningeal vasculature.
While there is clear documentation of NKCC1 expression by endothelial cells cultured in vitro [
59], the endothelia of arterioles and veins in the intact brain parenchyma is seemingly devoid of NKCC1 [
62]. However, the brain capillary endothelia have been shown to contain NKCC1, mostly located at its luminal membrane, playing a role in stroke-induced edema [
61] and in transendothelial ion uptake by the brain under ischemia [
60]. Our observations indicate that cells positive for both AQP1 and NKCC1 localize to the smooth muscle layer of arterioles and veins (Figs.
2,
3), specifically residing within the SAS. Moreover, AQP1 and NKCC1 immunoreactive cells were also observed in the endothelia of veins, venules and capillaries distributed within the SAS (Fig.
4). Remarkably, AQP1
+/NKCC1
+ endothelial cells were also present in the leptomeningeal vasculature of the spinal cord (Fig.
5). We therefore suggest that the co-distribution of AQP1 and NKCC1 in the smooth muscle cell layer and in the capillary and venular endothelia in the adult rodent brain and spinal cord is a unique feature of leptomeningeal vessels.
Several studies in different mammalian species, including non-human primates, have challenged the classical view of the choroid plexus as the sole CSF source [
4,
16‐
20]. Besides the ventricular system [
1,
2], the SAS has been postulated as an important secondary source of CSF, producing as much as 40% of the total volume [
18‐
20]. The results, based on measurements of CSF output, gave no insight into the molecular machinery supporting extra-choroidal CSF secretion. Our observations complement previous functional studies demonstrating that AQP1 and NKCC1, both proteins with acknowledged roles in CSF production by the choroid plexus (reviewed in [
13]), are not expressed by parenchymal capillaries, but only by the leptomeningeal vasculature distributed in the SAS, including a subset of penetrating arterioles and veins. In the smooth muscle cell layer of the vasculature, AQP1 and NKCC1 co-distribution might participate in the maintenance of smooth muscle contractility [
65], while also regulate transcellular transport of fluid [
66]. At the capillary level, NKCC1
+/AQP1
+ endothelial cells of the SAS vasculature may contribute to the generation of osmotic gradients and facilitation of water movement. Present results partially corroborate the hypothesis that the vast capillary network of the CNS not only subserves oxygen and nutrient supply, but also produces ISF, which is ultimately incorporated into the total circulating CSF [
4,
22‐
28]. Nevertheless, functional in vivo experiments with pharmacological or genetically-encoded alterations of AQP1 and NKCC1 expression and function, coupled to measurements of CSF production rate, are necessary to confirm that the leptomeningeal vasculature is an extra-choroidal source of CSF.
If proved correct, the existence of an extra-choroidal CSF source in the SAS vasculature may allow a comprehensive understanding of syndromes that relate to CSF production and circulation impairment, such as idiopathic intracranial hypertension (IIH) and idiopathic normal pressure hydrocephalus (iNPH), which are neurodegenerative diseases with non-determined causes [
67,
68]. In patients with IIH syndrome, increase in the expression level of perivascular AQP4 correlates with the degree of astrogliosis resulting in increased fluid turnover by mechanisms that remain to be determined [
69]. Alterations in fluid movement across the BCSFB along the SAS vasculature mediated by AQP1 and NKCC1, might relate to the changes in fluid turnover observed in this syndrome.
Acknowledgements
We would like to acknowledge M.Sc. Alba Redo Riveiro (Novo Nordisk Foundation Center for Stem Cell Biology, University of Copenhagen, Copenhagen, Denmark) for the aid with Western blotting imaging, and Pernille S. Froh (Department of Cellular and Molecular Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Denmark) for excellent technical immunohistochemical assistance. We would like to acknowledge Ph.D. Vikram Prasad and Ph.D. Gary E Shull (Department of Molecular Genetics, Biochemistry, and Microbiology, University of Cincinnati College of Medicine, Cincinnati, Ohio, USA) for kindly providing the NKCC1 knockout mice and Ph.D. Xiaowei Wang (Center for Translational Neuromedicine, Division of Glial Disease and Therapeutics, University of Rochester Medical Center, Rochester, New York, USA) for harvesting their brain tissue. We acknowledge Dan Xue for her aid on scheme graphic design. Also, we would like to thank Dr. Hajime Hirase for the valuable comments on the manuscript.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.