Molecular characterization of a putative K-Cl cotransporter in rat brain. A neuronal-specific isoform.

Using a combination of data base searching, polymerase chain reaction, and library screening, we have identified a putative K-Cl cotransporter isoform (KCC2) in rat brain that is specifically localized in neurons. A cDNA of 5566 bases was obtained from overlapping clones and encoded a protein of 1116 amino acids with a deduced molecular mass of 123.6 kDa. Over its full length, the amino acid sequence of KCC2 is 67% identical to the widely distributed K-Cl cotransporter isoform (KCC1) identified in rat brain and rabbit kidney (Gillen, C., Brill, S., Payne, J.A., and Forbush, B., III(1996) J. Biol. Chem. 271, 16237-16244) but only approximately25% identical to other members of the cation-chloride cotransporter gene family, including "loop" diuretic-sensitive Na-K-Cl cotransport and thiazide-sensitive Na-Cl cotransport. Based on analysis of the primary structure as well as homology with other cation-chloride cotransporters, we predict 12 transmembrane segments bounded by N- and C-terminal cytoplasmic regions. Four sites for N-linked glycosylation are predicted on an extracellular intermembrane loop between putative transmembrane segments 5 and 6. Northern blot analysis using a KCC2-specific cDNA probe revealed a very highly expressed approximately5.6-kilobase transcript only in brain. Reverse transcriptase-polymerase chain reaction revealed that KCC1 was present in rat primary astrocytes and rat C6 glioma cells but that KCC2 was completely absent from these cells, suggesting KCC2 was not of glial cell origin. In situ hybridization studies demonstrated that the KCC2 transcript was expressed at high levels in neurons throughout the central nervous system, including CA1-CA4 pyramidal neurons of the hippocampus, granular cells and Purkinje neurons of the cerebellum, and many groups of neurons throughout the brainstem.

The ability of cells to maintain water and electrolyte homeostasis is often dependent upon cation-chloride cotransport mechanisms. These cotransporters directly couple the movement of cations and Cl Ϫ ions in an electrically neutral fashion, and therefore, net ion movement is dependent solely on the chemical gradients of the transported ions. Two such cationchloride cotransporters (CCC) 1 include the Na-K-Cl and K-Cl cotransporters. Both of these transporters are sensitive to the sulfamoylbenzoic acid "loop" diuretics furosemide and bumetanide; however, the K-Cl cotransporter has a much lower affinity for these drugs than the Na-K-Cl cotransporter (1).
The K-Cl cotransporter was first described as a swellingactivated K ϩ efflux pathway in avian red blood cells (2). It has subsequently been described in numerous single cells, where it functions to reduce cell volume after swelling by promoting an efflux of K ϩ and Cl Ϫ and osmotically obliged water (for recent review, see Ref. 3). The mechanism and regulation of K-Cl cotransport has been extensively studied in vertebrate red blood cells. For example, the direct coupling and 1:1 stoichiometry of K ϩ and Cl Ϫ fluxes have been clearly established in human red blood cells (4,5), and a net dephosphorylation event has been shown to be necessary for activation of the red blood cell K-Cl cotransporter (6 -10). A K-Cl cotransporter has also been reported in secretory and absorptive epithelia, where, in addition to regulating cell volume, it may participate in net transepithelial salt and water movement (11)(12)(13)(14)(15).
In addition to volume regulation and net transepithelial salt movement, the K-Cl cotransporter appears to carry out a unique function in neurons: the maintenance of low intracellular [Cl Ϫ ] ([Cl Ϫ ] i ). Evidence supporting the role of a K-Cl cotransporter as a Cl Ϫ efflux pathway maintaining low [Cl Ϫ ] i has been reported in both invertebrate and vertebrate neuronal preparations (16 -22). Chloride-dependent synaptic inhibition in many neurons relies upon the maintenance of low [Cl Ϫ ] i and a favorable inwardly directed Cl Ϫ electrochemical gradient. The inhibitory postsynaptic potential (IPSP) elicited by inhibitory neurotransmitters such as ␥-aminobutyric acid (GABA) and glycine is dependent upon the conductive movement of anions through a ligand-gated channel in the membrane. For example, the GABA A receptor functions as an anion channel that is highly selective for Cl Ϫ . The IPSP results from an influx of Cl Ϫ through the ligand-gated channel that causes a hyperpolarization of the membrane, and thus drives the membrane potential away from its threshold value. In order to prevent the collapse of the Cl Ϫ electrochemical gradient, an "active" Cl Ϫ extrusion mechanism must be present in these GABA-and glycine-sensitive neurons. As a secondary active transport mechanism, the K-Cl cotransporter appears to carry out this important transport function in neurons.
In previous studies, we and others have reported the cloning, sequencing, and functional expression of multiple isoforms of the Na-K-Cl cotransporter in secretory (NKCC1, also called BSC2; Refs. [23][24][25] and absorptive epithelia (NKCC2, also called BSC1; Refs. 26 and 27). These NKCC isoforms along with the thiazide-sensitive Na-Cl cotransporter (NCC, also called TSC; Ref. 28) make up a new gene family of membrane transport proteins. Based on similarities in substrate transport and drug inhibition, the K-Cl cotransporter is a putative member of this gene family of transport proteins (29,30). Assuming that the K-Cl cotransporter would be related to the CCC gene family, we employed a cloning strategy based on data base searching, PCR, and library screening to identify cDNAs displaying homology to NKCC. Significantly, two closely related but distinctly different gene products were identified in rat brain that are distantly related to NKCC and NCC. The molecular and functional characterization of one of these gene products (KCC1) as a K-Cl cotransporter is presented in the accompanying article (31). In this report, we present the molecular characterization of the second gene product as a putative K-Cl cotransporter isoform (KCC2) that is expressed at very high levels in the brain and is specifically localized in neurons. Based on its high expression and neuronal specificity, we propose that this putative K-Cl cotransporter isoform is responsible for the maintenance of low [Cl Ϫ ] i in neurons and the proper function of Cl Ϫ -sensitive channels (inhibitory neurotransmitter receptors) in postsynaptic inhibition.

EXPERIMENTAL PROCEDURES
Polymerase Chain Reaction-Expressed sequenced tags (EST) from human brain (T16107) and human breast (H15821) were identified in the GenBank™ data base using the BLAST search programs (32) and the human colonic Na-K-Cl cotransporter (hNKCC1) as the queried sequence. Using sequence information from these human ESTs, forward (5Ј-CATCCTCATCGCCTCCCTCG-3Ј) and reverse (5Ј-CGTCAC-CCCACTCCTTCTCAGC-3Ј) oligonucleotide primers were synthesized and used to amplify a 286-bp fragment from double-stranded cDNA prepared from 1 g of poly(A) ϩ RNA isolated from whole rat brain (Marathon cDNA amplification kit, Clontech). The reaction volume of 50 l contained 5 l of a 1:250 dilution of the double-stranded rat brain cDNA, 0.2 M of each primer, 50 mM Tris-HCl (pH 9.2, 25°C), 16 mM (NH 4 ) 2 SO 4 , 1.75 mM MgCl 2 , 0.2 mM dNTP and 3 units of Expand™ long template DNA polymerase (Boehringer Mannheim). Twenty-eight cycles of PCR were performed (each consisting of incubation of 30 s at 94°C, 30 s at 60°C, and 2 min at 68°C). The 286-bp PCR product from five reaction vials was combined, precipitated, and isolated on a 1.5% agarose gel. After purification with glass beads (Geneclean), the 286-bp PCR product was cloned into the plasmid pCRII (Invitrogen) and sequenced in both directions.
Cloning and Sequencing Analysis-A rat (Sprague-Dawley) whole brain cDNA library in ZAP II was obtained from Stratagene (La Jolla, CA). Approximately 7.5 ϫ 10 5 plaques were screened with the 286-bp PCR product after labeling with [ 32 P]dCTP by random priming (Boehringer Mannheim). Plaque lifts were screened under low stringency conditions (16 h at 35°C in 50% formamide, 5 ϫ SSPE, 5 ϫ Denhardt's solution, 0.1% SDS, and 100 g/ml salmon sperm DNA) with a final stringent wash at 50°C in 0.5 ϫ SSC, and 0.1% SDS for 15 min. Positive clones were plaque-purified, excised into pBluescript SK Ϫ , and then characterized by 5Ј-and 3Ј-end sequencing, restriction mapping, and Southern blotting. Additional clones encoding the 5Ј-end of rtKCC1 and rtKCC2 were obtained by rescreening the library under similar conditions as above using PCR products derived from cDNA clones obtained from the initial screening (rtKCC1: nt 610 -950; rtKCC2: nt 546-1000).
The rtKCC2 cDNA was sequenced bidirectionally by the dideoxy chain termination method using a combination of manual sequencing with Sequenase II (U. S. Biochemical Corp.) and automated sequencing (Applied Biosystems Inc.) with synthetic oligonucleotide primers and fluorescent dideoxy terminators. Analysis of the nucleotide sequences and deduced amino acid sequence were performed with programs from the Genetics Computer Group software.
RNA Isolation and Northern Blot Analysis-Total RNA was isolated from fresh rat tissues and various cultured cells by the guanidinium thiocyanate method (34). Poly(A) ϩ RNA was purified from total RNA using magnetic beads (Poly(A)Ttract, Promega Corp.). Samples of rat tissue mRNA were denatured by heating to 65°C in formamide and formaldehyde and size-fractionated on a 1% agarose gel. Fractionated mRNA was transferred to a nylon membrane by semidry blotting. The rat tissue blot used in this study was the same as that used for the KCC1 Northern hybridization in Gillen et al. (31). This membrane was completely stripped of previously hybridized KCC1 cDNA probe by incubating in 0.1% SCC and 0.5% SDS at 75°C, and confirmation of probe removal was determined by film exposure (24 h at Ϫ70°C). A KCC2 isoform-specific cDNA probe was prepared from an 783-bp PstI-PstI (rtKCC2: nt 4668 -5451) fragment derived from the 3Ј-untranslated region contained in clone ERB10. The rat tissue Northern blot was prehybridized 4 h at 65°C in 0.25 M Na 2 HPO 4 (pH 7.2), 1 mM EDTA, and 7% SDS and then hybridized for 16 h in fresh hybridization solution containing 10 6 cpm/ml 32 P-labeled cDNA probe. The blot was subjected to a final stringent wash of 20 min at 58°C in 40 mM Na 2 HPO 4 (pH 7.2), 1 mM EDTA, and 1% SDS and analyzed by autoradiography at Ϫ70°C with an intensifying screen.
In Situ Hybridization-In situ hybridization was performed on sections from normal male Wistar rats. Rats were killed by decapitation, and the whole brain was rapidly removed, frozen on dry ice, and stored at Ϫ80°C, until processed for in situ hybridization, as described previously (35). A KCC2-specific probe similar to that used for Northern blot analysis, was generated by linearization of clone ERB10 with XhoI (rtKCC2: nt 4712-5451), and cRNA probe transcribed in vitro using T3 RNA polymerase (Boeringer Mannheim), [ 35 S]UTP (800 Ci/mmol, Amersham International, United Kingdom) and unlabeled nucleotides to a specific activity of 3-5 ϫ 10 8 Ci/mmol. In situ hybridization was performed on 10-m sagittal or coronal sections of brain mounted on poly-L-lysine subbed slides. Sections were postfixed in 4% paraformaldehyde in 0.1 M phosphate-buffered saline solution for 10 min and rinsed once in phosphate-buffered saline and three times in 2 ϫ SSC. All solutions were treated with diethylpyrocarbonate (0.02%) and autoclaved. 35 S-Labeled cRNA probes were denatured by heating at 70°C and added to hybridization mix (50% deionized formamide, 0.6 M NaCl, 10 mM Tris-Cl (pH 7.5), 0.02% Ficoll, 0.02% polyvinylpyrrolidone, 0.1% bovine serum albumin, 1 mM EDTA, 100 g/ml denatured salmon sperm DNA, 0.05 mg/ml yeast tRNA, 10% dextran sulfate, and 10 mM dithiothreitol) to give 10 7 counts/ml. Hybridization mix (200 l) was added to each slide, and hybridization was carried out overnight in sealed humid chambers at 50°C. After hybridization slides were rinsed in 2 ϫ SSC and treated with RNase A (30 g/ml) for 60 min at 37°C. Washes consisted of 2 ϫ SSC at room temperature and 0.1 ϫ SSC at 50°C for 60 min. Sections were then dehydrated in graded ethanols in 0.3 M ammonium acetate, air-dried, and exposed to ␤-max film (Amersham International) for 1 week for determination of gross localization. Slides were then dipped in K5 nuclear emulsion (Ilford, United Kingdom), exposed at 4°C for 3 weeks, developed in D19 developer (Eastman Kodak Co.), and counterstained with hemeatoxylin and eosin. Control sections were pretreated with RNase A (100 g/ml) for 60 min at 37°C prior to hybridization.

RESULTS
Cloning and Sequencing of KCC1 and KCC2 in Rat Brain-Two ESTs were identified in the GenBank™ data base that encoded a peptide ϳ35% identical to hNKCC1, spanning predicted transmembrane segments 9 -12 (see "Experimental Procedures"). Forward and reverse oligonucleotide primers were synthesized over this region and used to amplify a 286-bp PCR product from a population of double-stranded cDNAs prepared from whole rat brain. The 286-bp PCR product was cloned and sequenced and was 91% identical to the nucleotide sequence derived from the two human ESTs.
The 286-bp PCR product was 32 P-labeled by random priming and used to screen a whole rat brain cDNA library under low stringency conditions. Nineteen cDNA clones were identified and fully characterized from this initial screening. It was clear from restriction and sequence analysis that the clones represented two closely related but distinct gene products, KCC1 (7 cDNAs) and KCC2 (12 cDNAs). The molecular characterization of KCC1 is described in the accompanying article (31). In order to obtain the 5Ј-end of the coding regions of KCC1 and KCC2, the rat brain cDNA library was rescreened with PCR products derived from the 5Ј-ends of two of the initial cDNAs (rtKCC1-ERB24 and rtKCC2-ERB10). Ten additional cDNA clones were obtained from this second screening: KCC1 (2 cDNAs) and KCC2 (8 cDNAs). A full-length overlapping nucleotide sequence of 5566 bp was identified for rtKCC2 from clones ERB10 and 5ERB12 (Fig. 1). One full-length clone was obtained from the second library screening (5ERB14; Fig. 1).
The KCC2 cDNA includes a full-length open reading frame encoding a polypeptide of 1116 amino acids (Fig. 2). The initiating methionine was assigned to an in-frame ATG within a strong Kozak consensus site (GCCACCATGC) contained in the most 5Ј-end clone (5ERB12; Fig. 1).
The predicted molecular mass of KCC2 is 123.6 kDa. As shown in the sequence alignment (Fig. 2), the full-length amino acid sequence of KCC2 is 67% identical to KCC1, yet displays much lower identity (ϳ25%) to other members of the CCC gene family that includes both isoforms of the bumetanide-sensitive Na-K-Cl cotransporter (NKCC1 and NKCC2) and the thiazidesensitive Na-Cl cotransporter (NCC).
Structure of KCC2-Hydropathy analysis of KCC2 predicts a large central hydrophobic region bounded by N-and C-terminal hydrophilic domains (Fig. 3A), similar to the structure predicted for the previously identified CCC proteins (NKCC and NCC). Characteristic of this gene family, we predict 12 membrane-spanning ␣-helices and the placement of both the hydrophilic N terminus and large C terminus within the cytoplasm (Fig. 3B). Eleven potential N-linked glycosylation sites are present within the polypeptide, and four of these sites are located on the predicted extracellular loop between putative transmembrane segments 5 and 6 ( Figs. 2 and 3). There are no consensus phosphorylation sites for protein kinase A; however, five consensus phosphorylation sites for protein kinase C (Thr 34 , Ser 728 , Thr 787 , Ser 940 , and Ser 1034 ) as well as one consensus tyrosine protein kinase phosphorylation site (Tyr 1081 ) are located within the hydrophilic domains at the N or C terminus of the protein.
The N terminus is the most divergent region between the two KCC isoforms, and KCC2 is truncated 21 amino acids at this end of the primary sequence when compared to KCC1 (Fig. 2). Except for this truncation, as well as a single amino acid insertion in KCC2 within the predicted extracellular loop between putative transmembrane segments 5 and 6 and 51 amino acids that are inserted in KCC2 within the predicted intracellular C-terminal domain, KCC1 and KCC2 align very well over their entire length. The central core hydrophobic domain containing the 12 putative transmembrane (TM) segments is the region that displays the highest identity among the two KCC isoforms (Fig. 4A). Seven of the putative TM segments share Ն90% identity; these include: TM1, TM3, TM6, TM8, TM10, TM11, and TM12. In addition, a number of regions predicted to be outside the TM segments are very well conserved between KCC1 and KCC2, including a predicted intracellular intermembrane loop between TM2 and TM3 as well as a few small regions within the large hydrophilic C-terminal domain (Figs. 2 and 4A). Interestingly, the predicted intermembrane loop between TM2 and TM3 is the most highly conserved region among all members of the CCC gene family, and the high identity displayed by the KCC isoforms for the NKCC and NCC proteins over this region clearly places these new KCC proteins within the CCC gene family (Figs. 2 and 4B).
Tissue Expression of KCC2-The expression of KCC2 in various rat tissues was examined by Northern blot analysis (Fig.  5). An isoform-specific cDNA probe was prepared that contained sequence from the 3Ј-untranslated region of the rtKCC2 cDNA and displayed no significant identity to the rtKCC1 cDNA. This cDNA probe hybridized with a ϳ5.6-kilobase transcript in brain only (Fig. 5). The KCC2 transcript was not detected in any other tissue even after a longer exposure time (48 h at Ϫ70°C; data not shown). Thus, the KCC2 transcript is specific for brain tissue. In other experiments, we also observed this same ϳ5.6-kilobase transcript exclusively in brain following hybridization with cDNA probes prepared from the more highly conserved coding regions of rtKCC1 and rtKCC2 (data not shown). Significantly, the level of KCC2 expression in the brain was very high. A strong signal for the brain KCC2 transcript was observed after only a 4-h exposure of the Northern blot as shown in Fig. 5.
In an attempt to examine the distribution of KCC1 and KCC2 expression within different cell types derived from the rat central and peripheral nervous system, we used RT-PCR with isoform-specific oligonucleotide primers (rtKCC1 and rt-KCC2) and total RNA isolated from whole rat brain, rat primary astrocytes, a rat glioma cell line (C6 cell), and a rat pheochromocytoma cell line (PC12 cell). As shown in Fig. 6, each of the tissue samples displayed a RT-PCR product of the correct size (233 bp) in the reaction containing primers specific for the widely distributed KCC1 isoform; however, only whole rat brain RNA exhibited a RT-PCR product of the correct size (399-bp) in the reaction containing primers specific for the KCC2 isoform.
In Situ Hybridization of KCC2-In order to examine in greater detail the distribution of the KCC2 transcript within the rat brain, we used in situ hybridization with a KCC2specific 35 S-labeled riboprobe. Film autoradiographic analysis of sagittal brain sections hybridized with the KCC2-specific riboprobe showed strong expression of mRNA in all layers of the cortex, all areas of the hippocampus and the granular layer of the cerebellum (Fig. 7A). Expression is also evident in areas of the brainstem. Signal was not detected in sections treated with RNase A before hybridization (data not shown).
Microautoradiography of the brain sections showed specific hybridization signal in neurons in many areas of the brain (Fig.  7, B-E). White matter in the corpus callosum was devoid of signal, supporting a neuronal localization for KCC2 mRNA. High expression of KCC2 mRNA was evident in all layers of the cortex, all regions of the hippocampus (CA1-CA4 pyramidal neurons, and dentate gyrus), and Purkinje neurons of the cerebellum. Granular cells in the cerebellum also expressed KCC2 mRNA but at lower levels than the Purkinje neurons. Addi-tionally, many groups of neurons throughout the brainstem showed high expression of KCC2 mRNA. DISCUSSION This paper describes the molecular characterization of a cDNA from rat brain, which is proposed to encode a neuronalspecific isoform of the K-Cl cotransporter (KCC2). While KCC2 has not yet been functionally expressed, over the full length it displays high amino acid identity (67%) to the K-Cl cotransporter (KCC1) identified and functionally characterized in the accompanying article (31). Together, these two KCC proteins share distant relationship (ϳ25% amino acid identity) to a gene family of cation-chloride cotransporter proteins (see Fig. 2 in Ref. 31), which currently includes two isoforms of the bumetanide-sensitive Na-K-Cl cotransporter (NKCC1 and NKCC2; FIG. 2. Sequence alignment of the deduced primary structure of K-Cl cotransporters in rat brain (rtKCC2 and rtKCC1), Na-K-Cl cotransporters in human colon (hNKCC1) and rat kidney (rtNKCC2; F-variant), and thiazide-sensitive Na-Cl cotransporter in rat kidney (rtNCC). The 12 putative transmembrane segments are underlined below the alignment. Above the alignment, predicted N-linked glycosylation sites are identified by bars, potential phosphorylation sites for protein kinase C by crosses (ϩ), and one potential phosphorylation site for tyrosine kinase by double crosses ( ‡).
Analysis of the primary structures of the KCC proteins indicates that overall they are similar to the other members of the CCC gene family, i.e. hydrophilic N-and C-terminal regions bounding a central core hydrophobic domain containing 12 putative transmembrane segments. As expected for a family of FIG. 3. A, hydropathy profile of the putative K-Cl cotransporter from rat brain (rtKCC2). The hydropathy index was determined by the Kyte and Doolittle algorithm (49) using a 15-amino acid window. Above the hydropathy plot is a linear hypothetical model of the rtKCC2 protein indicating the transmembrane region. The 12 putative transmembrane segments are indicated by numbers above the line. B, proposed model of the putative K-Cl cotransporter from rat brain (rtKCC2). Circles symbolize amino acid residues. A potential site for tyrosine kinase phosphorylation is highlighted in black, and branched lines specify those asparagines presumed to be capable of anchoring oligosaccharide. membrane proteins, the central hydrophobic domain containing the TM segments is the region displaying the highest identity among all of the CCC proteins, including KCC1 and KCC2. Significantly, the KCC proteins display high identity to hNKCC1 (ϳ80%) over a short 28-amino acid intermembrane loop between putative TM segments 2 and 3, which is the most highly conserved region among all members of the CCC gene family. The function of this highly conserved region of these proteins is unknown, but it is significant that in the renal Na-K-Cl cotransporter (NKCC2) a portion of this intermembrane loop is encoded by a cassette exon that is alternatively spliced in a mutually exclusive fashion, leading to three distinctly different splice variants of this protein (26,39).
The KCC proteins do exhibit a distinct structural feature that sets them apart from the previously identified CCC proteins. KCC1 and KCC2 both have a large predicted extracellular intermembrane loop harboring four N-linked glycosylation sites between putative TM segments 5 and 6. In the Na ϩ -dependent CCC proteins (i.e. NKCC and NCC) the extracellular intermembrane loop containing glycosylation sites is predicted between TM segments 7 and 8 (23,(25)(26)(27)(28).
Previous molecular studies of the CCC proteins have noted that the N terminus is the region of greatest divergence among the different proteins (NKCC to NCC) and also among isoforms of the same protein (NKCC1 to NKCC2; Refs. 25 and 26). This divergence at the N terminus is also observed between the two KCC isoforms, where they display only 51% amino acid identity and an additional 21 amino acids are found at the N terminus of KCC1. We have suggested that this N-terminal variability in the Na-K-Cl cotransporter indicates that this region is not likely to be directly involved in ion translocation (25). In direct contrast to the N-terminal region, the large hydrophilic C terminus is more highly conserved between the KCC isoforms (70% identical). The K-Cl cotransporter of red blood cells has been shown to be regulated by phosphorylation-dephosphorylation events (6 -10); however, the kinase-phosphatase enzymes involved have not yet been clearly identified. A number of protein kinase C consensus sites are localized within the C terminus of the KCC proteins, but only one is common to both isoforms (rtKCC1-Thr 748 and rtKCC2-Ser 728 ). Additionally, KCC2 has a tyrosine kinase consensus phosphorylation site that is not present in KCC1. Significantly, in comparing the KCC isoforms, KCC2 has a large insertion within the C terminus, which is exceedingly rich in serines and negatively charged amino acids (Asp and Glu) and also contains a potential protein kinase C phosphorylation site (Ser 940 ). Based on these findings, KCC1 and KCC2 may display very distinct differences in their regulation. Northern blot analysis was used to examine the tissue dis- In upper panel, tissue mRNA (5 g each) was hybridized with a 783-bp cDNA probe specific for the KCC2 transcript (see "Experimental Procedures"). The blot was rehybridized with a ␤-actin cDNA probe as a control for RNA integrity (lower panel). The blot was exposed for 4 h (KCC2 probe) or 6 h (␤-actin probe) at Ϫ80°C with an intensifying screen. The second band in some lanes of the ␤-actin panel reflects a muscle-specific actin transcript of ϳ1.8 kilobases.
FIG. 6. Expression of KCC1 and KCC2 in various cell lines derived from rat nervous system. Reverse transcriptase-PCR was performed using total RNA from whole rat brain, rat C6 glioma cells, rat PC12 pheochromocytoma cells, and rat primary astrocytes. Oligonucleotide primers specific for KCC1 and KCC2 were used to amplify a 233-bp (KCC1) or 399-bp (KCC2) product. Negative controls for both sets of primers containing no template cDNA are shown at far left along with the DNA markers (values in base pairs). The ethidium bromidestained gel is depicted as an inverse image. tribution of the KCC2 transcript in the rat. In direct contrast to KCC1, which displays a very wide tissue distribution in rat (31), the KCC2 transcript was found only in brain. Consistent with a specific distribution of KCC2 within the nervous system is the finding that highly homologous nucleotide sequences have been identified in the GenBank™ EST data base that are derived either from whole brain or retinal tissues. As the K-Cl cotransporter has been described in various cells and tissues, we propose that KCC1 is likely the "housekeeping" form of this protein responsible for cell volume regulation (31). Reverse transcriptase-PCR was used to examine the distribution of KCC1 and KCC2 in rat primary astrocytes, rat C6 glioma cells, and rat PC12 pheochromocytoma cells. Each of these cells displayed a RT-PCR product of the correct size only for the oligonucleotide primers specific for KCC1. Only the whole rat brain RNA exhibited correctly sized RT-PCR products for both the KCC1 and KCC2-specific primers. These data support the role of KCC1 as the "housekeeping" isoform and indicate that KCC2 probably is not of glial cell origin.
In situ hybridization with a 35 S-labeled antisense cRNA probe specific for the KCC2 transcript demonstrated that the KCC2 transcript was found in neurons throughout the rat FIG. 7. In situ hybridization detection of KCC2 mRNA in rat brain. A, film autoradiograph of 10-m section of rat brain hybridized with a KCC2-specific riboprobe (see "Experimental Procedures") showing widespread distribution of KCC2 mRNA throughout the brain. Scale bar,ϭ 3 mm. Cort, cortex; Hipp, hippocampus; Cb, cerebellum; Bs, brainstem. B and C, dark field photomicrographs of hippocampus. B shows field denoted by solid box in A. Hybridization of probe to the whole hippocampus is shown by silver grains overlying the neuronal cell bodies. C shows a higher magnification of neurons in the dentate gyrus. D and E, dark field photomicrographs of cerebellum (dotted box in A). D shows a field containing both granular and Purkinje neurons (arrows). Note the high level of expression in the Purkinje neurons surrounding the granular neurons. E shows a higher power view of Purkinje neurons and granular neurons. Scale bars represent 100 m (C and E) and 500 m (B and D).
central nervous system, including pyramidal neurons of the hippocampus and Purkinje neurons and granular cells of the cerebellum. The fact that the white matter of the corpus callosum was devoid of signal is consistent with a neuronal-specific localization. The importance of this putative electroneutral cotransporter in neurons is underscored by its tremendous expression throughout the entire central nervous system as evidenced by Northern blot analysis (Fig. 5) and by in situ hybridization (Fig. 7A).
Physiological Function of K-Cl Cotransport in Neurons-The K-Cl cotransporter is a major Cl Ϫ efflux pathway in a number of cells, where it participates in cell volume regulation and net transepithelial salt movement. In neurons, the K-Cl cotransporter has been suspected to function as a "Cl Ϫ pump," maintaining low [Cl Ϫ ] i and a favorable inwardly directed Cl Ϫ electrochemical gradient (for review see Ref. 40). The importance of this function of K-Cl cotransport in neurons is clearly illustrated when one considers that the action of the inhibitory neurotransmitters, GABA and glycine, is dependent upon an influx of Cl Ϫ through ligand-gated channels, leading to the generation of the IPSP. Early work using microelectrodes and the giant neuron of the Aplysia abdominal ganglion clearly showed that E Cl was less negative than the membrane potential (E m ), demonstrating that Cl Ϫ must be actively extruded in these cells (41,42). Although reports have suggested the involvement of a primary active transport mechanism (i.e. Cl Ϫ -ATPase; 43, 44), Cl Ϫ movement against an electrochemical gradient in animal cells has often been attributed to electroneutral cation-chloride cotransport mechanisms where the electrical gradient is no longer a factor (e.g. Na-K-Cl cotransport in the thick ascending limb of mammalian kidney). In fact, evidence supporting the role of K-Cl cotransport as the Cl Ϫ efflux pathway in neurons has been reported in both invertebrate and vertebrate systems. The crayfish stretch receptor neuron (16,17), insect neurosecretory cells (22), and vertebrate cortical and hippocampal neurons (18 -21) appear to extrude Cl Ϫ against an electrochemical gradient by coupling Cl Ϫ transport to the favorable outwardly directed K ϩ chemical gradient, i.e. K-Cl cotransport. These studies have indirectly characterized this transport pathway by examining shifts in E Cl (as estimated from E GABA or E IPSP ) following alterations in [K ϩ ] o or furosemide treatment.
It is interesting to note that in some neurons GABA causes a membrane depolarization with an efflux of Cl Ϫ through the GABA A receptor (45)(46)(47). In these neurons, elevated [Cl Ϫ ] i is required (E Cl Ͼ E m ), and it is the Na-K-Cl cotransporter that appears to be involved in maintaining this high [Cl Ϫ ] i (for review, see Ref. 40). Under normal physiological conditions, the ion chemical gradients are such that Na-K-Cl cotransport operates in a net inward direction, and this cotransporter is known to participate in maintaining high [Cl Ϫ ] i in epithelial cells of Cl Ϫ -secreting tissues for the proper function of apical Cl Ϫ channels (48). Thus, it appears that electroneutral cationchloride cotransport processes (i.e. K-Cl and Na-K-Cl cotransporters) play an important role in regulating [Cl Ϫ ] i in neurons and determining the response of neurons to inhibitory neurotransmitters. The use of electroneutral transporters to maintain [Cl Ϫ ] i in neurons has obvious benefits for an excitable cell.
In the present study, we have identified a putative K-Cl cotransporter that is specifically localized to neurons and is highly expressed throughout the central nervous system. We proposed that this putative neuronal K-Cl cotransporter may be the so-called "Cl Ϫ pump" of neurons that maintains low [Cl Ϫ ] i for the proper function of ligand-gated Cl Ϫ -sensitive channels (GABA A and glycine receptors) in postsynaptic inhibition.