The breeding and genotyping of mouse models deficient in slc4a10 have previously been described (Jacobs et al., 2008). Mice were bred on c57bl/6 background, and female and male mice littermates aging 4–5 weeks were used in an ~50:50 ratio. All procedures conformed to Danish animal welfare regulations. The authors are licensed to breed the mouse strains by The Animal Experiments Inspectorate, Ministry of Food, Agriculture and Fisheries (j.n. 2012-15-2935-00004).

All mice were perfusion fixed in succession via the heart with 3% paraformaldehyde in a phosphate-buffered salt solution (PBS, in mM: 167 Na⁺, 2.8 H₂PO⁻₄, 7.2 HPO²⁻₄; pH 7.4). After fixation the brain was removed, post-fixed for 2 h, dehydrated, and embedded in paraffin wax, enabling 2 μm sections to be cut using a rotary microtome (Leica). The sections were de-waxed and stepwise rehydrated, before epitopes were retrieved by boiling the sections in 10 mM Tris buffer (pH 9) with 0.5 mM EGTA. The epitopes were quenched with 50 mM NH₄Cl in PBS, and unspecific binding was blocked by washing with 1% BSA in PBS with 0.2% gelatin and 0.05% saponin. Sections were incubated overnight at 4°C with primary antibody diluted in 0.1% BSA in PBS added 0.3% Triton X-100. Primary antibodies are listed in Table 1, and positive control tissues included kidneys, brain, vasculature, and red blood cells (not shown).

For bright-field light microscopy, the sections were incubated in horseradish peroxidase-conjugated secondary antibodies (DAKO, Glostrup, Denmark) diluted in PBS with BSA and Triton X-100. The staining was visualized using 0.05% 3,3′-diaminobenzidine tetrahydrochloride dissolved in PBS with 0.1% H₂O₂. Mayer's hematoxylin was used for counterstaining, and the sections were dehydrated in graded alcohol and xylene. Finally, sections were mounted in Eukitt mounting medium (O. Kindler, Freiburg, Germany). The images were acquired using a Leica DMRE bright-field microscope equipped with a Leica DM300 digital camera. For fluorescence visualization of the primary antibodies, AlexaFlour 488- or 555-coupled donkey anti-goat, -sheep, -rabbit, or -mouse secondary antibodies (Invitrogen) were used, and cell nuclei were visualized using Topro3 counterstaining (Invitrogen). Sections were mounted with a coverslip in Glycergel antifade medium (DAKO) and analyzed using a Leica DMIRE2 inverted microscope with a TC5 SPZ confocal unit using ×63/1.32 NA or ×100/1.4 NA HCX PI Apo objectives with 8-bit depth for illustration of localization and colocalization, while 12-bit image depth was applied for fluorescence semiquantitation.

Specific protein abundance was investigated by quantifying the immunofluorescence intensities from confocal micrographs. All tissues were carefully handled in parallel from the time of fixation throughout embedding, sectioning, staining, and imaging. To avoid saturation of the photomultiplier, the intensity dynamic range (gain and offset) was adjusted to span the intensities of the most intense sample for each antibody. Images were acquired in the focal plane with the highest signal intensity using fixed settings for magnification, laser power, gain, image depth, offset, and averaging for all images with a given antibody.

The immunofluorescence intensities of the stained choroid plexuses were quantified from 12-bit gray scale images using Image Pro (Media Cybernetics). For each image, the area of interest was manually defined to avoid counts from non-choroidal tissue or artifacts. For all quantifications, the number of cell nuclei within the area of interest was automatically counted with fixed settings for minimal cell nucleus area, density/intensity signal, smoothing, and intensity range. The sum of immunofluorescence intensities was then divided by the number of nuclei to normalize for differences in choroid plexus tissue size among the sections. All analyzed images were from 4th ventricle choroid plexus, and data from the right and left brain sections were averaged for each animal when possible. In bar graphs, data are normalized to the mean wild type fluorescence signal. Where indicated, linescan intensity profiles were generated from fluorescence micrographs using Image Pro. The lines of interest are marked in white on the respective micrographs and represent only examples of juxta-labyrinth profiles from the indicated line.

Data is expressed as means ± sem. A non-parametric test (Mann–Whitley–Wilcox rank sum test) was used to compare two groups. Values of p < 0.05 were considered statistically significant. Each n represents one animal.

The Na⁺ independent HCO⁻₃ exchanger AE2, as well as the Na⁺,K⁺-ATPase, is typically linked to the spectrin cytoskeleton via anchoring proteins, such as ankyrins, in the basolateral membrane. In the CPE, however, the α1 subunit of the Na⁺,K⁺-ATPase is expressed in the luminal brush border, along with the β1 subunit of the Na⁺,K⁺-ATPase (Figure 1A, left and right panels, respectively). AE2 is located in the basolateral plasma membrane, similar to other epithelia (Figure 1B, left panel).

As the expression of spectrin and ankyrin isoforms in the CPE have not been determined, a panel of antibodies reacting specifically against ankyrins 1-3, spectrins αI-II, and spectrins βI-V were employed for immunohistochemical analysis of the CPE. As illustrated in Figure 1B (right panel), ankyrin-3 immunoreactivity was predominantly observed in close proximity to the luminal membrane and in microvilli. No other ankyrins were localized to the CPE with the applied antibodies but did label control tissues (not shown).

The αII-spectrin immunoreactivity was mainly observed in the luminal membrane domain (Figure 1C, left panel). However, αI-spectrin immunoreactivity was most prominent in relation to the basal labyrinth and to a lesser extent in the luminal membrane domain of the CPE cells (Figure 1C, right panel). βI- and βII-spectrin specific antibodies also produced a mainly luminal staining pattern with minor staining in relation to the lateral membrane and basal labyrinth (Figure 1D, left and right panels, respectively). The βII-spectrin labeling was also observed in a supra-nuclear location, where the microtubule organizing center is situated in CPE cells (see Figure 8). No other spectrin antibodies reacted with the CPE (not shown). The CPE expresses ankyrin-3, αII-spectrin, βI-spectrin, and βII-spectrin in high abundance beneath the luminal membrane and the three spectrins at a lesser extent in proximity to the basal labyrinth. In contrast, the fourth spectrin expressed in the CPE, αI-spectrin, shows an opposite expression pattern, with most expression in the basolateral labyrinth and minor in the luminal membrane. In regard to the different cellular localizations, AE2 is unlikely to bind ankyrins in the choroid plexus.

The basolateral adhesion molecule E-cadherin is known to link the cytoskeleton through either ankyrins or catenins. From Figure 1B (right panel), it appears that ankyrin-3 would be an unlikely anchor for E-cadherin as it is located near the luminal membrane. Figure 2A shows that E-cadherin localizes to the basolateral membrane domain of the CPE, opposite ankyrin-3 in the luminal membrane domain. As shown in Figure 2B, α- and β-catenin immunoreactivity is also pronounced in the basolateral domain. Both α-catenin and β-catenin is known to link to the actin cytoskeleton. Figure 2C shows a predominant luminal membrane domain immunolabeling for γ-actin, which extends into the microvilli. Thus, in the CPE, it seems more likely that E-cadherin is linked to the general actin cytoskeleton through catenins rather than through ankyrin-3 and spectrins.

Adducins are alternatives to ankyrins for linking AE2 to the cytoskeleton at the basolateral membrane of the choroid plexus, as they co-sediment separately from ankyrin and the Na⁺,K⁺-ATPase in sucrose gradients (Marrs et al., 1993). Adducins bind both the spectrin and actin cytoskeleton and adducin immunoreactivity was previously shown in proximity to AE2 in the CPE (Alper et al., 1994). Adducin forms dimers consisting of α-β or α-γ subunits. The brain sections were immunostained for the three adducin forms in order to establish their relative locations in the CPE. The α-adducin immunoreactivity was observed near both the luminal and basolateral membrane (Figures 2D, 3A), with highest reactivity at the basal labyrinth as expected.

Immunofluorescence staining for β-adducin and γ-adducin failed, but peroxidase stained sections are seen in Figures 3B,C, respectively. Antibodies against β-adducin yielded staining in the luminal domain, while γ-adducin staining was more pronounced toward the basolateral cell domain. Despite the variable performance of the adducin antibodies, the cellular distribution of adducins in CPE seems comparable to what is observed in renal epithelia (not shown). Syntaxin-3 is a typical luminal membrane SNARE protein (Mellman and Nelson, 2008), which ascertains the insertion of specific vesicles destined to this part of the plasma membrane. Figure 3D shows that syntaxin-3 is also a luminal membrane protein in CPE. Commercially available antibodies against the basolateral SNARE protein syntaxin-4 did not produce reliable immunostaining in CPE or control epithelia (not shown).

Previously, we have characterized the slc4a10 ko mouse (Jacobs et al., 2008) and showed that the cellular localization of NHE1 in these mice was found in the basolateral plasma membrane instead of in the luminal membrane as in the slc4a10 wt mouse (Damkier et al., 2009). Therefore, antibodies reacting specifically against the ezrin and moesin proteins known to interact with NHE1 directly or indirectly were used for immunohistochemical analysis of CPE. As previously observed, ezrin staining is found in the cytoplasm close to the luminal plasma membrane in the wt CPE (Figure 4A, micrographs). Ezrin staining was more variable among slc4a10 ko mice, but was also here found predominantly in the subluminal domain in the CPE as assessed by immunofluorescence (Figure 4B). Figure 5 shows that the ezrin distribution appears more cytosolic in slc4a10 ko mice compared to slc4a10 wt in peroxidase stained sections from the same mice. Some labeling is observed in the luminal membrane domain, but the labeling is also found in intracellular compartments. The bar graph in Figure 4A shows that the semi-quantified immunofluorescence signal for ezrin did not differ between slc4a10 wt and ko mice, (p = 0.556, n = 5 and 4 for wt and ko, respectively).

As shown in the micrographs of Figure 4C, moesin is found at the same cellular localization as ezrin, close to the luminal membrane, with no apparent difference in subcellular localization between the slc4a10 wt and ko. The bar graph shows that the moesin signal did not differ quantitatively between slc4a10 wt and ko mice (p = 0.286, n = 5 and 4 for wt and ko, respectively). Thus, neither ezrin nor moesin colocalizes with NHE1 in slc4a10 ko mice.

The Na⁺,K⁺-ATPase is linked to the spectrin cytoskeleton through ankyrin-3 in the choroid plexus (Marrs et al., 1993). The expression of α1-Na⁺,K⁺-ATPase was greatly decreased in slc4a10 ko mice (Damkier and Praetorius, 2012). Therefore, it is feasible that the Na⁺,K⁺-ATPase β subunit as well as the anchoring proteins would display similar changes in abundance or in subcellular distribution. Figure 6A illustrates that the subcellular localization of the β1-Na⁺,K⁺-ATPase subunit is strictly luminal in CPE from both slc4a10 wt and ko mice, and that the abundance is ~75% decreased in slc4a10 ko CPE (Figure 6A bar graph, p = 0.0357, n = 5 and 3 for wt and ko, respectively). Ankyrin-3 abundance seemed lower in slc4a10 ko CPE compared to wt littermates (Figure 6B, micrographs). This observation was not paralleled by a significant decrease in ankyrin-3 expression in slc4a10 ko (Figure 6B bar graph, p = 0.250, n = 5 and 3 for wt and ko, respectively). Thus, the β1-Na⁺,K⁺-ATPase parallels the α1-Na⁺,K⁺-ATPase protein abundance and localization in the CPE of slc4a10 wt and ko mice, while the protein abundance of their colocalizing scaffolding protein ankyrin-3 is unaffected in slc4a10 ko CPE.

The expression and localization of Na⁺-transporting slc4 gene family members NBCn1 and NBCe2 did not change significantly in slc4a10 ko mice (Damkier et al., 2009). However, Figure 6C shows that the abundance of AE2 encoded by slc4a2 is increased in the CPE from slc4a10 ko mice relative to the wt mice. The semi-quantitation depicted in the bar graph indicates an ~50% increase in the relative AE2 abundance in slc4a10 ko compared to wt (p = 0.029, n = 4 for wt and ko). The increase in AE2 protein was not accompanied by a significant change in α-adducin abundance in the slc4a10 ko CPE (Figure 6D, p = 0.250, n = 5 and 3 for wt and ko, respectively). Nevertheless, we noted a non-significant trend toward increased total α-adducin abundance and a more pronounced basolateral staining relative to the luminal staining intensity. The luminal immunoreactivity was almost completely absent in the slc4a10 ko CPE (Figure 6E).

The αI-spectrin immunoreactivity shows an ~60% decrease in protein abundance in the slc4a10 ko compared to the wt (Figure 7A, p = 0.0357, n = 5 and 3 for wt and ko, respectively). This significant change in protein abundance was accompanied by a difference in cellular localization between the genotypes; from predominantly staining the basolateral labyrinth in the slc4a10 wt to almost exclusively luminal expression in the slc4a10 ko (Figure 7B). The immunofluorescence signal for αII-spectrin was not significantly altered in the CPE from slc4a10 ko mice as compared to wt (Figure 7C, p = 0.143, n = 5 and 3 for wt and ko, respectively). There was no apparent change in cellular localization of the protein apart from more cytosolic immunoreactivity (Figure 7D).

Analysis of the βI-spectrin abundance in the CPE resulted in a non-significant decrease in the slc4a10 ko compared to wt mice (Figure 8A, p = 0.057, n = 5 and 3 for wt and ko, respectively), where the cellular localization did not change markedly (Figure 8B). The immunofluorescence signal for βII-spectrin shows no difference between slc4a10 ko and wt CPE (Figure 8C, p = 0.85, n = 5 and 3 for wt and ko, respectively), whereas the cellular localization appears less basolateral in the ko mice (Figure 8D).

Major changes in protein abundances of E-cadherin and the catenins were not expected in the slc4a10 ko CPE, as the major changes in protein expression are observed in relation to the luminal plasma membrane. The micrographs in Figure 9A shows similar labeling intensity and cellular distribution of E-cadherin in slc4a10 wt and ko CPE (p = 0.686, n = 4 and 4 for wt and ko). Nevertheless, as seen in Figure 9B, the abundance of α-catenin is significantly decreased in the slc4a10 ko CPE as compared to the wt (Figure 9B Bar graph, p = 0.0357, n = 5 and 3 for wt and ko, respectively). Furthermore, the cellular localization of α-catenin seems to become more abundant in the lateral membrane domain and less in the basolateral labyrinth (Figure 9C). For β-catenin, similar expression patterns and protein abundances were observed in the two genotypes (Figure 9D, p = 0.143, n = 5 and 3 for wt and ko, respectively).

Claudin-2 expression in the tight junction of the CPE is most likely a determining factor in paracellular permeability in the epithelium. We hypothesized that decreased expression of ion and water transporters in CPE of slc4a10 ko mice and most likely also decreased CSF secretion capacity would minimize the need for paracellular movement of ions and water molecules through claudin-2. The representative micrographs in Figures 10A,B illustrate that the claudin-2 immunoreactivity in CPE is decreased in slc4a10 ko compared to slc4a10 wt mice. The occasional observations of continuous lines of claudin-2 labeling in the luminal cell domain probably corresponds to tangentially sectioned cells, as judged from the DIC overlay in Figure 10B. The bar graph shows an ~50% decrease in CPE anti-claudin-2 staining in the slc4a10 ko (p = 0.0357, n = 5 and 3 for wt and ko, respectively). Thus, it is likely that the capacity for selective movement of certain ions and perhaps water molecules via the paracellular route may be limited in slc4a10 deficient mice.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.