The polypeptide from the cytoplasmic interdomain 2–3 (ID_2–3, amino acids 753–917 from Rattus norvegicus sequence, accession number NP_113874) was synthesized in the CHUL research centre (Québec, QC, Canada). Briefly, the 495 bp corresponding cDNA sequence (accession number NM_031686) was amplified with the following PCR primers: forward, 5′-GGGGGATCCGCTTCATACGATGCTACCACAGAA-3′ and reverse, 5′-GGGCTCGAGAAGCTTCGACTTTTCATGTTCATAG-3′, included the restriction enzyme site BamHI and XhoI, respectively (underlined sequences).The amplified PCR fragment was cloned in plasmid expression pGEX-4T vector in frame with the glutathione S-transferase (GSH-T) tag, and transformed in E. coli BL21-Gold (D3) strain. The resulting DNA construction was purified and sequenced on the both strand to check unwanted mutation. Rat Na_X fusion protein was induced with 0.1 mM IPTG, 2 h at 37°C and purified on glutathione sepharose 4B column. The purified fusion protein was injected into rabbit for immunization and the antibodies were affinity purified by GenScript Company (GenScript Inc., NJ, USA).

Fifteen Wistar male rats (100–150 g) and fifteen C57BL/6 male mice (20–25 g) were purchased from Charles River (Wilmington, MA, USA) 1 week before experimentation. The animals were housed in plastic cages and acclimated to standard laboratory conditions (12 h light/12 h dark cycle; temperature controlled at 23°C). The animals had free access to tap water and regular rat chow (# 2018, Harlan Teklad, Montreal, QC, Canada). The experimental procedures were performed according to the guidelines of the Canadian Council on Animal Care and approved by the animal care committee of our institution (Université Laval).

The rats and mice were deeply anesthetized (ketamine, 87.5 mg/kg and xylazine 12.5 mg/kg solution, i.p. and ketamine, 8.75 mg/kg and xylazine 1.25 mg/kg solution, i.p., respectively) and euthanized by transcardiac perfusion with cold saline (4°C), immediately followed by the perfusion of a fixative solution [4% paraformaldehyde (PFA), pH 7.4; 4°C]. The brains were quickly removed from the skull, post-fixed 6 h in the fixative solution and then immersed for 40 h at 4°C in a sterile phosphate buffer salt (PBS) solution containing 20% sucrose. The brains were frozen on dry ice and mounted on a microtome stage. Coronal sections (30 μm for rats and 20 μm for mice) were collected in a cryoprotectant solution and stored at −20°C. Before use, the brain sections were rinsed in sterile PBS solution.

The primary and secondary antibodies used in this study are listed in Table 1. The free-floating brain sections were incubated 1 h at RT in a blocking solution (PBS solution containing 0.3% Triton X-100, 5% normal goat serum, 1% BSA), and then transferred in PBS solution containing primary antibodies overnight at 4°C. After washing in Tween-containing PBS solution (0.05%), the sections were incubated for 2 h at RT with the secondary antibodies. Finally, the brain sections were mounted onto poly-l-lysine-coated slides, air-dried and coverslipped with Vectashield^® mounting medium for fluorescence (Vector Laboratories, CA, USA).

Rats and mice were prepared as for tissue preparation for immunofluorescence without tissue fixation step. Briefly Brains were quickly removed from the skull and submerged in ice-cold (2°C) artificial CSF (aCSF) continuously bubbled with a gas mixture (95% O₂–5% CO₂) and containing 2 mM KCl, 1 mM CaCl₂, 3 mM MgCl₂, 26 mM NaHCO₃, 1.2 mM NaHPO₄, 10 mM d-glucose, 200 mM sucrose, pH 7.4. Osmolality was adjusted to 298–300 mOsm/l with mannitol. Coronal brain slices (350 μm thick) were cut with a vibratome (model VT1000S; Leica, Nusloch, Germany) then transferred in aCSF continuously bubbled with a gas mixture (95% O₂–5% CO₂). Brain regions of interest were micropunched, transferred in 1.5 ml microtube and freeze in dry ice. Tissue was kept at −80°C until to use.

Total RNA from brain micropunches were extracted and DNase-treated with the PureLink™ RNA Micro kit according to the manufacturer’s instructions (Invitrogen, Burlington, ON, Canada). The concentration was determined by measurement of absorbance at 260 nm with an absorbance reader (model Infinite F200; Tecan, Durham, USA). Reverse transcription (RT) PCR experiments were performed in two steps. First, cDNA synthesis was carried out with the Transcriptor first-strand cDNA synthesis kit (Roche Diagnostic, Quebec, QC, Canada). RT reaction used a mix of random hexamer and oligo-dT primers of the kit with 400 ng of total RNA in a final volume of 20 μl. Three microliters of the first-strand cDNA synthesis reaction was used as template for PCR amplification. PCR fragments were amplified with Maxima^® Hot Start Taq DNA Polymerase (Fermentas, Burlington, ON, Canada) and primers: MOUD66: 5′-ACTGTGTTCCGAATCCTCTG-3′/MOUD59: 5′-TCATGTCTCCATACTCCAGG-3′ for Na_X, and MOUD82: 5′-GTCTTCACTACCATGGAGAAGG-3′/MOUD83: 5′-TCATGGATGACCTTGGCCAG-3′ for GAPDH (Jia et al., 2010). The PCR program to amplify the 890 bp for Na_X and 200 bp for GAPDH products was as follows: one hold at 94°C for 5 min followed by 35 cycles of 94°C for 30 s, 58°C for 30 s, and 72°C for 60 s, and were visualized on a 2% (w/v) agarose gel.

High-resolution images were collected on a i90 Nikon fluorescence microscope (Nikon Canada Inc., Mississauga, ON, Canada) coupled to a Hamamatsu 1394 ORCA-285 monochrome camera and exploited by Simple PCI software version 5.3.0.1102 (Compix Inc. Imaging Systems, PA, USA). Finally, all images were adjusted for brightness and contrast using Adobe Photoshop, and the figures were assembled in Adobe Illustrator.

The aim of the present study was to determine the distribution and phenotype of the cells that express the Na_X channel in the rat and mouse brain. Accordingly, we developed a specific antibody directed against the cytoplasmic loop linking domains 2 and 3 (ID_2–3) of the Na_X channel. This region was targeted because protein sequence analysis revealed that ID_2–3 is unique among the voltage-gated Na⁺ channel family members. Indeed, the targeted 165 amino acids sequence (Figure 1A) shares a low percentage of identity (ranging 13–42%) with the other voltage-gated sodium channels (Na_v1.1–Na_v1.9, Table 2). Interestingly, this protein sequence identified in the rat shares a high percentage of identity (87%) with the protein sequence identified in the mouse. Taken together, we assumed that the polyclonal antibody produced from the ID_2–3 immunogenic peptide was Na_X specific and could be used to detect Na_X expression in both the mouse and rat.

The ID_2–3 immunogenic peptide was produced in E. coli and expressed using the recombinant plasmid pGEX-Na_X-ID_2–3. Na_X-GST fusion protein expression was induced using IPTG and purified by glutathione affinity chromatography as described in Section “Materials and Methods.” The quality of the protein purification was evaluated using SDS-PAGE (Figure 1B).

The purified Na_X-GST protein was used to immunize 2 rabbits to produce the polyclonal anti-Na_X antibody. The purified Na_X-GST protein was injected four times in 2 months. The rabbits were bled 1 week after the fourth injection of the immunogenic peptide and the immune serum was purified in two steps. The first purification step was to eliminate the antibodies that did not bind the Na_X-GST protein, which was then followed by a step to eliminate the antibodies that solely bound the GST-tag of the immunogenic peptide. Validation of the anti-Na_X antibody was achieved with several tests. First, the ability of the purified antibodies to bind the Na_X protein was performed using an ELISA test (results provided by GenScript Inc., Figure 2A). This test demonstrated that only the purified antibodies bound the Na_X protein, whereas the pre-immune serum did not bind the immunogenic peptide (negative control). Note that the antibody titer was defined as the maximal dilution giving an OD_450 nm signal greater than 0.1 and was calculated at a 1/512,000 dilution for the antibody batch used in the present study. Subsequently, the specificity of the anti-Na_X antibody was verified in situ with immunohistochemical staining. Using thin (30 μm) brain slices, we evaluated Na_X immunostaining in the rat MnPO (positive control) and BST (negative control; Figure 2B, upper panels). As expected, fluorescent immunostaining was restricted to the MnPO, which is known to express both Na_X mRNA and protein (Grob et al., 2004; Tremblay et al., 2011). In contrast, no fluorescent staining was observed in the BST, which has been shown to be devoid of Na_X mRNA as previously reported (Grob et al., 2004). Preadsorption carried out with increasing concentrations of the immunizing peptide (0–200 μg/ml) indicated that the fluorescent immunostaining in the MnPO could be totally abolished when the anti-Na_X antibody was pre-incubated with the immunizing peptide (50 μg/ml; Figure 2B, lower panels). Preadsorption carried out with an excess of GST protein (200 μg/ml) did not interfere with immunostaining in the MnPO (data not shown). Finally, the immunostaining was correlated with the presence or the absence of RT-PCR products isolated from the MnPO and BST, respectively (Figure 2C).

Together, this series of experiments led to the production and validation of a polyclonal rabbit anti-Na_X antibody. This antibody is highly specific toward the rat and mouse Na_X ID_2–3.

To examine the specificity in the Na_X distribution in the rat and mouse brain, the entire brains of these rodents were sliced in thin sections and the regions of interest, which included all of the CVOs, the MnPO and the supraoptic and paraventricular nucleus (SON, PVN), were incubated with the anti-Na_X antibody. In addition, fluorescent Na_X immunostaining was combined with various cell markers to identify the cellular phenotype that expressed the Na_X channel. Thus, Na_X immunostaining was performed in combination with anti-NeuN, anti-GFAP, or anti-vimentin immunohistochemistry to identify neurons, glial cells or ependymal cells and tanycytes, respectively. Immunostaining was also performed with oxytocin and vasopressin-associated neurophysin to distinguish between oxytocin and vasopressin-expressing magnocellular neurons.

Although rats and mice are phylogenetically close animals, our anatomical analysis revealed noticeable differences in terms of patterns and cell types in the brain that expressed Na_X.

Na_X immunostaining in both the rat and the mouse was detected in the CVOs, the OVLT, the SFO, the ME, and the area postrema (AP). The pattern and cellular phenotype for Na_X expression in both rats and mice showed similarities only in two CVOs, which were the ME and the AP. In the ME, Na_X immunostaining was essentially observed in the ventral portion of the wall of the third ventricle and in the radial processes penetrating the neuropil. Na_X immunostaining colocalized with vimentin immunostaining only (Figure 3; Table 3). Intriguingly, Na_X immunostaining was associated with none of the cell markers in the AP (Table 3). However, the clear regionalization of Na_X immunostaining (central, lateral, and ventral zone), of vimentin staining (funiculus separans, fs) and of GFAP staining (inverted pyramidal region extending from the fs to the central canal) ruled out possible colocalization of Na_X immunostaining with tanycytes or glial cells in the rat and mouse AP. In the other CVOs, Na_X cellular localization was strongly divergent (Table 3). Indeed, in the rat OVLT, Na_X immunostaining was dense in the cell layers boarding the third ventricle and in individual cells of the lateral periventricular tissue. Na_X immunostaining colocalized with NeuN immunostaining in the lateral part of the OVLT and with vimentin immunostaining lining the third ventricle (Figure 4). In the mouse OVLT, Na_X immunostaining was confined to a dense fiber network and only colocalized with vimentin immunostaining (Figure 5). Na_X immunostaining was not colocalized with NeuN or GFAP immunostaining. The distribution pattern of the Na_X-expressing cells was also different in the SFO of the two rodents. In the rat SFO, Na_X immunostaining was present throughout the organ with a clear staining of the ependymal cell layers boarding the ventricular surface. Na_X immunostaining colocalized with NeuN positive cells in both the core and periphery of the SFO, as well as with vimentin immunostaining in the ependymal cell layers (Figure 6). In the mouse SFO, Na_X immunostaining was present in clusters boarding the ventricular surface. Na_X immunostaining colocalized with vimentin immunostaining only (Figure 7). Overall, these results indicate that Na_X is expressed in neurons and ependymocytes in the rat OVLT and SFO, and only in ependymocytes in the mouse organs (Table 3).

The MnPO lies in a strategic position in order to monitor CSF ion composition and, thus, CSF osmolality. This functional role results from the absence of tanycytes and their tight junctions at the CSF–MnPO interface and from the presence of ependymal ciliated cells that permit passive diffusion of ions from the CSF. In the MnPO, the Na_X immunostaining was observed in individual cells homogeneously distributed in the nucleus. Na_X immunostaining solely colocalized with NeuN immunostaining in the rat (Table 3; Figure 8A). Interestingly, the mouse MnPO was characterized by the absence of Na_X immunostaining (Figure 8B).

Together, the present data demonstrate noticeable differences between the rat and mouse in the distribution pattern and the cell types expressing Na_X protein in the brain regions involved in the detection of osmolarity and Na⁺ concentration.

The SON and the PVN of the hypothalamus are the primary regions responsible for the control of hydromineral balance, which is because the magnocellular neurons in both the SON and PVN synthesize antidiuretic hormone (vasopressin) and antinatriuretic hormone (oxytocin) in rodents. Na_X immunostaining in the SON and the PVN was comparable between the two species, and it was primarily observed in the magnocellular neurons (Table 4). Na_X immunostaining in these nuclei was clearly supported by a positive Na_X RT-PCR signal in the micropunched tissues from the SON and PVN (Figure 9A). To further discriminate which population of magnocellular neurons expressed the Na_X channel, Na_X immunofluorescence was combined with anti-neurophysin associated vasopressin (VP–NP) and oxytocin (OT–NP) immunofluorescence (the anti-VP–NP and anti-OT–NP antibodies were provided by Dr. H. Gainer (NIH, Bethesda, MD, USA). These results show that Na_X immunostaining colocalizes with the magnocellular vasopressinergic and oxytocinergic neurons in the SON (Figures 9B,C) and in the PVN (Figures 10A,B). Moreover, most of the neuronal population showing Na_X immunostaining in the parvocellular part of the PVN was not immunofluorescent for VP–NP or OT–NP (Figure 10).

To compare the expression pattern and the phenotype of cells expressing Na_X in the rat and mouse brain, we produced an antibody against the ID_2–3 region of the rat Na_X protein. This region is highly conserved between the rat and mouse but shows weak homology with other voltage-gated sodium channels. A mandatory step in validating this anti-Na_X antibody was to assess the purification level and the specificity toward the Na_X channel. We showed that the polyclonal anti-Na_X antibody specifically bound the immunogenic peptide without interference from the GST protein. The anti-Na_X antibody labeled the rat MnPO nucleus, which has been previously described as being a Na_X-positive brain nucleus (Grob et al., 2004; Tremblay et al., 2011). Furthermore, this labeling was reduced to background levels after pre-incubation with an excess of the immunogenic peptide. Importantly, Na_X immunostaining was correlated with Na_X mRNA expression in Na_X-positive and Na_X-negative regions of the rat brain (MnPO and BST, respectively; Grob et al., 2004).

Overall, multiple tests validated the specificity of the antibody against the ID_2–3 region of the Na_X channel. Thus, this tool appears appropriate for evaluating Na_X distribution and for identifying the cell types expressing this distinct Na⁺ channel.

The function of the Na_X channel has been well characterized in the context of Na⁺ homeostasis in the forebrain CVOs of the mouse (Hiyama et al., 2002; Watanabe et al., 2003; Shimizu et al., 2007) and in the rat MnPO (Grob et al., 2004; Tremblay et al., 2011), where Na_X constitutes the molecular basis of the brain’s ability to sense Na⁺. Here, we show that Na_X distribution was highly similar in the two rodents because Na_X immunostaining was clearly observed in all the CVOs of the rat and mouse. However, a noticeable difference was the presence of Na_X immunostaining in the rat MnPO, whereas this nucleus was Na_X-immunonegative in the mouse. This Na_X distribution pattern was validated under similar experimental conditions for the two rodents, which demonstrates the presence of Na_X throughout the lamina terminalis of the rat and not just in the MnPO. Moreover, the specific Na_X expression pattern in the MnPO of the rat may contribute to a more complex organization of the sodium sensing network in this animal and may represent an additional detection site for CSF sodium levels. Note that the absence of Na_X immunostaining in the mouse MnPO has a direct physiological relevance: patch-clamp recordings of dissociated mouse MnPO neurons revealed the absence of membrane depolarization during the occurrence of hypernatriuric artificial CSF (Tremblay et al., 2011).

In addition to the differential Na_X distribution between these two rodents, the primary difference between the rat and mouse was in regard to the cell types displaying Na_X immunostaining in the CVOs. In the rat, Na_X was colocalized with NeuN, a neuronal marker, in the lateral periventricular part of the OVLT and in the core of the SFO. Na_X was also colocalized with vimentin, a marker of ependymocytes and tanycytes (Dahl, 1981; Schnitzer et al., 1981; Pixley, 1984; Petito et al., 1990), in the cell layers lining the third ventricle in the OVLT, SFO, and ME. In the mouse, Na_X immunostaining was mainly, if not exclusively, colocalized with vimentin in the OVLT, SFO, and ME. This result adds further support to previous data reporting the presence of Na_X in glial processes in the mouse CVOs (Watanabe et al., 2006). In that study, Na_X immunostaining was reported to colocalize with GLAST immunostaining, which is a glutamate transporter expressed in glial cells. However, GLAST is expressed in both astrocytes and tanycytes (Shibata et al., 1997; Berger and Hediger, 2001), and might not allow to distinguish between the two cell types. In the present study, there was clear colocalization of vimentin and Na_X immunostaining in the SFO, OVLT, and ME. Na_X immunostaining was also colocalized with GLAST positive cells in the mouse SFO (data not shown), like previously observed (Watanabe et al., 2006). Although vimentin and GFAP have been reported in tanycytes, the GFAP content is very low and tanycytes generally appear vimentin immunopositive and GFAP negative in the mediobasal hypothalamus and CVOs, like the ME and the subcommissural organ (Chouaf et al., 1989; Chauvet et al., 1995, 1998). Na_X immunostaining in the SFO, OVLT, and ME, concomitant with the absence of colocalization of GFAP, indicates that the Na_X channel is expressed in the ependymocytes and/or tanycytes lining the third ventricle in the rat and mouse. This observation was puzzling at a first glance because Na_X gene was isolated from cultured rat astroglia (Gautron et al., 1992). The cell cultures used in that study were prepared from 1-day-old pups and in the rat, tanycytes are generated during the late pregnancy and the first postnatal day (Rodriguez et al., 2005). Because, the radial glial cells are transformed into astrocytes and also differentiated into tanycytes during the perinatal period, Na_X RNA might have been isolated from a mixed population of glial cells. Alternatively, the presence of a subpopulation of astrocytes that express vimentin, GLAST and GFAP in the SFO and OVLT might be possible. This putative population is however weak in the forebrain CVOs because most of the GFAP immunostaining identified stellate cells and processes that were not Na_X immunopositive.

The expression of the Na_X channel in the neurons, ependymocytes and tanycytes raises the question of the degree of complexity for sodium sensing in the rat. In addition to the unique Na_X expression in the rat MnPO neurons, it is conceivable that the neuronal populations of the SFO and OVLT encode sodium levels in a similar way (Grob et al., 2004; Tremblay et al., 2011). In fact, the neuronal mechanism of sodium sensing in the MnPO allows the transduction of both hypo- and hypernatriuric signals into membrane hyperpolarization and depolarization, respectively (Grob et al., 2004; Tremblay et al., 2011). Therefore, neuronal Na_X in the rat CVOs may transduce sodium levels ranging below and above the sodium set point into distinct firing patterns. In addition, a second mechanism involving ependymocytes and/or tanycytes may lead to lactate release and a resulting change in the excitability of neighboring neurons, which has been reported in the mouse SFO (Shimizu et al., 2007). This mechanism will be engaged under hypernatriuric conditions only and may possibly bolster the initial neuronal signaling.

Interestingly, our results show that Na_X is expressed in the AP of both the rat and the mouse. Na_X immunostaining did not colocalized with the tanycytes of the funiculus separans zone or with GFAP-positive cells observed in the inverted pyramidal area extending from the ventral zone of the AP to the central canal. Na_X immunostaining was restricted to cells present in the three zones of the AP, which are the central, the lateral, and the junctional zone of the AP. The absence of vimentin and GFAP-positive cells in these zones strongly suggests that Na_X is expressed by neurons in the AP. The observation of Na_X immunostaining in the AP indicates that Na⁺ sensing intrinsically occurs in the medullary CVO, which supports the role of the AP in the central network of sodium sensing. The present data indicate that AP neurons not only relay the chemosensitive information regarding plasma sodium levels from the hepatoportal osmoreceptors but also likely serve as genuine sodium sensors similar to those found in the lamina terminalis. Hence, these neurons could directly forward sensory information to higher integrators such as the aldosterone-sensitive neurons in the nucleus tractus solitarius (NTS) or the FoxP2-immunoreactive neurons in specific divisions of the parabrachial nucleus (PBN; Geerling et al., 2006, 2011; Geerling and Loewy, 2007; Stein and Loewy, 2010). Moreover, the presence of Na_X-expressing neurons in the AP may partly account for the neural activity observed in the rat AP after sodium ingestion but not after sodium depletion (Geerling and Loewy, 2007). Indeed, high plasma sodium levels will depolarize these sodium sensing neurons, which would likely lead to the observed Fos immunostaining in the AP. By contrast, sodium levels below the set point would hyperpolarize these neurons, thereby reducing the neural activity and the presence of Fos in the CVO.

The presence of Na_X immunostaining in the magnocellular neurosecretory cells (MNCs) of the SON and PVN in the rat and mouse represents a major finding of this anatomical study. The expression of the Na_X channel by both the vasopressinergic and oxytocinergic MNCs was unexpected according to a previous mapping study of the Na_X gene in the mouse brain (Watanabe et al., 2000). However, this discrepancy may be related to the labeling method used in the two studies or to the abundance of the Na_X-expressing cell population in the two rodent species. Watanabe et al. (2000) used a Na_X deficient mouse in which a lacZ reporter gene was inserted in frame into the mouse Na_X gene; accordingly, the sensitivity of this technique may not have been adequate for Na_X detection in smaller cell populations. Indeed, our data clearly indicate that the size of the Na_X-immunopositive MNC population is higher in the rat than in the mouse. Intriguingly, in situ hybridization carried out with a Na_X riboprobe failed to reveal the presence of Na_X mRNA in the rat SON (Grob et al., 2004). However, RT-PCR experiments performed on micropunched tissues from both the SON and PVN in the present study showed a clear Na_X-positive signal in the two nuclei. Similar differences in the level of Na_X expression were reported in the human hippocampus where the presence of Na_X mRNA was revealed by RT-PCR and not by in situ hybridization (Gorter et al., 2010). Taken together, our results indicate that basal level of Na_X mRNA in the MNC is weak, thereby suggesting a low turn-over of the protein in these neuroendocrine cells. Functionally, Na_X does appear to participate in the process of Na⁺ detection in the SON and PVN. It was previously shown that the MNCs were permeable to the Na⁺ ion, which was demonstrated with a N-terminal variant of the TRPV1 (transient receptor vanilloid 1; Voisin et al., 1999; Naeini et al., 2006). These electrophysiological recordings demonstrated that an increased extracellular Na⁺ concentration triggered an inward current with an amplitude that was strongly reduced (∼88%), but not abolished, by the application of gadolinium, which is a specific TRPV blocker (Voisin et al., 1999). The residual current was quantitated to be about 5 pA and might represent sodium influx through the Na_X channel. Indeed, such a current amplitude is well in the range of the sodium inward current flowing through Na_X in the MnPO neurons recorded in situ (Grob et al., 2004).