Differential expression of Na+/K+/Cl− cotransporter 1 in neurons and glial cells within the superficial spinal dorsal horn of rodents

Although convincing experimental evidence indicates that Na+/K+/Cl− cotransporter 1 (NKCC1) is involved in spinal nociceptive information processing and in the generation of hyperalgesia and allodynia in chronic pain states, the cellular distribution of NKCC1 in the superficial spinal dorsal horn is still poorly understood. Because this important piece of knowledge is missing, the effect of NKCC1 on pain processing is still open to conflicting interpretations. In this study, to provide the missing experimental data, we investigated the cellular distribution of NKCC1 in the superficial spinal dorsal horn by immunohistochemical methods. We demonstrated for the first time that almost all spinal axon terminals of peptidergic nociceptive primary afferents express NKCC1. In contrast, virtually all spinal axon terminals of nonpeptidergic nociceptive primary afferents were negative for NKCC1. Data on the colocalization of NKCC1 with axonal and glial markers indicated that it is almost exclusively expressed by axon terminals and glial cells in laminae I–IIo. In lamina IIi, however, we observed a strong immunostaining for NKCC1 also in the dendrites and cell bodies of PV-containing inhibitory neurons and a weak staining in PKCγ-containing excitatory neurons. Our results facilitate further thinking about the role of NKCC1 in spinal pain processing.

Distribution of NKCC1 immunostaining in the dorsal horn of the spinal cord. We observed a strong immunostaining for NKCC1 in the lumbar spinal cord of rats. Punctate immunostained profiles were distributed throughout the dorsal horn but the staining was more intense in laminae I-II than that in deeper laminae (Fig. 1d). Glial-like immunostained profiles were also observed in the white matter (Fig. 1d).

Expression of NKCC1 in neurons and glial cells. Colocalization of NKCC1 with markers of nocicep-
tive primary afferents. On the basis of their neurochemical characteristics, nociceptive primary afferents are divided into peptidergic and nonpeptidergic subgroups in the superficial spinal dorsal horn 27 . Most of the peptidergic ones contain calcitonin gene-related peptide (CGRP), whereas the nonpeptidergic afferents selectively binds isolectin-B4 (IB4) 27 . Thus, to reveal the expression of NKCC1 on nociceptive primary afferent terminals in the spinal cord, we studied the colocalization of NKCC1 with CGRP and IB4-binding.
Colocalization of NKCC1 with markers of axon terminals of glutamatergic and GABAergic spinal neurons. Vesicular glutamate transporter 2 (VGLUT2) in immunofluorescence studies can be used as a marker for axon ter- Photomicrographs showing immunoreactivity for NKCC1 in wild-type (a) and knockout (b) mice. NKCC1 immunostaining can be observed in the dorsal horn of the wild-type mouse, while the immunoreactivity is completely abolished from the dorsal horn of the NKCC1 knockout animal. c. Western blot analysis reinforces the specificity of the anti-NKCC1 antibody. The single immunoreactive band on the full-length running gel indicates that the antibody detects a protein with a molecular mass of ~ 160 kDa that corresponds to the molecular mass of NKCC1. For the molecular weight calibration, the precision plus protein dual color standards were used on which the blue and red colors appear as gray and white, respectively, after the black and white conversion (Bio-Rad, Hercules, California, USA) d. Photomicrographs  www.nature.com/scientificreports/ minals of intrinsic excitatory spinal neurons [31][32][33] , whereas GABAergic and glycinergic inhibitory interneurons transport both GABA and glycine into synaptic vesicles with the aid of the vesicular GABA transporter (VGAT) 34,35 . Therefore, to study the expression of NKCC1 on axon terminals of excitatory, glutamatergic and inhibitory, GABAergic and/or glycinergic spinal neurons we investigated the colocalization between NKCC1 and VGLUT2 as well as VGAT immunoreactivity. As expected, VGLUT2-IR profiles were evenly scattered in laminae I-II [31][32][33] . Despite the strong staining obtained for both NKCC1 and VGLUT2, the colocalization between the two molecules was low. Only 3.77 ± 0.52% of NKCC1-IR puncta were found in axon terminals immunostained for VGLUT2, and 11.87 ± 1.31% of VGLUT2 profiles were also immunostained for NKCC1 (Figs. 3,4c).
Similar to earlier reports 34 , VGAT-IR puncta were densely and evenly scattered throughout the superficial spinal dorsal horn. We recovered more NKCC1-IR puncta in VGAT-positive axon terminals than in those immunostained for VGLUT2, but the colocalization between NKCC1 and VGAT was still moderate. We found that 10.51 ± 1.26% of NKCC1-IR puncta in axon terminals were stained for VGAT, whereas NKCC1-IR puncta were recovered in 20.76 ± 0.82% of VGAT-IR axon terminals (Figs. 3, 4c,d).
More than half of the investigated glial cells expressed NKCC1, but the density of spots immunostained for NKCC1 was sparse on both types of glial cells. We found that 53.23 ± 1.84% of GFAP-IR profiles were immunoreactive for NKCC1, but only 16.86 ± 1.67% of NKCC1-IR puncta were scattered over profiles stained for GFAP (Figs. 3, 5a-c,g). Similar to GFAP, 62.99 ± 2.46% of IBA1-IR profiles were immunostained for NKCC1. However, only 6.85 ± 0.25% of NKCC1-IR puncta were recovered in profiles stained for IBA1 (Figs. 3, 5d-f,h).
Colocalization of NKCC1 with parvalbumin and protein kinase C gamma. The colocalization data of NKCC1-IR puncta in lamina I-IIo, where the peptidergic (CGRP-IR) nociceptive primary afferents terminate 28,29 , showed that NKCC1 is almost exclusively expressed by axon terminals and glial cells, primarily due to the remarkably high expression of NKCC1 in CGRP-IR axon terminals. In lamina IIi, where the nonpeptidergic (IB4-binding) nociceptive primary afferents terminate 30 , however, only a minor proportion of the IB4-binding axon terminals showed positive staining for NKCC1. Taking low axonal and moderate glial expression into consideration, we www.nature.com/scientificreports/ hypothesized that, in contrast to lamina I-IIo, the somatodendritic compartment of neurons may show strong expression for NKCC1 in lamina IIi. To test this notion, we studied the distribution of NKCC1-IR puncta on the dendrites of parvalbumin (PV)-containing inhibitory and protein kinase C gamma (PKCγ)-containing excitatory neurons, which are located in lamina IIi, and play major roles in spinal pain processing [41][42][43] . Fulfilling our expectations, we recovered NKCC1-IR puncta in high numbers in the dendrites of PV-IR ( Fig. 6a-c,g) neurons. However, immunostaining for NKCC1 was weak in PKCγ-IR dendrites ( Fig. 6d-f,h). In addition to the dendrites, we observed robust immunostaining for NKCC1 in the cell bodies of PV-IR neurons ( Fig. 7a-c). As in the dendrites, most of the immunostained puncta were scattered within the cytoplasm, and only a few of them were located so close to the border of the PV-immunostained areas that they could be regarded as cell membrane-associated staining (Figs. 6a-c,f, 7a-c). In contrast to the heavy staining of PV-IR neurons, PKCγ-IR neurons showed sparse immunostaining for NKCC1 not only in dendrites but also in the cell bodies ( Fig. 7d-f).
In addition to PV-IR and PKCγ-IR neurons, other types of neurons may also contribute to the formation of the neuronal assembly of lamina IIi, and the dendrites as well as cell bodies of these neurons may also express NKCC1; therefore, we did not perform any quantitative analysis concerning the distribution of NKCC1-IR puncta in the PV-IR and PKCγ-IR neurons. However, on the basis of the colocalization data of NKCC1-IR puncta with axonal and glial markers we assume that approximately 60% of NKCC1 immunoreactivity might be confined to the dendrites and cell bodies of neurons in lamina IIi.

Discussion
By studying the cellular expression of NKCC1 with immunohistochemical methods, we provided the first detailed account of the distribution of NKCC1 in neurons and glial cells in the superficial spinal dorsal horn of rats. We provide the first demonstration in the literature that almost all spinal axon terminals of CGRP-IR peptidergic nociceptive primary afferents in laminae I-IIo express NKCC1. In contrast, virtually all spinal axon terminals of IB4-binding nonpeptidergic nociceptive primary afferents in lamina IIi were negative for NKCC1. The www.nature.com/scientificreports/ colocalization data of NKCC1 indicated that it is almost exclusively expressed by axon terminals and glial cells in laminae I-IIo. In lamina IIi, however, we also revealed strong immunostaining for NKCC1 in the dendrites and cell bodies of selected neurons, which was stronger in PV-IR inhibitory cells and weaker in PKCγ-IR excitatory cells.

Differential distribution of NKCC1 in neurons and glial cells.
The expression of NKCC1 protein and mRNA in dorsal root ganglion (DRG) neurons has been well documented 12,19,23,25,44 . The experimental findings, however, are contradictory concerning the identification of subgroups of DRG neurons that are positive or negative for NKCC1. It has been reported that NKCC1 protein can be detected in virtually all DRG neurons in the frog, cat, rat 23 and mouse 19,24 . Price et al. 26 , however, found that NKCC1 mRNA expression is largely confined to small and medium diameter DRG neurons showing positive immunostaining for peripherin, CGRP or TRPV1. In contrast to the well-documented expression of NKCC1 in DRG neurons, NKCC1 protein has not yet been localized to terminals of primary sensory afferents in the spinal cord. Here, we demonstrated robust immunostaining in the axon terminals of CGRP-containing peptidergic nociceptive primary afferents in laminae I-IIo of the superficial spinal dorsal horn. The heavy staining of CGRP-IR axon terminals validates our other novel finding, namely that virtually all axon terminals of the IB4-binding nonpeptidergic nociceptive primary afferents were negative for NKCC1. Seemingly, this is in conflict with the results of those who found NKCC1 expression in all DRG neurons 19,23,24 . One may assume, however, that although NKCC1 is expressed in IB-4 binding DRG neurons, it is not transported from the cell bodies to the spinal axon terminals, similar to some ion channels that are expressed in the cell bodies of DRG neurons but missing from their central axon terminals 45,46 . NKCC1 protein and mRNA have already been found in the somata of some neurons in the superficial spinal dorsal horn 22,26 . In the present paper, we extended these previous reports with new findings. On the one hand, our colocalization studies strongly suggest that the cell bodies and dendrites of neurons the axon terminals of which express NKCC1 in the superficial spinal dorsal horn should be located almost exclusively in lamina IIi and deeper layers of the dorsal horn, reinforcing the notion that there are neurons in the spinal dorsal horn that have their cell bodies and dendrites in laminae IIi, III and send their axons into laminae I-IIo 47 . These neurons could give rise to axon terminals in laminae I-II that were VGLUT2-IR or VGAT-IR and were also stained for NKCC1, representing 11.8% of VGLUT2-IR excitatory and 20.7% of VGAT-IR inhibitory axon terminals in laminae I-II. We have to add, however, that in addition to axon terminals of spinal origin, VGLUT2-IR had been detected also in some nociceptive primary afferent fibers. Some authors reported a noticeable decrease in VGLUT2 immunoreactivity in the spinal dorsal horn after dorsal rhizotomy 31 . In addition, Todd et al. 33 showed that axon terminals of primary afferents retrogradely labeled with cholera toxin beta subunit in lamina I were positive for VGLUT2. Thus, the 11.8% proportion of VGLUT2-IR axon terminals in which we observed NKCC1 immunostaining can be partly axon terminals of intrinsic neurons, partly terminals of primary sensory neurons.
On the other hand, we identified two types of neurons in lamina IIi, PV-IR inhibitory and PKCγ-IR excitatory neurons 41-43 the cell bodies and dendrites of which were clearly stained for NKCC1. It is important, however, to note that the intensity of NKCC1 expression was remarkably different in these two types of cells. PV-IR neurons www.nature.com/scientificreports/ were heavily stained whereas PKCγ-IR neurons showed only very weak staining for NKCC1. In addition, most of the NKCC1 immunostained puncta were recovered in the cytoplasm of PV-IR dendrites and cell bodies, which may have an important functional consequence (see next chapter). Confirming earlier findings 44,48 we also found NKCC1 immunoreactivity in spinal glial cells. Although NKCC1-IR puncta were recovered in large populations of processes of both GFAP-IR astrocytes (53.2%) and IBA1-IR microglial cells (62.9%), immunostaining for NKCC1 was very weak in both cell types, especially in microglial cells.
Functional consideration. NKCC1 expression on CGRP-IR primary afferents. Some twenty years ago, it was demonstrated that the thermal nociceptive thresholds was altered in NKCC1 KO mice 19 . This finding was substantiated by several subsequent behavioral studies supporting the pro-nociceptive role of NKCC1 in persistent pain 17,49,50 , showing that the administration of the NKCC1 antagonist bumetanide has a significant antinociceptive effect 49 . It has also been demonstrated that in chronic pain conditions provoked by intradermal capsaicin injections, touch-evoked pain was reduced in NKCC1 deficient mice 17 , and this reduction could be mimicked by blocking NKCC1 via spinal application of bumetanide in wild-type animals 20,51 . These results generated the www.nature.com/scientificreports/ hypothesis that NKCC1 plays a role in spinal nociceptive information processing, and can be a factor that may participate in the mediation of touch-evoked pain. www.nature.com/scientificreports/ A number of experimental evidence substantiates the idea that stimulation of low-threshold mechanoreceptors or their afferent fibers (Aβ) can evoke depolarization of nociceptive (C/Aδ) primary afferents 6 . It has been hypothesized that this phenomenon, known as primary afferent depolarization (PAD), is generated by spinal interneurons that provide interactions between Aβ and C/Aδ primary afferents. According to the most widely accepted notion, activities of Aβ fibers excite inhibitory GABAergic neurons in the spinal dorsal horn. The activated inhibitory neurons make contacts with the spinal terminals of C/Aδ fibers through axo-axonic synapses. Then, GABA released in these synapses depolarizes the terminals of C/Aδ fibers through the activation of their GABA A receptors 6,52-54 . The administration of bumetanide onto the spinal cord reduces PAD 51 , suggesting that NKCC1 is involved in the generation of PAD evoked by GABA release 54 .
Thus, Aβ-fiber-mediated touch-evoked allodynia and the generation of PAD can be based on the same NKCC1-dependent synaptic mechanism 6,16,21,54 . Our present results support this earlier postulation. However, on the basis of our observations, we have to add that it is likely that PAD can be evoked in peptidergic (CGRP-IR), but not in nonpeptidergic (IB4-binding) nociceptive primary afferents. Almost all CGRP-IR axon terminals (94.45 ± 0.43%) showed strong immunostaining for NKCC1; thus, their intra-axonal Cl − concentration can be so high that the activation of their GABA A receptors may evoke membrane depolarization and consecutive PAD. In contrast, due to the very low level, virtual lack of NKCC1 expression, the intra-axonal Cl − concentration in nonpeptidergic (IB4-binding) nociceptive primary afferents can be so low that the activation of their GABA A receptors may mediate inward (hyperpolarizing) Cl − currents, which can also lead to the inhibition of the postsynaptic neuronal activity, but cannot generate PAD.
It must be emphasized, however, that this study investigated naive animals, in which PAD results in presynaptic inhibition of nociceptive primary afferents 52,55 and therefore attenuates acute pain. Under chronic pain conditions, however, both NKCC1 expression and intracellular chloride concentration may increase in the nociceptive axon terminals, as has already been observed in DRG neurons 21,24,25,56 . A Western blot study by Galan and Cervero 21 reinforces this notion. They found that the quantity of NKCC1 protein increased by 50% in the plasma membrane prepared from the lumbosacral spinal cord following intracolonic installation of capsaicin in mice. As a consequence of the increased NKCC1 expression, the excitability of C/Aδ primary afferent terminals may increase to such an extent that the GABA A receptor-mediated PAD can be so excessive that it may generate action potentials. The evoked action potentials of primary afferents can activate spinal neurons, resulting in touch-evoked pain (allodynia) 6,16,18,57 . The findings that NKCC1 KO mice display 17 and spinal application of bumetanide results in 18 reduced touch-evoked allodynia reinforces this hypothesis. According to our results, it is tempting to assume that touch-evoked allodynia can be more characteristic of those chronic pain states that are mediated by CGRP-IR peptidergic C/Aδ nociceptive primary afferents. Reinforcing this notion, touch-evoked allodynia has been reported in most cases in chronic pain conditions evoked by thermal, chemical or neuroinflammatory stimuli 6,[16][17][18][19][20][21]51,58 , which are conducted by peptidergic primary afferents. It conducts low-threshold Aβ fiber input from laminae IIi-III onto lamina I nociceptive neurons 47 , resulting in the generation of behavioral allodynia 62 . Under control conditions, this ventral-to-dorsal excitatory pathway is blocked by tonic inhibition 65 . However, in chronic pain, the inhibition is downregulated, which enables Aβ-fiber-mediated innocuous mechanical inputs to activate nociceptive projection neurons 42,59,66 . PV-IR inhibitory neurons in lamina IIi seem to be important constituents of neural circuits regulating the transmission of Aβ fiber activities to nociceptive specific neurons in lamina I 42,44 . They receive monosynaptic inputs from Aβ fibers and inhibit those excitatory neurons, which are also activated by Aβ fiber inputs and participate in the formation of the ventral-to-dorsal excitatory pathway 41,43 . In chronic pain conditions, the excitability of PV-IR neurons is reduced 41 , and the loss of inhibition can open up the gate for the Aβ fiber polysynaptic activation of the pain pathway. It is one of our most interesting findings that the cytoplasm of PV-IR neurons is loaded with NKCC1 but its expression in the cell membrane is limited. It is likely that cytoplasmic NKCC1 cannot influence the intracellular chloride concentration. But then, why does NKCC1 accumulate in the cytoplasm of PV-IR neurons? We do not know the answer to this question, but would like to put forward the notion that the cytoplasmic NKCC1 can be inserted into the cell membrane in chronic pain. If this happens, the influx of chloride into the PV-IR cells may increase. Consecutively, the excitability of PV-IR cells may decrease 67 , which may open up the gate for the flow of Aβ fiber activities through the ventral-to-dorsal excitatory pathway.

Methods
Animals and preparation of tissue sections. Animals were handled and tissue sections were prepared according to Antal et al. 68 with some modifications. Experiments were carried out on 8 adult rats (Wistar-Kyoto, 250-300 g, Gödöllő, Hungary), two wild-type and two NKCC1 knockout mice 69 . All animal study protocols were approved by the Animal Care and Protection Committee at the University of Debrecen and were in accordance with the European Community Council Directives. Animals were deeply anesthetized with sodium pentobarbital (50 mg/kg, i.p.) and transcardially perfused with Tyrode's solution (oxygenated with a mixture of 95% O 2 and 5% CO 2 ), followed by a fixative containing 4% paraformaldehyde dissolved in 0.1 M phosphate buffer (PB, pH 7.4). After transcardial fixation, the lumbar segments of the spinal cord were removed, postfixed in the original fixative for 3-4 h, and immersed in 10% and 20% sucrose dissolved in 0.1 M PB until they sank. To enhance reagent penetration the removed spinal cord was freeze-thawed in liquid nitrogen, sectioned at 50 µm on a vibratome, and extensively washed in 0.1 M PB.
Immunohistochemistry. Single and double immunostaining were carried out by following the protocols of Hegyi et al. 70 .
Single immunostaining. A single immunostaining protocol was performed to study the laminar distribution of NKCC1 in rats and wild-type as well as NKCC1 knock-out 69 mice. Free-floating sections were first incubated in mouse anti-NKCC1 antibody which was generated against a fusion protein fragment encompassing the C-terminus (S760-S1212) amino acids of the human colonic NKCC1 (diluted 1:3,000, T4 antibody, Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, USA) for 48 h at 4 °C, and then transferred into biotinylated goat anti-mouse IgG (diluted 1:200; Vector Labs., Burlingame, California, USA) for 12 h at 4 °C. Thereafter, the sections were treated with an avidin biotinylated horseradish peroxidase complex (diluted 1:100, Vector Labs., Burlingame, California, USA) for 5 h at room temperature, and the immunoreaction was completed with a 3,3′-diaminobenzidine (Sigma, St. Louis, Missouri, USA) chromogen reaction. Before the antibody treatments, the sections were kept in 20% normal goat serum (Vector Labs., Burlingame, California, USA) for 50 min. Antibodies were diluted in 10 mM Tris-phosphate-buffered isotonic saline (TPBS, pH 7.4) to which 1% normal goat serum (Vector Labs., Burlingame, California, USA) was added. Sections were mounted on glass slides, dehydrated and covered with Permount neutral medium.
Double immunostaining. Double-immunostaining protocols were performed to study the colocalization of NKCC1 immunoreactivity with various markers of nociceptive primary afferents, axon terminals of glutamatergic and GABAergic spinal neurons, somatodendritic compartment of neurons, astrocytes and microglial cells. Confocal microscopy and analysis. Confocal microscopy and qualitative as well as quantitative evaluation of the data were performed according to the method of Dócs et al. 71 . Single and serial 1 µm thick optical sections were scanned with an Olympus FV1000 confocal microscope. Serial optical sections were scanned with an overlap of 0.5 µm. Scanning was carried out using a 60× oil-immersion lens (NA: 1.4). The confocal settings (laser power, confocal aperture and gain) were identical for all sections, and care was taken to ensure that there were no saturated pixels corresponding to puncta immunostained for NKCC1 and all the markers applied for the visualization of nociceptive primary afferents, axon terminals of glutamatergic and GABAergic spinal neurons, somatodendritic compartment of neurons, astrocytes and microglial cells. The scanned images were processed in Adobe Photoshop CS5 software. The colocalization of NKCC1 with the investigated markers was quantitatively analyzed in the double-stained single 1 µm thick sections. A 10 × 10 standard square grid in which the edge length of the unit square was 5 µm was placed onto the regions of the confocal images corresponding to laminae I-II of the superficial spinal dorsal horn. The proper placement of the grid was based on the following criteria: (a) The border between the dorsal column and the dorsal horn was easily identified on the basis of the intensity of immunostaining. (b) The border between laminae II and III was approximated on the basis of previous observations [72][73][74] . It has been repeatedly demonstrated in ultrastructural studies that there are almost no myelinated axons in lamina II, while they are abundant in lamina III. Thus, the border between laminae II and III can be defined quite precisely in ultrastructural studies, and the thickness of laminae I-II can be measured. For this reason, immunoreactivities and colocalizations were investigated in the most superficial 150 µm thick zone of the dorsal horn that had previously been identified as a layer of the gray matter corresponding to laminae I and II in the L3-L5 segments of the spinal dorsal horn.
Profiles that showed immunoreactivity for NKCC1 over the edges of the standard grid were counted. The selected profiles immunoreactive for NKCC1 were then examined to determine whether they were also immunoreactive for axonal, neuronal or glial markers. Since the NKCC1 antibody utilized in the present study was raised against the intracellular domain of the enzyme, NKCC1 immunolabeled puncta were expected to be located within the confines of the area immunostained for the markers. Thus, to define the colocalization values, we counted only those NKCC1-immunolabeled puncta that were located within the confines of the areas immunostained for the marker. The colocalization for all investigated markers was analyzed in three animals. Three sections from each animal were randomly selected and the quantitative measurement was carried out in two regions of interest (ROIs) that were randomly selected from each section. One of the ROIs was placed in a more superficial region and the other in a deeper region of the layers stained for the neuronal or glial markers, but always within the confines of laminae I-II. Thus, the calculation of quantitative figures in each case was based on the investigation of 18 ROIs. From quantitative data obtained in the 18 ROIs, box-and-whisker plots were generated by using Origin Pro 8.0 software. The mean and standard error of the means (SEM) were also calculated for each value.
In order to substantiate the glial and neuronal localization of NKCC1, short series of confocal optical sections were also investigated in the X-Z and Y-Z projections. The X-Z and Y-Z images were drawn through points of colocalization between the two markers, and the two orthogonal views were investigated for overlap.

Controls.
The specificity of the antibody raised against NKCC1 has been extensively characterized 23,75 . It has been shown that the antibody recognizes NKCC1 protein in a wide variety of cell types [75][76][77] , including DRG 23 and cortical 78 neurons.
We further tested the specificity of the NKCC1 antibody in the spinal dorsal horn according to the protocols of Hegyi et al. 79 . (1) Free-floating sections obtained from NKCC1 knock-out 69 and wild-type mice were immunostained according to the single immunostaining protocol described above. (2) In order to obtain a more global view about the specificity of the anti-NKCC1 antibody, Western-blot analysis was performed on rats. While the animals were deeply anesthetized with sodium pentobarbital (50 mg/kg, i.p.), the spinal dorsal horn at the level of L3-L5 lumbar segments were dissected. The dorsal horn was sonicated in 20 mM Tris lysis buffer (pH 7.4) containing the following protease inhibitors (mM): EDTA (4.0), EGTA (2.5), PMSF (0.002) benzamidine (0.013), pepstatin A (0.004), soybean trypsin inhibitor (0.001), leupeptin (0.001) and aprotinin (0.001). After removing cell debris from the sonicated samples with centrifugation (1,500 rcf for 10 min at 4 °C), the supernatant was centrifuged again (12,000 rcf for 20 min at 4 °C). The pellet was resuspended in lysis buffer containing 1% Triton X-100 and 0.1% SDS, and the samples were run on 10% SDS-polyacrylamide gels according to the method of Laemmli 80 . The separated proteins were electrophoretically transferred onto PVDF membranes (Millipore, Billerica, MA, USA), and the membranes were immunostained according to the single immunostaining protocol described above. The immunostaining revealed only one immunoreactive band corresponding to the molecular mass of the glycosylated NKCC1 protein as seen in many types of cells 75,78 .
In order to test the specificity of the immunostaining protocol, free-floating sections were incubated according to the single immunostaining protocol described above with primary antibodies omitted or replaced with 1% normal goat serum. No immunostaining was observed in these sections.

Data availability
All relevant data are available from the authors.