Three‐plane description of astroglial architecture and gliovascular connections of area postrema in rat: Long tanycyte connections to other parts of brainstem

The study demonstrates the astroglial and gliovascular structures of the area postrema (AP) in three planes, and compares them to our former findings on the subfornical organ (SFO) and the organon vasculosum laminae terminalis (OVLT). The results revealed long glial processes interconnecting the AP with deeper areas of brain stem. The laminin and β‐dystroglycan immunolabeling altered along the vessels indicating alterations of the gliovascular relations. These and the distributions of glial markers displayed similarities to the SFO and OVLT. In every organ, there was a central area with vimentin‐ and nestin‐immunopositive glia, whereas GFAP and the water‐channel aquaporin 4 were found at the periphery. This separation supports different functions of the two regions. The presence of nestin may indicate stem cell capabilities, whereas aquaporin 4 has been suggested by other studies to be a possible participant of osmoperception. Numerous S100‐immunopositive glial cells were found approximately evenly distributed in both parts of the AP. Frequency of glutamine synthetase‐immunoreactive cells was similar in the surrounding brain tissue in contrast to that found in the OVLT and SFO. Our findings on the three sensory circumventricular organs (AP, OVLT, and SFO) are compared in parallel.


INTRODUCTION
The aim of the study was to describe the glial architecture and the gliovascular connections of the area postrema (AP) in three planes and fourth ventricle (rostrally), and the brain substance (basolaterally and posteriorly). The attachment of the choroid plexus separates the first two. The AP consists of dorsal mantle, central, ventral junctional, and lateral zones based on cytological and neurochemical properties and neural connections (Gross, 1991;used also in further studies, Furube et al., 2015;Jeong et al., 2021;McKinley et al., 2003;Morita et al., 2016;Price et al., 2008). A glial border zone, the funiculus separans, separates the AP from the subpostremal area of the nucleus of the solitary tract (NTS; Barraco et al., 1992;Dallaporta et al., 2010;McKinley et al., 2003).
The AP is a "sensory" circumventricular organ (CVO) like the SFO and OVLT; it participates in the control of several processes: salt and volume homeostasis and blood pressure, but in the case of AP, mainly the feeding. As a sensory CVO, it monitors the serum levels of humoral factors in the feedback mechanisms controlling these processes (for reviews, see Dallaporta et al., 2010;Jeong et al., 2021;Kaur & Ling, 2017;MacDonald et al., 2020;McKinley et al., 2003;Price et al., 2008;Sisó et al., 2010;Smith & Ferguson, 2010;Troadec et al., 2022). Besides neurons, astroglia also participates in receptor functions (Dallaporta et al., 2010;MacDonald et al., 2020;Troadec et al., 2022).
The vessels invade the organ from its dorsocaudal aspect and form a primary system of wide "sinusoids." At the AP-NTS, border this system coalesces into short connecting capillaries of regular size, which break up again into a second capillary plexus of NTS, forming a "portal" circulation (Dempsey, 1973;Roth & Yamamoto, 1968).
Monitoring requires free access to the cerebral blood, that is, "leaky" vessels with a less restrictive blood-brain barrier (BBB). Fenestrated capillaries with wide perivascular spaces and permeability were described by Dempsey (1973), Krisch et al. (1978), Gross (1991), and Willis et al. (2007). Willis et al. (2007) and recently Miyata (2015) and Morita et al. (2016) proved by fluorescent tracers and immunohistochemistry that the permeability belongs to the lamininimmunopositive vessels of the central zone of AP.
Besides the humoral input, the AP has reciprocal neural connections with NTS, receives neural input from visceral vagal afferents as a part of the so-called dorsal vagal complex, and has connections with hypothalamic nuclei (Jeong et al., 2021;McKinley et al., 2003;Price et al., 2008;Sisó et al., 2010). The anterior cingulate cortex and insula receive neural inputs from the AP regulating the cortical motivation of feeding (McKinley et al., 2019).
Laminin and β-dystroglycan were also investigated along the cerebral vessels, since they may be indirect indicators of the gliovascu-lar connections. Where cerebrovascular laminin immunopositivity is detectable, it indicates a separation of the vascular and glial basal laminae, that is, a loosening of the gliovascular coupling (Krum et al., 1991).

Animals
Sixteen adult rats (Wistar) of either sex, weighing 250-300 g, were used. The animals were supplied with food (rat food from Charles River) and water ad libitum and were kept on a 12/12-h light-dark cycle.

Fixation and sectioning
The animals were deeply anesthetized with ketamine and xylazine (20 and 80 mg/kg, respectively, IM), and perfused through the aorta with 100 mL 0.9% sodium chloride followed by 300 mL 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). After perfusion, brains were removed and were postfixed in the same fixative for 1 day at 4 • C.
Tissue blocks containing the AP were embedded into agarose, and serial sections (thickness 50 m, three to four sections per animal) were cut by a vibration microtome (Leica VT 1000S) in the coronal (11 animals), horizontal (two animals), or sagittal plane (three animals).

Immunofluorescent reactions
The method was applied as in our previous studies (e.g., Pócsai & Kálmán, 2015). Briefly, floating sections were pretreated with 20% normal goat serum diluted in PBS (phosphate buffered saline; Sigma) for 90 min to block the nonspecific binding of antibodies.
This and the following steps were followed by an intensive wash and stirring in abundant PBS (30 min, at room temperature). Primary immunoreagents were diluted as shown in Table 1 in PBS containing 0.5% Triton X-100, and 0.01% sodium azide was added as a conserving agent. Sections were incubated for 40 h at 4 • C. Fluorescent secondary antibodies (Table 2) were used at room temperature for 3 h. Then, the sections were washed in PBS (1 h, at room temperature), mounted onto microscope slides, cover-slipped in a mixture of glycerol and double distilled water (1:1), and finally sealed with lacquer. As negative controls, sections were treated identically, but the primary antibodies were omitted. No structure-bound fluorescent labeling was observed in these specimens. For double-labeling reactions, the two antibodies were applied in parallel; otherwise, the protocol was as described above.

Confocal laser scanning microscopy and digital imaging
Slides were photographed with a Radiance-2100 (BioRad, Hercules, CA, USA) confocal laser scanning microscope. On the photomicrographs, the green and red colors correspond to the emitted colors of the fluorescent dyes as shown in Table 2. Digital images were processed using Photoshop 9.2 software (Adobe Systems, Mountain View, CA, USA; RRID:SCR_014199) with minimal adjustments for brightness and contrast.

Semithin sections
The brain pieces with AP were cut from the vibratome sections under a light microscope. They were immersed for 30 min into a 1% osmium tetroxide solution in phosphate buffer (0.1 M, pH 7.4), rinsed in phosphate buffer, and then dehydrated through a graded series up to absolute ethanol. Following immersion in propylene oxide (10 min), the tissue pieces were embedded into epoxy resin (Durcupan, Fluka).
Semithin sections were cut with a Reichert Ultracut S ultramicrotome.

2.6
Pre-embedding electron microscopical immunohistochemistry to localize laminin To make it simple, this reagent is incorporated here despite not being a secondary antibody.
in PBS for 5 min at room temperature to inactivate endogenous peroxidase. Triton X-100 detergent was reduced to 0.1% to decrease the tissue destruction; otherwise, incubations with normal serum and primary antiserum were as described above. Then, the procedure continued by applying a biotinylated antibody followed by the avidinbiotinylated peroxidase complex (see Table 2). Both incubations lasted for 90 min at room temperature and were followed by PBS (30 min, at room temperature). To visualize the immunohistochemical reaction product, 0.05% 3,3′ diaminobenzidine-tetrahydrochloride (DAB) and 0.01% H 2 O 2 in Tris-HCl buffer (0.05 M, pH 7.4, at room temperature) were used. The peroxidase reaction was stopped under the visual control of color, replacing the solution by changes of PBS.
Following the immunoreaction, the sections were treated as described in Section 2.5 to get semithin sections, then having selected proper areas ultrathin sections were cut with the same ultramicrotome. The photomicrographs were taken with a JEOL 100B electron microscope equipped with a Sys Morada digital camera.

The localizations of laminin and β-dystroglycan
The laminin-immunoreactive vessels marked conspicuously the territory of the AP. In the surrounding cerebral tissue, no vessels were immunopositive for laminin ( Figure 1a). The laminin-immunopositive vessels were surrounded by "outer walls," which were also laminin immunopositive, so they seemed to be "double walled." These "outer walls" were irregular and frequently surrounded two to three vessels.
Around the deeper vessels, perivascular spaces were not visible; their walls were single. The pial surface was also immunopositive to laminin.
The β-dystroglycan immunostaining (Figure 1b) labeled vessels both in the AP and around it throughout the brain. The lumina inside the AP were much wider than outside of it, but the two vessel systems were continuous with each other.
Combining the two labelings ( Figure 1c,d), in the "outer wall," yellow color indicated the close localization of laminin (green) and β-dystroglycan (red) immunoreactivities, whereas the "inner wall" remained free of β-dystroglycan immunoreactivity. Outside of the AP in the surrounding brain substance, the vessels were "single walled" and they were immunopositive only to β-dystroglycan but not to laminin. "Single-walled" but yellow vessels interconnected the two systems. The pial surface was also immunopositive to both laminin and β-dystroglycan. The extracerebral vessels, that is, pial vessels, however, were immunopositive only to laminin.
In the semithin sections (Figure 1g,h), it was visible that in the AP one or more vessels took place in every cavity; the perivascular spaces were wide around them. In the surrounding brain tissue, there were no similar perivascular spaces. Immunoelectron microscopy (Figure 1i The immunoreactive vessels sharply delineate the territory of the AP. Most vessels seem to have "double walls." Both the outer walls (arrowheads) and the inner walls (arrows) are laminin immunopositive (green, see also left side inset). The outer walls can surround two or three vessels (see right side inset). The pial surface is also immunopositive (double arrowhead). NTS: the subpostremal part of the nucleus of the solitary tract. Marks point to identical details in panels (a-c). Scale bar: 60 μm; for the insets: 15 μm. (b) The immunolabeling of β-dystroglycan in the same territory. Within the AP, the lumina are wider than outside of it, but the two vessel systems are continuous with each other. The pial surface is also immunopositive (double arrowhead). Scale bar: 60 μm. (c) Double labeling against laminin (green) and β-dystroglycan (red). Laminin immunopositivity is only found around vessels in the AP. Where the vessels have "double walls," in the outer wall the yellow color indicates the close localization of laminin and β-dystroglycan. Outside of the AP, the vessels are "single walled" and immunopositive only to β-dystroglycan. "Single-walled" but yellow vessels (colocalization) interconnect the two systems (broken arrows). The extracerebral (pial) vessels are immunopositive only to laminin (double arrow). Scale bar: 60 μm. (d) Enlarged part of a territory similar to that seen in panel (c) with similar labeling. Arrow with a circle: a vessel entering from the pia, immunopositive only to laminin. Scale bar: 30 μm. (e) Z-stack to panel (c); for scale, see that panel. (f) Z-stack to panel (d); for scale, see that panel. (g) Right half of the AP; semithin section, toluidine blue. Marking is similar to that in panels (a-c). One or more vessels (arrows) take place in wide cavities ("outer wall," arrowheads) with wide perivascular spaces. Outside of the AP, (Continues) F I G U R E 1 (Continued) in the surrounding brain tissue there are no similar spaces. Scale bar: 75 μm. (h) Enlarged detail of a semithin section similar to that seen in the previous panel. Scale bar: 15 μm. (i) Electron photomicrograph on a vessel and the surrounding cavity. Immunoelectron microscopy against laminin. Arrow: laminin-immunopositive lamina basalis around the vessel. Arrowhead: laminin-immunopositive lining of its cavity. E: endothelium; C: cell in the perivascular space. Scale bar: 5 μm. (j) Electron photomicrograph on a vessel. Here, the perivascular space almost disappeared; the two basal laminae (arrow, arrowhead) are only separate at a small segment. Note that this vessel is narrower than that seen in panel (i), since its scale bar is 2 μm. Abbreviations indicate on the images what substances are labeled. The color of the abbreviation corresponds to the color of labeling (green colors: polyclonal reagents [see Table 1]; red colors: monoclonal reagents [see Table 1]. Dys, β-dystroglycan; Lam, laminin. was easier to follow along the vessels (Figure 2e-i). The pial surface was immunopositive to both laminin and β-dystroglycan. Where the vessels entered the brain substance, it was visible that their "outer wall" was a continuation of the pial surface. Where the vascular wall became single, it was first yellow, then it continued in red where the vessels left the AP.
Some vessels, mainly in their entrance at the dorsorostral part of AP, displayed very weak immunoreactivity to β-dystroglycan, which was masked by the laminin immunostaining at double labeling, and was detectable when the β-dystroglycan immunoreactivity was only on display. The small, red β-dystroglycan-immunopositive globules, which accompanied the ependymal surface throughout the ventricular system ( Figure 2c, see also inset), were missing corresponding to the AP.

The distribution of GFAP and vimentin in frontal sections
For different double labelings, two anti-GFAP (produced in mouse and rabbit) and two anti-vimentin reagents (produced in mouse and chicken) were applied (see Table 1). It made possible their combinations with reagents produced either in mouse (against β-dystroglycan, nestin, appearing in red in the pictures) or in rabbit (against aquaporin 4, laminin, S100, appearing in green). The origin of the reagent applied actually is mentioned in the legends.
The GFAP immunopositivity is confined mainly to the lateral border zones (Figure 3a-d). On both sides, the border zones coursed ventraland medialward forming together a letter of "V." Here, the GFAPimmunopositive elements were mainly small, round, oval, or irregular profiles, which referred to their being cross or oblique sections of processes. In contrast, in the surrounding brain substance, astrocytes were numerous. The shape of the AP changed going caudalward. In the beginning, its dorsal surface was moderate convex, and ventrally the bilateral border zone extended to the central canal. More caudally its dorsal surface "sunk," and got a more convex shape. Here, the border zones fused in the midline and extended to the central canal forming a letter of "Y" above it ( Figure 3c). The border zone was passed laterally by long glial processes, which interconnected the AP and NTS In the former one, the vimentin immunopositivity did not surround the vessels, and its colocalization with β-dystroglycan was scarce. In contrast, the GFAP immunopositivity surrounded the vessels and colocalized with the β-dystroglycan.

GFAP and vimentin in sagittal and horizontal sections
In sagittal sections, the central part of the AP was poor in vimentin immunoreactivity; its ependyma was actually devoid of it, although on the other parts of the ventricular system, the ependyma was intensely immunopositive ( Figure 5a). From the ventricle and the central canal, thin and long vimentin-immunopositive processes coursed up-and caudalward; the ependymal origins were recognized at few processes, so they showed the characteristics of tanycytes. This population of processes formed the lower border of AP; they delineate a rostrocaudal slope. The GFAP immunolabeling visualized a system of processes similar to that seen in the case of vimentin, but in part, it origi- In horizontal sections, the immunostaining against vimentin revealed straight processes from the ventricle into the brain stem ( Figure 6a,b). Their length was about 800-1000 μm, so they extended beyond the AP. These processes were also immunopositive to nestin (not shown). The straight processes were positioned laterally and flanked a territory with an unarranged glial population in the middle.
GFAP staining ( Figure 6c) visualized a similar situation. The size and shape of the flanked territory changed at the different section levels.

F I G U R E 2
The localization of laminin and β-dystroglycan in sagittal sections. In sagittal section, the position of the AP above the central canal is well visible. The choroid plexus joins its rostrodorsal edge. (a) β-Dystroglycan (red) and laminin (green), double labeling. The laminin immunoreactive vessels delineate the territory of AP. Outside of it, the vessels are immunopositive only to β-dystroglycan. The vessels interconnecting the two systems are yellow (colocalization, broken arrows). Double arrowheads: pial surface, immunopositive to both laminin and β-dystroglycan; large arrowhead: the attachment of choroid plexus-the vessels are pure green here; 4V: fourth ventricle. Scale bar: 150 μm. (b) When only the β-dystroglycan immunoreactivity is on display, its weakening can be followed toward the attachment of choroid plexus-the immunopositivity of vessels is hardly visible here. Marks are as in panel (a). Scale bar: 150 μm. (c) β-Dystroglycan (red) and laminin (green), double labeling. The vessels interconnecting the two systems are yellow (colocalization); see the rectangle area enlarged in panels (e), (g), and (h). The ependymal surface (dashed line) is free of immunolabeling, but is accompanied by small β-dystroglycan-immunopositive globules a little below the surface (enlarged in the upper inset; the position of the inset is marked with "+" in the main panel). These globules are missing corresponding to the surface of the AP. Double arrow: an extracerebral vessel; double arrowheads: pial surface, immunopositive to both laminin and β-dystroglycan; 4V: fourth ventricle; asterisk marks the same place as in panel (d) The color of the abbreviation corresponds to the color of labeling (green colors: polyclonal reagents [see Table 1]; red colors: monoclonal reagents [see Table 1]. Dys, β-dystroglycan; Lam, laminin.

Distribution of aquaporin 4 and its colocalizations
The aquaporin 4 distribution was rather even in the cerebral tissue surrounding the AP, but it was confined to vessels in the central part of the AP (Figure 7a). Here, the vascular localization was not continuous, in contrast to the vessels throughout the brain substance (Figure 7b,c).
Double immunolabeling for aquaporin 4 and β-dystroglycan (Figure 7d) showed colocalization along vessels (yellow color). Like β-dystroglycan, aquaporin 4 was only found in the "outer wall" but not in the "inner" one. On the pial surface, there was also colocalization with β-dystroglycan.

Other markers
The distribution of nestin immunopositivity was similar to that of vimentin (Figure 8a

Separation of GFAP-and vimentin-containing glial populations
The glial markers distinguish two areas: a central area and a border zone (Figure 10a). Comparing it to our former studies, in the AP these GFAP-and vimentin-predominated areas are not so separate as it was found in the SFO (Pócsai & Kálmán, 2015) and OVLT (Kálmán et al., 2019). In the AP, both components occur in the lateral border zone, whereas in the SFO and OVLT only narrow strips along the borderlines of the peripheral and central areas displayed immunopositivity to both GFAP and vimentin. In the OVLT this characteristic was shown by the rostroventral part adjacent to the optic chiasm (Prager-Khoutorsky & Bourque, 2015), whereas in the SFO by the ventromedially located glial population (Hicks et al., 2021) . In the AP, the junctional ventral zone (at Several authors mentioned the rich GFAP immunopositivity in the lateral zone (Dallaporta et al., 2010;Maolood & Meister, 2009;Pecchi et al., 2007;Wang et al., 2008;Willis et al., 2007). Pecchi et al. (2007) found that there is a population of "radial glia-like cells" in contrast to the common astrocytes. The cells were immunopositive to GFAP and vimentin and resembled tanycytes, but this group (Dallaporta et al., 2010;Troadec et al., 2022) coined the name "vagliocytes" referring to the dorsal vagal complex. Maolood and Meister (2009) detected the tanycyte marker DARPP-32 in them.
The characteristics of glial processes depend on their cytoskeletal composition, for example, the content and proportion of GFAP and/or vimentin (Galou et al., 1997;Menet et al., 2001); the former may provide firmness, whereas the latter was described in motile cells (Dahl et al., 1981) and supposed to give flexibility.  suggested that vimentin promotes intra-and transcellular transport capability. Menet et al. (2001) demonstrated that the presence of GFAP influences the composition of the extracellular matrix (ECM) and the cell adhesion molecules of astrocytes. Nehmé et al.
(2012) described a Na(x) channel, which occurs in vimentin-containing glial processes but lacks in GFAP-containing ones. In contrast, Miller and Loewy (2013) found that the epithelial sodium channel (ENaC) gamma-subunit only occurs in GFAP-immunopositive astrocytes in the border zones of CVOs. Pekny (2001) found that in the perivascular astrocytes, GFAP is necessary to support BBB formation. GFAP colocalizes with aquaporin 4, whereas vimentin does it with nestin (see later).

Limiting diffusion of blood-borne substances
It was suggested that the astroglia in the border zone forms a barrier limiting the diffusion of blood-borne substances from the AP into the surrounding tissue (Langlet et al., 2013;Maolood & Meister, 2009;Miyata, 2015;Morita et al., 2016;Wang et al., 2008;Willis et al., 2007). The immunoreactivities of the tight junction proteins claudin F I G U R E 3 The distribution of GFAP and vimentin (Vim) in the AP in frontal sections. (a) The GFAP immunopositivity is confined mainly to the border zone (double arrowhead); the original color was red (monoclonal anti-GFAP). The glial elements forming the border zone are mainly small, round, oval, or irregular profiles, that is, cross-sections of processes (inset). The GFAP immunopositivity is dense outside of the AP and along its border but not within its central part. Here, at its anterior part, the AP extends to the central canal (CC). NTS: the subpostremal part of the nucleus of the solitary tract. Scale bar: 120 μm; for the inset: 40 μm. (b, c) The shape of the AP changes going caudalward; the "V" arrangement of the border zones changes to a "Y" extending (double arrow) to the central canal (CC), which does not contact the AP here. Laterally, the border zone is passed by long processes interconnecting the AP and NTS (arrows). The original color was green (polyclonal anti-GFAP). Scale bars: 120 and 60 μm. and ZO-1 were found to be intense between the tanycyte processes here (Maolood & Meister, 2009;Miyata, 2015;Morita et al., 2016;Wang et al., 2008;Willis et al., 2007).
In our former studies, the binding of Wisteria floribunda agglutinin and immunostaining of tenascin marked the territory of AP, whereas brevican and neuracan delineated its border (Pócsai & Kálmán, 2014a).
The ECM is important in the absorption of soluble factors, and in water binding (Aumailley & Gayraud, 1998;Dow &Wang, 1998;Franco & Müller, 2011). In general, in the brain the extracellular spaces are narrow. The rich ECM provides relatively large intercellular spaces for the diffusion of blood-borne substances (Syková et al., 2005). On the other hand, at the border of the AP (in general, the CVOs) the decrease of ECM components, that is, the narrowing of intercellular spaces and the appearance of other ECM components may limit the diffusion of The glial processes only extend to the "outer walls" but never to the "inner" ones (faintly seen, arrows). Scale bar: 30 μm. (c) Enlarged part of the border zone of a specimen double labeled against vimentin (red; the monoclonal anti-vimentin) and laminin (green). The glial processes only extend to the "outer walls" (arrowheads) but never to the "inner" ones (faintly seen, arrows). Scale bar: 30 μm.
F I G U R E 4 (Continued) which is surrounded with vimentin-immunopositive ependyma. Scale bar: 60 μm. (h) Double labeling against GFAP (green; the polyclonal anti-GFAP) and β-dystroglycan (red) in a detail of the border zone. Note that GFAP colocalizes with β-dystroglycan (yellow color), and these double-labeled processes surround the vessels. Scale bar: 120 μm. Abbreviations indicate on the images what substances are labeled. The color of the abbreviation corresponds to the color of labeling (green colors: polyclonal reagents [see Table 1]; red colors: monoclonal reagents [see Table 1]. Dys, β-dystroglycan; Lam, laminin; Vim, vimentin. blood-borne substances to the surrounding brain tissue (Zamecnik et al., 2012).
About the vasculogenesis in CVOs, see the end of Section 4.7.

Glial connections
In the OVLT, we described long and short glial connections, which interconnected its different subdivisions with different hypothalamic areas.
Similarly, the AP has longer tanycyte connections to the caudal brain stem; actually, these processes form the main part of the border zone of the AP, and their cross-sections are visible in the frontal sections. These long processes course longitudinally, not radially, that is, not toward the pial surface but along the brain stem. with the AP, mainly with its vessels, represent a separate population, but may also have transport functions. Note that these shorter "borderpassing" processes display only GFAP immunopositivity, whereas in the long processes coursing in the border zone along the brain stem GFAP and vimentin colocalize.

Significance of immunoreactivities of laminin, β-dystroglycan, and utrophin
In general, the cerebral vessels do not show laminin immunoreactivity. To explain the phenomenon, we accept the opinion of Krum et al. (1991): as two basal laminae, an astroglial and an endothelial, fuse into a common gliovascular basal lamina around the vessels, it makes the laminin epitopes inaccessible for immunoreagents. Laminin immunopositivity, therefore, may indicate that the two basal laminae are not completely fused, although the gap between them may be undetectable at the level of light microscopy. In accordance with this, the laminin immunopositivity is detectable in immature vessels, and following lesions (Krum et al., 1991;Shigematsu et al., 1989;Sixt et al., 2001;Szabó & Kálmán, 2004, 2008, when the basal laminae separate. In other CVOs, the vessels are also immunopositive to laminin (see Kálmán et al., 2019;Krum et al., 1991;Morita et al., 2016;Pócsai & Kálmán, 2015).
The distribution of laminin immunopositivity was similar to the distribution of the "leaky" BBB as described by Miyata (2015) and Morita et al. (2016). The separation of the glial and vascular basal laminae and the "leaky" vessels mutually depend on each other: a minimal perivascular space is required to receive the product of extravasationand this product of extravasation inhibits the fusion of basal laminae. Bouchaud et al. (1989) as well as Gross (1991) supposed that the wide perivascular gap, i.e. the distance from vessel weakens the ability of perivascular astrocytes to induce the formation of BBB.
Immunostaining of β-dystroglycan accompanied the cerebral vessels but not the extracerebral ones. The β-dystroglycanimmunopositive contours indicated the presence of vascular glial endfeet even where vimentin immunopositivity did not visualize them (Guadagno & Moukhles, 2004;Tham et al., 2016;Tian et al., 1996;Wolburg et al., 2009). Along the pial surface, the laminin immunopositivity is attributed to the subpial basal lamina, whereas the β-dystroglycan immunopositivity is due to the subpial glial endfeet (Szabó & Kálmán, 2008). The β-dystroglycan-immunopositive "globuli" along the ependymal surface proved to be the "basal labyrinth" of ependyma (described by Leonhard, 1970)   Our former studies (Pócsai & Kálmán, 2014b) suggested that the detectability of utrophin immunoreactivity is parallel with that of laminin in the SFO and OVLT as well as in the AP. The utrophin, however, is localized intracellularly; therefore, the fusion/separation of basal laminae does not influence access to immunoreagents. According to Haenggi et al. (2004): ". . . data suggest the presence of an alternative, yet unidentified, anchoring protein" between utrophin and laminin-it could modify the conformation of utrophin. As a membrane cytoskeletal protein, utrophin could play an important role in the mechanical stress of endothelial cells (Matsumura et al., 1993). Via a
Entering the AP, first the vessels become "double walled": the accompanying "outer wall" is continuous with the pial surface, whereas the perivascular space is continuous with the subarachnoidal space (see also Krisch et al., 1978;McKinley et al., 2003).
Both the outer and inner walls are immunopositive to laminin corresponding to the basal laminae of the pial surface and the vascular endothelium, respectively. The β-dystroglycan immunoreactivity along "the outer wall" indicates perivascular glial endfeet (Haenggi et al., 2004;Tian et al., 1996); the close localization of laminin and β-dystroglycan may result in yellow color. Toward the NTS, the perivascular spaces become obliterated; the vessels are "single walled." The yellow color indicates a transitory stage with an incomplete fusion of the pial and vascular basal laminae; although perivascular space is not detectable at the light microscopic level, the laminin immunoreactivity has not been masked completely yet: its green manifestation turns to yellow as it is mixed with the red color of the β-dystroglycan immunoreactivity.
Where the wall is red, the close gliovascular contact, the fusion of their basal laminae "hides" the laminin epitopes (Krum et al., 1991), the capillaries have fused outer and inner basal laminae. This fusion of basal laminae was found in the NTS by the electron microscopic investigations of Dempsey (1973) and Willis et al. (2007). In the SFO and OVLT, similar immunohistochemical alterations were observed (Kálmán et al., 2019;Pócsai & Kálmán, 2015).
In the electron microscopical study of the SFO, Gross (1991) described three types of capillaries. They are in accordance with the F I G U R E 9 Other markers: (a) Double immunolabeling against utrophin (red) and laminin (green), an enlarged detail of a frontal section. Utrophin labels the "inner walls" (vessels, arrows) within the laminin-lined "outer walls" (arrowheads). Since the vessels have also laminin, their walls in some places are yellowish. Double arrowheads: pial surface. Scale bar: 30 μm. (b) Glutamine synthetase immunolabeling; the immunopositive elements gather at vessels (arrowheads); for larger magnification, see inset. Dotted line: the border of the AP; double arrowheads: pial surface, NTS: the subpostremal part of the nucleus of the solitary tract. Scale bar: 60 μm; for the inset: 10 μm. (c) S100 immunolabeling; it marks the whole territory of the AP (arrows: cells). Inset: S100-immunopositive cells enlarged. Arrowheads: border glia, double arrowheads: pial surface, S100-immunopositive glial endfeet. Scale bar: 60 μm; for the inset: 15 μm. (d) S100 (green, arrows: cells) and NeuN (red), double labeling. No colocalization is visible; the S100 is not localized in neurons. Arrowheads: border glia, double arrowheads: pial surface, S100-immunopositive glial endfeet; asterisk: the position of the inset. Scale: 50 μm; for the inset: 20 μm. (e) GFAP (red; monoclonal anti-GFAP) and S100 (green) double immunolabeling. Colocalization in the border zone (arrowheads); see enlarged in the inset. Scale bar: 60 μm; for the inset: 10 μm. (f) Vimentin (red; (Continues) F I G U R E 9 (Continued) monoclonal anti-vimentin) and S100 (green) double immunolabeling. Colocalization is found in the border zone (arrowheads); see enlarged in the inset. Scale bar: 60 μm; for the inset: 10 μm. Abbreviations indicate on the images what substances are labeled. The color of the abbreviation corresponds to the color of labeling (green colors: polyclonal reagents [see Table 1]; red colors: monoclonal reagents [see Table 1]. Utr, utrophin; GluSy, glutamine synthetase; Lam, laminin; Vim, vimentin. b-d types (Figure 10b) also found by us in the AP as well as in the SFO (Pócsai & Kálmán, 2015) and OVLT (Kálmán et al., 2019). Gross (1991) classified the typical cerebral vessels as type II ("d" in our study).
His type III had wide perivascular spaces (like "b" in our study) with fenestrated endothelium. Other studies found that this type of vessel showed immunopositivity to the MECA32 marker of lacking BBB (Langlet et al., 2013) but not to tight junction components (Willis et al., 2007), and proved to be permeable to silver microgranules (Dempsey, 1973). Gross' (1991) type I was the transitory form between III and II and therefore may correspond to the "c" type in the present study ( Figure 10b). (Note: Our "type a" is the vessel just entering the extracerebral segment (see Figure 10b). This type was not classified by Gross [1991].) In the SFO, Dellmann (1998) also distinguished three types of capillaries: one has large perivascular space and the other two have not, but one of them has a fenestrated "leaky" endothelium.

Differences in the perivascular glia of the different vascular segments
There are differences between the vessels of the central area (which are "leaky" and immunopositive to laminin according to Miyata, 2015;Morita et al., 2016) and the vessels of the border zone, which show the structure of vessels throughout the brain (Gross, 1991). In the former, the perivascular glia is immunopositive to vimentin but not to GFAP, and the vimentin immunopositivity does not surround the vessels (see also Willis et al., 2007). Fine astrocyte branches, and even whole astrocytes, can be free of GFAP (Connor & Berkowitz, 1985;Linser, 1985).
The wide perivascular spaces keep the perivascular endfeet at a distance from the "leaky" vessels (for a correlation between leakiness and perivascular gap, see Section 4.5). The lack of direct contact may inhibit the effect of astrocytes on the formation of the BBB (Bouchaud et al., 1989;Janzer & Raff, 1987). The lack of GFAP in these cells may also be disadvantageous for BBB formation (Pekny, 2001;Willis et al., 2007).
The presence of vimentin and the absence of GFAP characterize the immature glia (Pixley & deVellis, 1984). Around the vessels of the central zone, laminin immunopositivity is detectable, which is characteristic of immature vessels in contrast to mature ones (Krum et al., 1991). Willis et al. (2007) mentioned that these vessels "resemble the appearance of the basement membranes during vascular development." It may correlate with the finding of Morita et al. (2015) and Furube et al. (2015) that there is a dynamic vascular plasticity, a permanent VEGF-dependent angiogenic activity, which is associated with the states of the fenestrated features and leakiness of the vessels.

Aquaporin 4: Correlation with GFAP and a possible role in the osmometric function
The presence of aquaporin 4 in the AP we published formerly (Goren et al., 2006). The present study emphasizes its uneven distribution, which accompanied that of GFAP in different subdivisions. A correlation between the distribution of aquaporin 4 and GFAP throughout the brain has been published by Venero et al. (2001). The cytoskeleton may affect the water channel proteins: in tissue cultures, aquaporin 4 immunopositivity appears together with the cytoskeleton (Yoneda et al., 2001).
A meaningful difference is that its immunoreactivity along the vessels is discontinuous in the central part of the AP but continuous in the border zone, such as throughout the brain. It may relate to a discontinuity of the glial cover of the central vessels. A similar phenomenon was found in the SFO (Pócsai & Kálmán, 2015) and OVLT (Kálmán et al., 2019) (see Table 3) as well as the aquaporin 4 immunonegativity of the ependyma covering the central area of these organs.
Aquaporin 4 is thought to increase the osmosensitivity of neurons since it facilitates water diffusion toward higher osmolarity. It results in a deformation of the astrocytes that may be a trigger (Noda & Sakuta, 2013;Venero et al., 2001;Wells, 1998) for the osmoreceptor neurons to which Bourque (2008) confines the direct osmosensitivity. In a recent publication, a similar effect is described by Wang et al. (2021) that hyperosmotic stress evokes the loss of water from astrocytes through aquaporin 4 molecules, cell shrinkage, and therefore a reversible retraction of astrocytic processes. It decreases the glial coverage of neurons and influences their activity.
The retraction is controlled by GFAP filaments.
On the other hand, astrocytes are equipped with a mechanism to respond to osmotic swelling ("regulatory volume decrease"). They contain a TRPV4/AQP4 complex (TRPV4: transient receptor potential vanilloid 4), which is sensitive to the membrane stretch activated by water influx (Benfenati et al., 2011;Mannari et al., 2013). Note that this latter mechanism is general in the brain and not confined to the CVOs.

S100
The S100-immunopositive elements were rather dense in the AP in agreement with the finding of Bauer et al. (2005). Besides processes, round large cell bodies were also found. S100 was described originally as an astroglial marker (Ludwin et al., 1976). The lack of colocalization with anti-NeuN ruled out its neuronal localization. Brizzee and Klara (1984) found very few oligodendroglia in the AP. S100 colocalized with GFAP and vimentin. S100 is an important Cabinding protein; it controls the free Ca level of astroglia, which plays a role in the polymerization of GFAP units (Bianchi et al., 1995;Ziegler et al., 1998). For other functions, see the recent review of Donato et al. (2013). In contrast to the SFO and OVLT, in the AP the distribution of S100 did not differ in the central and border areas. It is to be noted that S100 is a family of proteins (Donato et al., 2013), but our antibody was sensitive against S100 proteins in general.

Glutamine synthetase-immunoreactive cells
This enzyme is an accepted marker of astroglia (Martinez-Hernandez et al., 1977). The present results disagree with those obtained in the SFO (Pócsai & Kálmán, 2015) and OVLT (Kálmán et al., 2019), where the glutamine synthetase-containing cells were surprisingly infrequent. Perivascular "cuffs" were published in the AP by D'Amelio et al. (1987), whereas Berger and Hediger (2000) detected an intense GLAST (glutamate-aspartate transporter) but a moderate GLT-1 (glutamate transporter) activity. In the case of the SFO and OVLT, the poor occurrence of glutamine synthetase-containing cells was then attributed to that in CVOs; the relatively large extracellular space allows a less intense elimination of ammonium ions (Kálmán et al., 2019;Pócsai & Kálmán, 2015). Why the AP is different at this point remains unclear.
F I G U R E 1 0 Sketches summarizing the results. (A) A frontal sketch: the distribution of the immunoreactivities in the AP. The lines represent the planes of the sections shown in Figure 5 (sagittal) and Figure 6 (horizontal). The surrounding brain area (NTS: the subpostremal part of the nucleus of the solitary tract) is marked by immunoreactivities of GFAP and aquaporin 4 (blue-green stippling); the vessels are immunopositive for β-dystroglycan and aquaporin 4 but not for laminin. Vessels of this type are found throughout the brain, but they are not shown here. The central area is immunopositive for vimentin and nestin (orange-pink stippling) but not for GFAP. In the middle (asterisk), even the vimentin immunopositivity is weak. Its vessels (circles) are also immunoreactive for laminin besides the aquaporin 4 (green) and β-dystroglycan. The mantle zone (MZ; McKinley et al., 2003) is not demarcated with these immunostainings. The border zone (the funiculus separans [FS]) contains both GFAP and vimentin immunopositivity colocalized. The immunostaining of the junctional zone (JZ; McKinley et al., 2003; "commissural NTS" according to Dallaporta et al. [2010] and Troadec et al. [2022]) displays no difference from that of the border zone. The dashed line indicates the pial surface with vimentin (orange) or GFAP and aquaporin 4 (blue and green) immunopositive glial endfeet; it is also immunopositive for laminin and β-dystroglycan. CC: central canal; its ependyma is immunopositive to vimentin (orange), aquaporin 4 (green), and S100; SAS: subarachnoid space. Lines 5a-6c represent the levels of the sections shown in the corresponding panels in Figures 5 and 6. (B) The segments of a vessel, enlarged and highly schematic.
(1) The inner (proper) wall of the vessel immunopositive to laminin.
(2) Pial surface with colocalization of laminin and β-dystroglycan; it continues into the "outer wall" of the vessel (2′). (a-d) The segments of the vessels (for they descriptions. see Section 4.6). (c) The two walls are fused incompletely: the laminin immunopositivity (green) manifests yet indirectly in yellow color due to its colocalization with β-dystroglycan (red). (d) The fusion is complete; the laminin immunopositivity is not detectable. Black dashed line: the outer border of the central area of the AP; dotted lines: the borders of the segments of the vessel. (C) A sagittal sketch of AP is better to demonstrate the vascular system and (Continues) F I G U R E 1 0 (Continued) the surfaces. Color code: type (a) dark green; type (b) light yellowish-green; type (c) yellow, type (d) red. Arrow marks the site of attachment of the choroid plexus; 4V: fourth ventricle; CC: central canal; SAS: subarachnoid space. The dotted line represents the inferoposterior border of the central area of AP. The dashed blue line indicates the pial surface, which is immunopositive for GFAP or vimentin according to the underlying astroglial endfeet. It is also immunopositive for laminin, β-dystroglycan, and aquaporin 4; arrow: the attachment of the choroid epithelium of the fourth ventricle. Continuous line marks the ependymal (ventricular) surface, which is immunopositive for vimentin (orange), S100, and aquaporin 4 (green line), but this latter is missing corresponding to the ventricular surface of the AP.

CONCLUSION-"SENSORY" CVOs HAVE SIMILAR CHARACTERISTICS
In the three sensory CVOs, the AP, SFO, and OVLT, the distributions of astroglial markers and gliovascular connections follow the same principle (Table 3) according to our present and former results (Kálmán et al., 2019;Pócsai & Kálmán, 2015). Each organ has a central area with vimentin-and nestin-immunopositive glia, whereas GFAP and the water-channel aquaporin 4 were found in a peripheral zone. In the inner part, the vessels are immunopositive to laminin and have "double walls" with large perivascular spaces. Along their wall, the aquaporin 4 immunopositivity is discontinuous in contrast to the vessels in the periphery and throughout the brain (Goren et al., 2006). Utrophin immunopositivity is only detectable in the lamininimmunopositive vessels of the inner parts (see also Pócsai & Kálmán, 2014b). The BBB is lacking in these vessels (see, e.g., Morita et al., 2016). The ependyma, which is aquaporin 4-immunopositive throughout the ventricular system, is immunonegative corresponding to the central areas of the sensory CVOs. The peripheral zones are intensely immunopositive for several components of ECM (Pócsai & Kálmán, 2014a). Minor differences, however, occur: the inner part is only immunonegative to S100 in the SFO (its so-called "core"), and glutamine synthetase-immunopositive astrocytes are rather confined in the SFO and OVLT.

ACKNOWLEDGMENTS
The technical assistance of AndreaŐz and Szilvia Deák and the graphics and photoshop works of Endre Nemcsics are thankfully appreciated.

CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.

DATA AVAILABILITY STATEMENT
The original photomicrographs are available upon reasonable request at kalmanprof@gmail.com.

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1002/cne.25470