点击显示 收起
【关键词】 heterogeneity
1Renal Division, Brigham and Women’s Hospital, 2Division of General Internal Medicine, Veterans Affairs Boston Healthcare System, Harvard Medical School, Boston, Massachusetts
Departments of 3Medicine and 4Anesthesiology, Vanderbilt University Medical Center, Nashville, Tennessee
5Molecular Physiology Unit, Instituto Nacional de Ciencias Medicas y Nutricion Salvador Zubirán and Instituto de Investigaciones Biomedicas, Universidad Nacional Autonoma de Mexico, Mexico City, Mexico
ABSTRACT
The SLC12A6 gene encoding the K+-Cl cotransporter KCC3 is expressed in multiple tissues, including kidney. Here, we report the molecular characterization of several NH2-terminal isoforms of human and mouse KCC3, along with intrarenal localization and functional characterization in Xenopus laevis oocytes. Two major isoforms, KCC3a and KCC3b, are generated by transcriptional initiation 5' of two distinct first coding exons. Northern blot analysis of mouse tissues indicates that KCC3b expression is particularly robust in the kidney, which also expresses KCC3a. Western blotting of mouse tissue using an exon 3-specific antibody reveals that the kidney is also unique in expressing immunoreactive protein of a lower mass, suggestive evidence that the shorter KCC3b protein predominates in kidney. Immunofluorescence reveals basolateral expression of KCC3 protein along the entire length of the proximal tubule, in both the mouse and rat. Removal of the 15-residue exon 2 by alternative splicing generates the KCC3a-x2M and KCC3b-x2M isoforms; other splicing events at an alternative acceptor site within exon 1a generate the KCC3a-S isoform, which is 60 residues shorter than KCC3a. This variation in sequence of NH2-terminal cytoplasmic domains occurs proximal to a stretch of highly conserved residues and affects the content of putative phosphorylation sites. Kinetic characterization of KCC3a in X. laevis oocytes reveals apparent Kms for Rb+ and Cl of 10.7 ± 2.5 and 7.3 ± 1.2 mM, respectively, with an anion selectivity of Br > Cl > PO4 = I = SCN = gluconate. All five NH2-terminal isoforms are activated by cell swelling (hypotonic conditions), with no activity under isotonic conditions. Although the isoforms do not differ in the osmotic set point of swelling activation, this activation is more rapid for the KCC3a-x2M and KCC3a-S proteins. In summary, there is significant NH2-terminal heterogeneity of KCC3, with particularly robust expression of KCC3b in the kidney. Basolateral swelling-activated K+-Cl cotransport mediated by KCC3 likely functions in cell volume regulation during the transepithelial transport of both salt and solutes by the proximal tubule.
K-Cl cotransporter; cell volume; SLC12
THE ELECTRONEUTRAL MOVEMENT of K+ and Cl across the plasma membrane is accomplished by a group of secondary active transport proteins known as K+-Cl cotransporters (KCCs). This transport system was originally described in red blood cells as a swelling- and N-ethylmaleimide (NEM)-activated K+ efflux mechanism (16, 34), with a role in the regulation of cell volume. K+-Cl cotransport is clearly very widespread, with evidence for its presence in kidney (43), neurons (49), vascular smooth muscle (1), endothelium (52), epithelia (43), heart (60), and skeletal muscle (63). K+-Cl cotransport is mediated by four members of the SLC12 electroneutral cation-chloride cotransporter gene family, denoted KCC1 (18), KCC2 (50, 55), KCC3 (23, 44, 53), and KCC4 (44) (SLC12A4, SLC12A5, SLC12A6, and SLC12A7, respectively). The four predicted proteins are 6575% identical and share a common predicted topology, with a large glycosylated extracellular loop between transmembrane domains 5 and 6 (44). Whereas KCC1, KCC3, and KCC4 are expressed in several tissues (18, 23, 44, 53), KCC2 is neuron specific (49, 50). Physiological roles are beginning to emerge from the comprehensive analysis of each isoform, encompassing tissue localization (38, 51), functional characterization (40, 55, 59), genetic analysis (25, 57), and gene targeting (25, 26, 64). KCC2, in particular, functions as a crucial regulator of neuronal chloride, with dramatic downstream effects on the neuronal response to GABA and glycine (49). Dysfunction of KCC2 has thus been variously implicated in epilepsy (49, 64), hypoxic and traumatic brain injury (49), and the genesis of neuropathic pain (13).
KCC3 also plays a fundamental role in the development and function of the central and peripheral nervous system, given that inactivating mutations in SLC12A6 are the cause of "motor and sensory neuropathy with agenesis of the corpus callosum" (ACCPN) (25). KCC3 is expressed in multiple tissues outside of the nervous system, with particularly abundant transcript in heart, muscle, kidney, and placenta (23, 44, 53). Moreover, the KCC3 protein is expressed in red blood cells (36), suggesting that it may encode the major red cell K+-Cl cotransporter. Although this suggests a function for KCC3 in red cell volume regulation, in both health and disease (9), the role of this and other KCCs in the kidney is as yet unclear. There is, however, a significant body of evidence for basolateral swelling-activated K+-Cl cotransport in the proximal nephron, with proposed roles in cell volume regulation and/or transepithelial salt and solute transport (43). Targeted deletion of KCC4 results in a renal tubular acidosis, suggesting other roles for this paralog in distal nephron function (6).
KCC3 is unique among the KCCs in having two distinct NH2-terminal domains, generated by the use of two different first coding exons, 1a and 1b (23, 25, 44); we have defined KCC3a isoforms as those utilizing exon 1a and KCC3b isoforms as those utilizing exon 1b (25). We present the cloning and functional characterization of these and other NH2-terminal isoforms of human and mouse KCC3, along with immunolocalization of KCC3 in rodent kidney. We utilize the label "isoform" herein to denote alternative forms of KCC3 transcripts and proteins, rather than using this term to distinguish the four KCC paralogs.
METHODS
Cloning of mouse KCC3a and KCC3b. We used a stepwise approach to clone mouse KCC3 cDNAs by RT-PCR, initially with human KCC3 oligonucleotides. The first primer pair utilized, KC31S/1A (see Table 1), is by coincidence identical to the mouse sequence; subsequent RT-PCR of mouse kidney utilized mouse sense primers with human anti-sense primers to march along the open reading frame (see Table 1 and Fig. 1). The cloning of two different human KCC3 cDNAs, herein denoted KCC3a (44) and KCC3b (23), was independently reported in 1999 (see also RESULTS); both isoforms were cloned in mouse. The primer pairs utilized were as follows: KC32S/2A for the 5' end of KCC3a (codons 56507), KC33S/3A for codons 1459, KC34S/4A for codons 428907 of KCC3 (numbered for KCC3a), and KC35S/5A for codons 8311150. The 3'-UTR was then identified within the IMAGE EST #1209431, which was sequenced in entirety. Concurrent screening of a FIXII genomic library with human exon 1a and 1b probes yielded genomic clones containing the corresponding murine exons; sequencing of genomic subclones determined the complete exon 1a and 1b sequences, along with 3 kb of 5'-flanking DNA for each exon. We also utilized 5'-RACE RT-PCR to map the transcriptional start sites for mouse KCC3b; 5'-RACE RT-PCR was not successful with KCC3a anti-sense primers, in both mouse and human. Mouse kidney RACE-ready template (Clontech) was amplified with the KC36A anti-sense primer and the sense adaptor primer KC36S, using AmplitaqGold (PerkinElmer) and 35 cycles of 94°C for 1 min, 68°C for 2.5 min, and 72°C for 10 min. The amplified DNA was subcloned into pCR2.1 by TA cloning (Invitrogen). Sixty recombinant clones were then screened by PCR, using sequentially more 5' sense primers, and then sequenced.
For Northern blot analysis of mouse KCC3, 10 μg/lane of total RNA from several tissues was resolved by electrophoresis (5% formaldehyde, 1% agarose), transferred to a nylon membrane (Stratagene), and probed with 32P-labeled randomly primed probes specific for exon 1a and exon 1b. Hybridization was performed at 50°C in 4x SSCP/40% formamide/4x Denhart’s solution/0.5% SDS/200 μg salmon sperm DNA, and membranes were washed twice for 20 min at room temperature in 2x SSCP/0.1% SDS, and twice for 1 h at 65°C in 0.5x SSCP/0.1% SDS.
Identification and cloning of NH2-terminal isoforms of KCC3. For RT-PCR cloning of the various human 5' isoforms, poly-A+ RNA from multiple human tissues was purchased from Clontech and reverse transcribed as described (44). The human KCC3b 5' end was cloned from human brain by RT-PCR, using the KC37S and KC33A primer pair, followed by TA cloning into pCR2.1. The KCC3a- and KCC3b-specific sense primers KC38S and KC39S, each paired with an anti-sense primer from within exon 4 (KC37A), were utilized for RT-PCR of the KCC3b and KCC3a isoforms in both mouse and human tissues. PCR products differing in 45 bp were observed after RT-PCR of the 5' end of both KCC3a and KCC3b (see RESULTS); the lower MW fragments were subcloned into pCR2.1 and sequenced. Subsequent blastn searches of the human EST database revealed the existence of multiple 5' ESTs from testis and brain, with 5' ends distinct from full-length KCC3a but incorporating part of exon 1a (see RESULTS). The corresponding ESTs (IMAGE clone nos. 4826538, 4829214, 5164098, 5269073, and 5298663) were obtained from Research Genetics and partially sequenced. To verify that the corresponding variants of human KCC3 (KCC3a-S1 and KCC3a-S2, see RESULTS) were expressed in vivo, RT-PCR of multiple tissues was performed using specific primer pairs. These included the KCC3a-S1 sense primer KC310S paired with an anti-sense primer within exon 1a (KC38A), and the KCC3a-S2 sense primer KC311S paired with anti-sense primers within exon 3 (KC39A) and exon 4 (KC310A). Again, DNA fragments of the appropriate size were subcloned and identity was confirmed by sequencing.
Generation of human KCC3 expression constructs. Functional expression of full-length mouse KCC3 in Xenopus laevis oocytes has not been successful (data not shown), hence our functional characterization of KCC3 was restricted to the human ortholog. Our previous publications reported the functional characteristics of mouse KCC4, rabbit KCC1, and human KCC2, expressed in the context of the pGEMHE X. laevis expression vector (40, 44, 55); to generate a KCC3a-pGEMHE construct, the 5'-UTR of pGEMHE was transferred to a partial KCC3a 5' muscle cDNA (44) by PCR, adding a NheI site to the multicloning site. This 5' cDNA was then subcloned into pGEMHE using SacI and EcoRI. The TIGR clone 150620 and an intervening 5'-RACE clone (44) were joined in pBluescript using KpnI and HindIII, and then joined to the 5' end in pGEMHE using XhoI and EcoRI. All sequence generated by PCR was resequenced to confirm identity with the KCC3a sequence (44). To generate the KCC3b-pGEMHE construct, the NheI site in the 5' multicloning site of KCC3a-pGEMHE was modified to a SacII site, using QuikChange (Stratagene). The KCC3b 5' end cloned from human brain (see above) was then subcloned into KCC3a-pGEMHE using SacII and XhoI. The 3' end of KCC3 was subsequently subcloned back into this construct using XhoI and EcoRI, to replace segments generated by QuikChange. To generate an expression construct for KCC3a minus exon 2 (denoted KCC3a-x2M), the 5' end of KCC3a SacII-modified KCC3a expression construct was subcloned into pBluescript using SacI and XhoI. A 676-bp PCR fragment generated by RT-PCR of human brain with the KC38S and KC311A primers, missing exon 2, was trimmed with BamHI and MunI and subcloned into this 5' pBluescript subclone; this fragment was then subcloned into KCC3a-pGEMHE using SacII and XhoI, substituting the KCC3a-x2M 5' end for that of KCC3a. A similar strategy was used to generate an expression construct for KCC3b minus exon 2 (denoted KCC3b-x2M). Finally, to generate a KCC3a expression construct corresponding to the NH2-terminal truncated isoforms KCC3a-S1 and KCC3a-S2, which have identical open reading frames, we replaced the 5' end of the KCC3a expression construct with that of the 4826538 IMAGE clone, using SacII and BamHI. All sequences generated by PCR for these various constructs were resequenced, to confirm identity with the relevant sequences (23, 44).
Preparation of X. laevis oocytes. Adult female X. laevis frogs were purchased from NASCO (Fort Atkinson, MI) and maintained under constant control of humidity and room temperature at 65% and 16°C, respectively. Oocytes were surgically collected from anesthetized animals under 0.17% tricaine that were laid on ice during surgery; after several such procedures, frogs anesthetized with tricaine were killed by cardiac puncture. The use and care of the animals in these experiments were approved by the Institutional Animal Care and Use Committee of both institutions. After extraction, oocytes were incubated for 1 h under vigorous shaking in frog Ringer ND96 without calcium (in mM: 96 NaCl, 2 KCl, 1 MgCl, and 5 HEPES/Tris, pH 7.4, plus 2 mg/ml of collagenase A). Then, oocytes were washed four times in ND96, defolliculated by hand, and incubated overnight in ND96 at 17°C. On the next day, mature oocytes were injected with 50 nl of water containing 0.1 μg/μl of cRNA in vitro transcribed from each of the KCC3 cDNA isoforms. Oocytes were incubated during 4 days in ND96 at 17°C supplemented with 2.5 mM sodium pyruvate and 5 mg/100 ml of gentamicin. The incubation medium was changed every 24 h. The day of the experiment, oocytes were switched to a Cl-free ND96 (in mM: 96 Na+-isothiocyanate, 2 K+-gluconate, 6.0 Ca2+-gluconate, 1.0 Mg2+-gluconate, 5 mM HEPES, 2.5 sodium pyruvate, 5 mg% gentamicin, pH 7.4) for 2 h before the assay.
To prepare cRNA for injection, each clone was linearized at the 3' end using NotI and cRNA was transcribed in vitro, using the T7 RNA polymerase mMESSAGE kit (Ambion). RNA integrity was confirmed on agarose gels and concentration was determined by absorbance reading at 260 nm (DU 640, Beckman, Fullerton, CA) as well as by densitometry of the corresponding bands in agarose gel. cRNA was stored frozen in aliquots at 80°C until used.
Western blotting. Western blotting of KCC3 protein was performed using an affinity-purified rabbit polyclonal antibody specific for a 19-residue peptide (KKARNAYLNNSNYEEGDEY) encoded within exon 3 of the SLC12A6 gene (25); this epitope is 100% conserved in mouse KCC3. Positive control antibodies included polyclonal rabbit antibodies generated against KCC1 (35) and KCC2 (38) fusion proteins, in addition to a peptide-specific NH2-terminal KCC4 antibody (28). Antibodies were affinity-purified by linking the peptide or fusion protein antigens to Affi-Gel 15 support (Bio-Rad, Hercules, CA), followed by incubation of this complex with 1.5 ml of antiserum at 4°C overnight in 5x PBS. After several washes in 5x PBS, specific antibodies were eluted with 0.1 M Na-citrate, pH 2.5, neutralized with Tris 1 M, pH 8.8, and dialyzed overnight at 4°C in 1x PBS, using a 12- to 14-kDa molecular mass cutoff dialysis membrane (Spectrum, Gardena, CA). The purified antibody was then concentrated using a 30-kDa cutoff centriplus column (Amicon, Beverly, MA) and stored at 20°C.
Crude membrane protein was isolated from various mouse tissues, as described (42). Total cellular protein for Western blot analysis was also prepared from groups of 510 X. laevis oocytes injected with H2O or with cRNAs for each of the five KCC3s, rabbit KCC1, human KCC2, or mouse KCC4. Three days after injection, oocytes were transferred to Eppendorf tubes on ice and were lysed in 7.5 mM Na2HPO4, 1 mM EDTA (pH 7.4) supplemented with protease inhibitors. After the lysate of yolk and cellular debris was cleared by centrifugation at 750 g for 5 min, the supernatant was stored at 80°C. Western blotting was performed as described (42), and antigen-antibody complexes were visualized using enhanced chemiluminescence (ECL and ECL-Plus systems, Amersham Life Science).
Immunofluorescence. Mouse (C57BL/6J) and rat (Sprague-Dawley) kidney was processed for immunofluorescence as described (65). Tissue sections were incubated for 30 min in PBS with 1% BSA and 4% normal goat serum (NGS), and then incubated with primary antibodies diluted in PBS/BSA/NGS. An antigen retrieval step (8) was included, using 5 min of incubation in 0.1% SDS diluted in PBS/BSA/NGS. Control sections were incubated with preimmune serum or immune serum immunoabsorbed with antigen. Sections were washed with PBS/2.8% NaCl followed by standard PBS, and then incubated for 1 h with fluorescein-labeled (FITC) or Alexa 594-labeled anti-rabbit antibody. Dual staining utilized the following primary antibodies directed against antigens from specific proximal and distal nephron segments: anti-aquaporin-1 (gift of Dr. P. Agre), anti-NKCC2 (gift of Dr. S. Hebert), and anti-NCC (gift of Dr. S. Hebert). Dual-stained slides (2 rabbit primaries) were incubated with AffiniPure Fab fragment goat anti-rabbit at 1:100 dilution for 1 h before we proceeded to the second primary antibody (65). Sections were washed as above and mounted with Vectashield mounting medium (Vector Labs). Slides were examined with a Nikon Eclipse 800 research microscope, connected to an Optronics (DEI-750) CCD camera and video imaging system.
Measurement of K+-Cl cotransport. K+-Cl cotransport was assessed by measuring tracer 86Rb+ uptake (New England Nuclear) in experimental groups of at least 15 oocytes, as described previously (40, 44, 55). In brief, oocytes were exposed to a 30-min incubation period in a hypotonic K+- and Cl-free medium [in mM: 50 N-methyl-D-glucamine (NMDG)-gluconate, 4.6 Ca2+-gluconate, 1.0 Mg2+-gluconate, 5 HEPES/Tris, pH 7.4] with 1 mM ouabain, followed by a 60-min uptake period in hypotonic Na+-free media with varying concentration of K+-Cl-containing medium. The concentrations of K+ and Cl were independently varied, using NMDG-Cl or NMDG-gluconate, for a maximal total concentration of 50 mM. For example, an uptake solution with 50 mM K+-Cl contained no NMDG-Cl or NMDG-gluconate. All uptake solutions also contained 1.8 mM CaCl2, 1 mM MgCl2, and 5 mM HEPES, at pH 7.4, and were supplemented with 1 mM ouabain and 2.5 μCi 86Rb+. Given the known sensitivity of the cotransporter to extracellular osmolality and cell volume, experiments were also performed in isotonic conditions using the same solutions, supplemented with sucrose at 3.5 g/100 ml to reach isosmolar conditions for X. laevis oocytes (210 mosmol/kgH2O). Ouabain was added in both incubation and uptake periods to prevent 86Rb+ uptake via the Na+-K+-ATPase. All uptakes were performed at 32°C. At the end of the uptake period, oocytes were washed five times in ice-cold uptake solution without isotope to remove extracellular fluid tracer. Oocytes were dissolved in 10% sodium dodecyl sulfate and tracer activity was determined for each oocyte by -scintillation counting. To characterize ion transport kinetics, we performed experiments using varying concentrations of K+ and Cl. The sensitivity for furosemide was assessed by exposing groups of oocytes to the inhibitor at concentrations varying from 20 μM to 2 mM. The desired concentration of the inhibitor was present in both the incubation and uptake periods.
Statistical analysis. Statistical significance is defined as two-tailed P < 0.05 and the results are presented as means ± SE. The significance of the differences between groups was tested by one-way ANOVA with multiple comparisons using Bonferroni correction.
RESULTS
Cloning of mouse KCC3a and KCC3b. We previously reported the cloning of human KCC3a, mouse KCC4, and human KCC4 (44); human KCC3b was first identified by Hiki et al. (23). KCC3a and KCC3b refer to transcripts that utilize two alternative first coding exons, exons 1a and 1b, respectively, of the SLC12A6 gene encoding this transporter (25). Of note, many of the peptide regions in mouse KCC4 suitable for antibody targeting are poorly conserved in the human ortholog, which is otherwise 91% identical. Therefore, to aid in the generation of both KCC3-specific antisera and a KCC3 (Slc12a6) knockout mouse (25), we characterized the full-length sequences for murine KCC3a and KCC3b, using a combination of RT-PCR and genomic cloning. The predicted murine KCC3a and KCC3b proteins are 98.1 and 97.7% identical to their human orthologs and are identical in length (1,150 and 1,099 residues, respectively). A blastn search of the murine genomic sequence database identified the supercontig NW_000178, which encompasses the murine Slc12a6 gene on chromosome 2; the relative position of intron-exons boundaries is conserved in the human SLC12A6 and murine Slc12a6 genes (data not shown). As in the human SLC12A6 gene (25), the 1a and 1b coding exons are separated by 18 kb of genomic DNA. Exon 1b is, however, separated from exon 2 by 32 kb in the mouse gene, vs. 42 kb in human SLC12A6 (25); exons 2 and 3 are separated by 16 kb in mouse and 14 kb in human, and the remaining introns are considerably shorter in both mouse and human. Single bands are seen on both agarose (Fig. 1A) and acrylamide gels in a representative overlapping RT-PCR of the KCC3 open reading frame from kidney. This indicates that there are no major splicing events that alter transcript size between the end of the NH2 terminus and the extreme COOH terminus of the KCC3 proteins.
Northern blot analysis of several tissues with 1a- and 1b-specific probes indicates that KCC3b is particularly abundant in kidney, although KCC3a transcript is also detectable in this blot (Fig. 1B). Of note, the KCC3b transcript is 1 kb shorter than that of KCC3a, providing a potential identity for the different transcripts detected originally with a human KCC3 probe from within the 3'-UTR (44); the relative contribution of differences in the 5'- and/or 3'-UTR to this divergence in transcript size is however not yet known. Western blot analysis of various tissues with our exon 3-specific KCC3 antibody indicates the unique expression of a lower MW band in the kidney (Fig. 1C), supportive but not definitive evidence (see DISCUSSION) that the KCC3b protein predominates in the kidney. This antibody is highly specific for KCC3, capable of discriminating between this and the other three KCCs when expressed in X. laevis oocytes (Fig. 1D).
To verify that the KCC3b isoform is generated by transcriptional initiation at a distinct promoter flanking exon 1b, as opposed to alternative splicing of a transcript initiated 5' of exon 1a, we mapped the transcriptional start site by 5'-RACE. 5'-RACE RT-PCR was not successful for either mouse or human KCC3a. Mouse KCC3b 5'-RACE products were subcloned into the pCR2 vector, and sequencing analysis of progressively more 5' cDNA clones indicates that the transcription start site is 106 bp 5' of the exon 1b start codon (data not shown).
Identification and tissue distribution of NH2-terminal isoforms of KCC3. We identified a total of five different NH2-terminal isoforms of the human KCC3 protein (see Fig. 2C), with an analogous repertoire for mouse KCC3. These isoforms are generated by a combination of alternative transcriptional initiation sites flanking two different first coding exons (exons 1a and 1b), mRNA splicing at an internal acceptor site within exon 1a (see Fig. 2, A and B), and alternative splicing of exon 2. The use of two different first coding exons was apparent in the first reports describing the human KCC3 cDNAs (23, 44, 53) and was subsequently confirmed by our structural characterization of the human SLC12A6 gene (25).
RT-PCR analysis of the tissue distribution of both KCC3a and KCC3b (Fig. 3) using an anti-sense primer within exon 4 indicates the presence of two distinct bands, differing in the inclusion of exon 2. Exon 2 is a cassette exon, with an even number of codons, such that these transcripts predict an in-frame deletion of 15 residues between exon 1a or 1b and exon 3. This splicing event was detected in both human (Fig. 3, A and B) and mouse (Fig. 3, C and D); confirmation in both species was obtained by cloning and sequencing the various amplificons. In contrast to the Northern blot result for mouse KCC3b (Fig. 1B), KCC3b is detectable in multiple tissues by RT-PCR, hence this is not a renal-specific isoform. Deletion of exon 2 in KCC3b transcripts is evidently much less frequent than in KCC3a transcripts (Fig. 3). The variant isoforms generated by removal of exon 2 were denoted KCC3a-x2M and KCC3b-x2M.
The human exon 1a promoter is imbedded within a CpG island, as defined by GrailEXP analysis (http://compbio.ornl.gov/grailexp/), extending from 1,529 to 679 nucleotides 5' of the first start codon in exon 1a. BLAST searches using the CpG island identify a number of 5' human KCC3 ESTs, differing from human KCC3a (44) in splicing of the 5'-UTR and exon 1a. These alternative transcripts give rise to predicted proteins that are either identical to KCC3a or have a 59-residue truncation of the NH2 terminus ("KCC3-S," S for "short," see below). A group of EST cDNAs, collectively corresponding to the KCC3a-S1 isoform (GenBank AF531259), begin at 1,080 bp 5' of the KCC3a start codon (1,080 bp) and use a donor site at 652 bp for alternative splicing of the 5'-UTR (Fig. 2B). These KCC3a-S1 cDNAs indicate the use of an alternative acceptor site embedded within exon 1a, 5'-attccaggttt-3' (coding sequence included in the cDNA is underlined in bold), which is found at +80 bp relative to the KCC3a start codon (see Fig. 2B); the three potential KCC3a start codons (44) are thus deleted from these transcripts (see below). The 5' noncoding exon from this transcript is denoted exon 1c, following the nomenclature initiated for the first two coding exons 1a and 1b, along with the other 24 coding exons (25); the truncated version of exon 1a used in this transcript is denoted exon 1a (see Fig. 2, A and B). A second group of alternative KCC3a cDNAs, denoted KCC3a-S2 (GenBank AF531260), is evidently generated by a start site at 1,384 relative to the KCC3a start codon; these transcripts use a donor site at 1,281, with splicing to the alternative donor site at +80 bp. The 5' noncoding exon from this KCC3a-S2 transcript is denoted exon 1d, with the truncated exon 1a common to this and KCC3a-S1 denoted exon 1a (see Fig. 2).
Our RT-PCR analysis (Fig. 4) of the alternative 5' ends of KCC3a transcripts concentrated on the KCC3a-S1 and KCC3a-S2 isoforms, as they predict a novel NH2-terminal truncation of the KCC3a protein (see Fig. 2C). We find that the KCC3a-S1 transcript is expressed predominantly in testis, with weak expression elsewhere, whereas the KCC3a-S2 transcript appears to be exclusive to testis (see Fig. 4B). The predicted open frames of the KCC3a-S1 and KCC3a-S2 transcripts are identical, as the differing 5'-UTRs are spliced into the same acceptor site of exon 1a; this protein isoform is denoted KCC3a-S. There are three potential start codons for the full-length KCC3a open reading frame, as discussed previously (44); we provisionally utilize the first in-frame ATG as the start codon for KCC3a, although the second and third ATGs are arguably in the context of better Kozak sites (29). Regardless, all three start codons are 5' of the alternative splicing that generates KCC3a-S and are thus not available for translation of the KCC3a-S open reading frame. The first in-frame start codon in KCC3a-S is in a poor context for a Kozak translation start site, with C at position 3 and A at position +4 (29). However, a second in-frame ATG, corresponding to codon 4 of the KCC3a-S open reading frame, has a G at +4 and is perhaps preferred; we provisionally use the first in-frame ATG in predicting the KCC3-S protein. As noted before (44), the unique NH2 terminus of the KCC3a protein contains a cluster of potential serine phosphorylation sites. A number of these are removed in the KCC3a-S protein, including S41 and S47, potential phospho-acceptor sites for protein kinase C (PKC).
Alternative splicing of the KCC3a 5' end is complex, with at least three different acceptor sites within and flanking exon 1a. Sequence analysis of the 5269073 I.M.A.G.E. clone, which shares the same extreme 5'-UTR with KCC3a-S1, reveals use of a second alternative acceptor site ttgcagtcaa flanking exon 1a (see Fig. 2B); this site is however 72 bp 5' of the first KCC3a start codon and thus results in the same predicted open reading frame as KCC3a. RT-PCR of KCC3a-S1 using the KC310S and KC38A primers (Fig. 4A) indicates the presence of two amplified bands between 200 and 300 bp, with the lower band corresponding to the KCC3a-S1 transcript. Cloning and sequencing the upper band reveal an insertion of 66 bp, due to alternative splicing to a third acceptor site (tcctccagaaacc) that is 13 bp 3' of the first KCC3a start codon (see Fig. 2B); as the second and third potential KCC3a start codons (see above) are retained in this transcript, denoted KCC3a-S3, it is conceivable that the relevant open reading frame has the same NH2 terminus as the "full-length" KCC3a. Finally, one of the human KCC3 ESTs sequenced, IMAGE clone 5298663, has a 1,379-bp 5'-UTR that is an unspliced version of the flanking genomic DNA shown in Fig. 2B; two other reported EST sequences, GenBank BC051706 and BX648195, have a similarly unspliced 5'-UTR. It is thus evident that transcription and alternative splicing of KCC3a are complex, in that the various EST cDNAs indicate both alternative transcriptional start sites and alternative splicing of 5' noncoding exons. Characterization of transcriptional start sites by 5'-RACE was unsuccessful in both mouse and human, hence the precise repertoire of 5'-UTRs and transcriptional start sites in KCC3a remains uncharacterized. Regardless, only the KCC3a-S transcripts that utilize the most 3' alternative acceptor site in exon 1a will definitively generate a different protein from full-length KCC3a.
Intrarenal localization of KCC3 proteins. Immunofluorescence of mouse and rat kidney reveals that KCC3 expression is specific to the proximal tubule, with intense immunoreactivity at the basolateral membrane of S1 to S3 (Fig. 5, A-C). Although not quantitative, expression appears lower in S3 segments, at least in mouse (Fig. 5C). Reactivity is abolished by preabsorption with the immunizing peptide (not shown). KCC3 expression does not colocalize with NKCC2 in thick ascending limbs (not shown) nor with NCC in distal convoluted tubule cells (Fig. 5D). Costaining with aquaporin-1 antibody, which labels apical and basolateral membranes of the proximal tubule in addition to descending thin limb cells (46), reveals an abrupt transition in KCC3 expression between S3 and aquaporin-1-positive thin limbs (Fig. 5, D and E).
Heterologous expression of KCC3 isoforms in X. laevis oocytes. Microinjection of all five of the human KCC3 isoforms cRNAs in X. laevis oocytes resulted in the expression of immunoreactive KCC3 proteins of the appropriate molecular weight (see Fig. 6). High-mannose glycoproteins (25) with apparent masses close to that of the core unglycosylated proteins were thus detected, corresponding to the KCC3a, KCC3a-x2M, KCC3b, and KCC3a-S proteins (predicted core MWs of 127,608, 125,985, 122,083, and 121,117 Da, respectively). Notably, the KCC3a and KCC3a-S proteins differ in evident mass, indicating that these proteins do not share the same start codon (see above). Detection of KCC3b-x2M protein required much higher amounts of protein (4 oocytes/lane, Fig. 6B, vs. 1 oocyte/lane for the other 4 isoforms) and the use of ECL-Plus; again, however, the high-mannose protein was close to the core MW of the predicted KCC3b-x2M protein (120,049 Da).
Expression of all five KCC3 cRNAs resulted in significant exogenous K+-Cl cotransport, compared with control oocytes that were injected with water. Shown in Fig. 7 is a representative experiment in which X. laevis oocytes were injected with water or 5 ng/oocyte of cRNA transcribed from each isoform. 86Rb+ uptake was assessed using a hypotonic uptake solution containing 10 and 50 mM extracellular K+ and Cl, respectively. Microinjection of KCC3a and KCC3b cRNA resulted in robust 86Rb+ uptake, 45.80 ± 1.96 and 41.61 ± 1.58 nmol?oocyte1?h1, respectively, vs. 0.94 ± 0.05 nmol?oocyte1?h1 in water-injected control oocytes. The absence of exon 2 in KCC3a or KCC3b (KCC3a/b-x2M), as well as the truncation of the NH2-terminal domain in KCC3a (KCC3a-S), did not affect absolute uptake in oocytes; 86Rb+ uptake in KCC3a-x2M, KCC3b-x2M, and KCC3a-S oocytes was 45.76 ± 1.69, 19.78 ± 2.66, and 46.14 ± 1.69 nmol?oocyte1?h1, respectively. As shown in Fig. 7, in all cases the 86Rb+ influx was Cl dependent. The increased 86Rb+ uptake in KCC3 isoforms was observed only under hypotonic conditions, such that 86Rb+ uptakes in all isotonic groups were not different from that of water-injected controls (data not shown and Fig. 8). Thus the five alternative isoforms of human KCC3 function exclusively as swelling-activated K+-Cl cotransporters. The consistently lower absolute value of uptakes in oocytes injected with KCC3b-x2M (one-third to a half of that seen with other isoforms) is likely due to the much lower expression level of this particular isoform, as judged by Western blotting of injected oocytes (see Fig. 6B).
Differential activation by cell swelling. K+-Cl cotransport is controlled by phosphorylation and dephosphorylation, such that activation by multiple stimuli is blocked by phosphatase inhibition (32). Because the five KCC3 isoforms differ in the sequence of the NH2-terminal domain, a region containing multiple putative phospho-acceptor sites, we reasoned that the various isoforms could differ in the response to cell swelling. We first analyzed the response to graded decreases in medium osmolality, to determine whether the isoforms differed in the "set-point" of activation by hypotonic conditions. To maintain similar concentrations of extracellular K+ and Cl, at a consistent ionic strength, all solutions contained 10 mM KCl and 40 mM NMDG-Cl, with a baseline osmolality of 100 mosmol/kgH2O; the appropriate amount of sucrose was added to this stock solution to generate media with osmolalities between 120 and 200 mosmol/kgH2O. As shown in Fig. 8, all KCC3 isoforms were inactive when uptakes were performed at extracellular osmolalities of 200 and 180 mosmol/kgH2O. The response to hypotonic conditions was identical between KCC3a, KCC3a-x2M, and KCC3a-S, with 20% activation at 160 mosmol/kgH2O, 75% at 140 mosmol/kgH2O, and full activation at 120 mosmol/kgH2O. KCC3b had a similar response, albeit with a lesser activation at 140 mosmol/kgH2O (60%). In contrast, KCC3b-x2M exhibited a significantly weaker response to hypotonic conditions, as activation only began at 120 mosmol/kgH2O. As shown in Fig. 8B, however, KCC3b-x2M did in fact function as a swelling-activated K+-Cl cotransporter under these conditions, as the uptake was 2.4 ± 0.18 nmol oocyte/h at 100 mosmol/kgH2O compared with 0.11 ± 0.01 at 200 mosmol/kgH2O. No significant differences were observed in the sensitivity to extracellular osmolarity among the five KCC3 isoforms. Because it is well described that the activation of red cell K+-Cl cotransport by cell swelling is not instantaneous, but occurs after a significant lag time (27), we compared the rate of activation by cell swelling; due to the lower activity of KCC3b-x2M, it was excluded from this comparison. Groups of 300 X. laevis oocytes from a single frog were injected with 5 ng/oocyte of cRNA transcribed from each KCC3 isoform, the 86Rb+ uptake was assessed 4 days later at an osmolality of 160 mosmol/kgH2O; K+-Cl cotransport was robust but not maximal at this osmolality in our prior experiments using graded changes in osmolality (Fig. 8). For each group, 150 oocytes were incubated at the same time in the uptake solution containing 2 μCi 86Rb+ and groups of 15 oocytes were retrieved at different time points. The other 150 oocytes were incubated in parallel in an uptake solution without Cl. As shown in Fig. 9, the speed of activation differed among the KCC3 isoforms. KCC3a and KCC3b exhibited identical patterns of activation, such that at 30 min the percentage of activity in KCC3a was 30 ± 2.8% and in KCC3b it was 30 ± 1.6%. Activation of cotransporter activity was significantly higher in all assessed points during the first 60 min in KCC3a-x2M and in KCC3a-S. At 30 min, activity in these two isoforms was 69 ± 3.7 and 55 ± 2.9%, respectively. These values were significantly higher compared with the percentages observed in KCC3a and KCC3b (P < 0.001).
Kinetic properties of human KCC3. The remaining functional properties of KCC3 were only assessed in KCC3a- and KCC3b-injected cells. To determine and compare the kinetic properties of KCC3 in the same expression system utilized for the other three KCCs (40, 55), we measured 86Rb+ uptake in KCC3a- and KCC3b-injected oocytes as a function of the concentration of each transported ion. The results of these series of experiments are depicted in Fig. 10. Uptakes were performed with Rb+ or Cl concentration fixed at 50 mM, with changing concentrations of the counterion from 0 to 50 mM. Values obtained in simultaneous uptakes measured in water-injected oocytes (data not shown) were subtracted from corresponding KCC3a or KCC3b groups to assess only the 86Rb+ uptake due to each KCC3 isoform. In KCC3a, 86Rb+ influx increased as the concentration of each transported ion was raised, until a plateau phase was reached at ion concentrations greater than 2040 mM, compatible with Michaelis-Menten behavior. The calculated apparent Kms for extracellular Rb+ and Cl concentration were 10.7 ± 2.5 and 7.3 ± 1.2 mM, respectively. KCC3b exhibited similar Michaelis-Menten behavior, with apparent Kms for extracellular Rb+ and Cl concentration of 17.2 ± 3.8 and 8.2 ± 0.9 mM, respectively.
Inhibitor profile of KCC3. We previously observed significant differences in affinity for loop diuretics in KCC1, KCC4, and KCC2 cotransporters (40, 55). The dose-dependent effect of the loop diuretic furosemide was thus assessed in KCC3a- and KCC3b-injected cells in the presence of a 10 mM concentration of extracellular K+; prior data also indicated that the efficacy of loop diuretics is dependent on extracellular K+, with maximal effect at this concentration (40). As Fig. 11A illustrates, KCC3a exhibits apparent half-maximal inhibition (K0.5) value of 200 μM for furosemide, which is slightly lower than the respective value for KCC3b, 350 μM. Thus the two major isoforms of KCC3 exhibited similar affinities for furosemide. The observed inhibitory values are similar to those in KCC1 but clearly different from the K0.5 in KCC2- and KCC4-expressing cells, 50 and 900 μM, respectively (40, 55).
The sensitivity of KCC3 to other diuretics or inhibitors of red cell K+-Cl cotransport was also assessed in oocytes injected with KCC3a or KCC3b. Both isoforms were inhibited by bumetanide and [(dihydroindenyl)oxy] alkanoic acid (DIOA) to a similar extent as other KCC cotransporters. When incubated in 2 mM bumetanide, the KCC3a and KCC3b activity was reduced to 14.3 ± 1.4 and 10.2 ± 2.8%, respectively. Similarly, when incubated in 100 μM DIOA, 86Rb+ uptake observed in KCC3a and KCC3b was just 9.9 ± 1.5 and 19.5 ± 1.6% of uptake in control conditions, respectively. In addition, we also tested the effect of 2 mM concentration of the thiazide diuretic trichlormethiazide on the percentage of Cl-dependent 86Rb+ uptake. As was previously observed with KCC1 and KCC4, both KCC3 isoforms are slightly but significantly inhibited by 2-mM concentration of this thiazide-type diuretic. Compared with uptake in control conditions, activity in the presence of this drug was 72.5 ± 5.4 for KCC3a and 75.4 ± 6% for KCC3b.
Influence of other extracellular anions on the transport activity of human KCC3. We previously showed that extracellular anions other than Cl can support cation transport mediated by KCC1 (40), KCC2 (55), and KCC4 (40); similar data have been reported for sheep and human erythrocytes (14). We therefore measured 86Rb+ transport mediated by KCC3a and KCC3b in the absence of Cl and in the presence of different anions; 86Rb+ uptake in oocytes exposed to 40 mM K-gluconate and 10 mM KCl served as the control, compared with uptake medium containing 40 mM K-gluconate and 10 mM of KBr, KH2PO4, KI, K-gluconate, or KSCN. Figure 11B shows the percentage of KCC3a function when uptakes were performed using the different anion substitutions; similar data were obtained with KCC3b (not shown). Both isoforms show slightly higher 86Rb+ influx in the presence of 10 mM KBr, compared with 10 mM KCl. The uptake in KCC3a in the presence of KBr was thus 126 ± 9%, a value that was significantly higher to uptake in the presence of KCl (P < 0.05). 86Rb+ influx was almost completely inhibited in the presence of other anions. Therefore, of the anions tested, only Br and Cl support K+/Rb+ transport by KCC3.
Regulation of KCC3 activity. It is known that inhibition of serine-threonine protein phosphatases prevents the activation of red cell K-Cl cotransport by either cell swelling or NEM (22). In previous studies, we showed that inhibition of protein phosphatase 1 activity with calyculin A prevents the activation of KCC1 (40), KCC2 (55), and KCC4 (40) by hypotonic conditions. In contrast, no effect was observed when protein phosphatases 2A and 2B were inhibited with okadaic acid and cypermethrin, respectively. We therefore assessed the effect of these drugs on KCC3. We used calyculin A at 100 nM (a concentration that inhibits the function of protein phosphatases 1 and 2A, PP1 and PP2A), and okadaic acid at 1 nM (which inhibits only PP2A), and cypermethrin at 100 pM (which selectively inhibits protein phosphatase 2B, PP2B). As shown in Fig. 12, the addition of calyculin A prevents the activation of KCC3a and KCC3b by cell swelling. In contrast, neither okadaic acid nor cypermethrin prevented this activation. These results suggest that inhibition of PP1 is required to prevent the activation of both KCC3 isoforms.
DISCUSSION
At least six different KCC3 transcripts are generated from the SLC12A6 gene via the utilization of alternative transcriptional start sites, alternative first exons, and/or alternative splicing; KCC3a, KCC3b, KCC3a-x2M (minus exon 2), KCC3b-x2M, KCC3a-S1, and KCC3a-S2. As shown in Fig. 2C, these six transcripts encode five different NH2-terminal domains (the KCC3a-S1 and KCC3a-S2 have the same open reading frame, denoted KCC3a-S). The presence of two mutually exclusive first exons generates KCC3a, originally described by Mount et al. (44), and KCC3b, identified by Hiki et al. (23) and Race et al. (53). We determined by 5'-RACE that KCC3b is generated by transcriptional initiation immediately 5' of the start codon in exon 1b, rather than at a promoter flanking the upstream exon 1a. Exon 1a encodes 90 amino acids, with 29 serine/threonine residues and 7 putative PKC sites that are not present in the 39-residue exon 1b. Alternative splicing of exon 2 removes 15 amino acids, including 2 potential phospho-acceptor sites; S96 and T98 (KCC3a sequence), potential substrates for casein kinase-1 and GSK3, respectively. Although exon 2 is alternatively spliced in both KCC3a and KCC3b transcripts, RT-PCR suggests that this splicing event is less common in KCC3b transcripts (see Fig. 3). Finally, a fifth isoform results from the alternative utilization of an internal splicing acceptor site exon 1a, in both the KCC3a-S1 and KCC3a-S2 transcripts (Figs. 1 and 4), resulting in a KCC3a isoform that lacks the first 51 residues.
Notably, there are at least three alternative acceptor sites within and flanking exon 1a (see RESULTS and Fig. 2), with alternative splicing into the most 3' acceptor site in the KCC3a-S1 and -S2 transcripts. KCC3A-S1 and -S2 differ in their respective 5'-UTRs, using two distinct noncoding exons (1c and 1d, respectively); presumably this divergence is generated by alternative transcriptional start sites; however, we have thus far not been successful in mapping start sites 5' of exon 1a. Regardless, the presence of several alternative transcriptional start sites within a single 1-kb CpG island has been previously described (47). Future issues of interest include the effect of promoter methylation on KCC3a transcription and the effect of variation in 5'-UTR on translational efficiency of KCC3a transcripts (45, 62).
Deletion of the first 46 and 89 amino acids of mouse KCC1 has been reported to reduce transport by 50 and 100%, respectively (11); hence it is superficially surprising that all of the NH2-terminal KCC3 isoforms are functional (see below). However, a closer inspection reveals that sequence variation in the various KCC3 isoforms occurs proximal to an NH2-terminal motif (-alfeee-) that is absolutely conserved in all four mammalian KCCs; with the exception of the first six residues of KCC3b (MPHFTV-), which are identical in KCC1, this sequence is the most proximal conserved motif within the NH2 termini of KCC proteins. This conserved sequence is partially disrupted in the 46-residue KCC1 truncation and abolished in the 89-residue truncation induced by Casula et al. (11) in mouse KCC1 and is likely required for some critical aspect of expression and/or transport activity.
None of the KCC3 isoforms mediate K+-Cl cotransport under isotonic conditions but are, however, strongly activated by cell swelling induced by hypotonic conditions. These findings are similar to previous observations made in X. laevis oocytes expressing KCC1 and KCC4 (40, 58, 59). Rat and human KCC2 mediate significant K+-Cl cotransport in X. laevis oocytes under isotonic conditions, with further activation of this baseline activity by cell swelling (55, 58); KCC2 remains the only mammalian paralog with activity under isotonic conditions, due to a KCC2-specific COOH-terminal domain (Mercado A and Mount DB, unpublished observations). We reported that X. laevis oocytes have a low endogenous swelling-activated K+-Cl cotransport activity and express at least one KCC with significant COOH-terminal homology to the mammalian KCCs (39). This endogenous activity is readily distinguishable from that of heterologously expressed KCCs, such that the KCC3 isoforms studied here mediate up to a 40-fold higher K+-Cl cotransport under hypotonic conditions. In contrast, KCC3 expressed in NIH-3T3 cells is reportedly not swelling activated (54). Relatively modest activation of KCC1 (18, 19, 24) and KCC3b (53) by hypotonic conditions has also been reported in HEK293 cells, in which KCC2 is not swelling activated (48). However, given that regulatory volume decrease (RVD) in response to hypotonic cell swelling is reduced in both neurons and proximal tubular cells from KCC3 knockout mice (7), it appears that KCC3 is activated in vivo by hypotonic cell swelling. Presumably the appropriate signaling pathways for the activation of KCC3 and other KCCs by cell swelling are weakly expressed in mesenchymal cell lines such as HEK293 and NIH-3T3.
Northern blot analysis of mouse tissues suggests that KCC3b is a renal-specific transcript; however, this isoform is easily detected in several tissues by RT-PCR in both mouse and human. Moreover, the mouse kidney clearly expresses detectable KCC3a transcript by Northern blot analysis (Fig. 1B). However, the kidney is unique in showing a lower molecular weight KCC3 protein on Western blots (Fig. 3B), suggesting that KCC3b is the dominant isoform in this tissue. This observation cannot be more definitive in the absence of KCC3a- and KCC3b-specific antibodies and/or a direct measurement of protein abundance by proteomic approaches. However, PCR amplification and sequencing of the entire open reading frame of KCC3 from kidney mRNA (Fig. 1A) did not reveal alternative splicing events other than the deletion of exon 2, leaving a relative predominance of shorter KCC3b protein as the most likely explanation for the lower mass of KCC3 protein in kidney. Because KCC3a and KCC3b are generated by transcriptional initiation at two separate promoters, we expect that these isoforms are differentially regulated in the kidney, as is the case in vascular smooth muscle cells (15).
The combined data from this and other reports (7) indicate that KCC3 functions as a swelling-activated K+-Cl cotransporter at the basolateral membrane of the proximal tubule. Thus we have localized KCC3 expression from S1 to S3 of both mouse and rat and have demonstrated by heterologous expression that KCC3 isoforms function as swelling-activated K+-Cl cotransporters; Boettger et al. (7), in turn, reported localization of KCC3 in murine proximal tubule, with a marked reduction in RVD in perfused proximal tubules of KCC3-deficient mice. The prior physiological and functional evidence for the existence of basolateral K+-Cl cotransport in the proximal tubule has been reviewed in detail elsewhere (43). Of particular significance, apical Na+-glucose transport in proximal tubular cells strongly activates a Ba2+-resistant K+ efflux pathway that is 75% inhibited by 1 mM furosemide (2), pharmacology that is consistent with a K+-Cl cotransporter. Therefore, isosmotic cell swelling in the proximal tubule in response to apical Na+ absorption was postulated to activate a volume-sensitive basolateral K+-Cl cotransporter (43). Preliminary data from our own KCC3 knockout mice indicate that this transporter is crucial for transepithelial fluid transport, presumably by functioning as a swelling-activated exit pathway for Cl during transepithelial Na+-Cl transport (61). The role of swelling-activated K+-Cl cotransport in preserving cellular integrity of the proximal tubule is a particularly intriguing issue for future study, given the involvement of KCC3 in ACCPN, a hereditary neurodegenerative syndrome (25).
The "set point" osmolality for the activation of K+-Cl cotransport, found to be 160 mosmol/kgH2O, does not differ for the five KCC3 isoforms (Fig. 8). Notably, isotonic conditions for X. laevis oocytes are considerably lower than for mammalian cells, such that this set point is not directly translatable to human physiology. Completely analogous experiments on the set point for red cell K+-Cl cotransport have not been reported. However, data on the effect of osmolality on the rate of inactivation of swelling-activated K+-Cl cotransport suggest a set point of between 225 and 258 mosmol/kgH2O, i.e., at a similar percentage of isotonic osmolality seen in oocytes for KCC3 (27). This previous study of red cell K+-Cl cotransport had also revealed a lag time of several minutes during the activation by cell swelling (27); hence we compared the time course of swelling activation for the KCC3 isoforms. Data from Figs. 8 and 9 show that hypotonic media activate KCC3a and KCC3b to a similar extent, with an essentially identical time course. In contrast, although maximal K+-Cl cotransport is reproducibly equivalent in oocytes expressing KCC3a, KCC3a-x2M, or KCC3a-S, the time curve of activation for the shorter isoforms is shifted to the left (see Fig. 9). Simplistically, the lesser content of (putative) phospho-acceptor sites in these shorter KCC3a isoforms is related to their greater speed of activation, as dephosphorylation is involved in the activation of KCCs by cell swelling in oocytes and other cells (32, 40, 55). Alternatively, trafficking and/or expression of these isoforms at the cell membrane in response to cell swelling are altered by removal of exon 2 or the first 50 residues of KCC3a. We assume that the alternative splicing of exon 2 has some biological role, given that removal of this cassette exon appears to be particularly prominent in mouse brain (Fig. 3C). The functional expression of KCC3b-x2M was significantly attenuated (Figs. 8 and 9); this is likely due to the drastically reduced expression level of KCC3b-x2M protein in X. laevis oocytes, as assessed by Western blot analysis (see Fig. 6). The reduced expression of this particular isoform is somewhat surprising, as it is essentially identical in length to KCC1, with which it shares 76% identity. There does, however, appear to be selection against the expression of this isoform in vivo, since alternative splicing of exon 2 is considerably less frequent in KCC3b transcripts than in KCC3a (see Fig. 3).
Kinetic analysis reveals that KCC3a and KCC3b exhibit very similar affinities for extracellular Rb+, a surrogate for K+ (apparent Kms of 10.7 ± 2.5 and 17.2 ± 3.8 mM, respectively) as well as for extracellular Cl (apparent Kms of 7.3 ± 1.2 and 8.2 mM ± 0.9, respectively). We did not expect the kinetic characteristics of these two isoforms to differ as they do not differ in the central core of TM domains, which likely confer ion transport characteristics. Our kinetic results for Rb+ are similar to those of Race et al. (53), who reported a Km for K+ of 9.5 ± 1.4 mM, but differ markedly for Cl, for which the Km is 51 ± 9 mM in HEK293 cells transiently transfected with human KCC3b. The reason for this discrepancy in anion affinity is not clear but may result from the different expression systems. First, we assume that the higher KCC3 activity in X. laevis oocytes compared with transiently transfected HEK293 cells (53) leads to greater accuracy in kinetic measurements. Moreover, as Race et al. demonstrated that HEK293 cells have a significant baseline K+-Cl cotransport activity, it is likely that these cells express several other human KCCs. Given the modest expression levels of KCC3 achieved by Race et al. and the potential for cotransporter heteromultimerization (10, 41), it is conceivable that "hybrid" kinetics resulted from heteromultimerization of endogenous HEK293 KCCs with transfected KCC3. Although there is an endogenous K+-Cl cotransporter in X. laevis oocytes (39), such hybrid kinetics are less likely to have a significant quantitative effect, given the 40-fold greater activity of heterologously expressed KCC3 and the reported differences in kinetics for the endogenous X. laevis KCC(s) (the Km values for K+ and Cl in native oocytes are 22.2 ± 6.4 and 15.4 ± 4.7 mM, respectively) (39).
We previously reported apparent cation Km values for KCC1, KCC2, and KCC4 of 25.5, 9.3, and 17.5 mM, respectively (40, 55); the relevant Km values for Cl were 17.2, 16.1, and 6.8 mM. As reviewed previously (40, 48, 55), it is likely that variation in key transmembrane domains, in particular TM2, TM4, and TM7, determines this divergence in ion affinities. Comparing the "anion series" of Rb+ uptake (40, 55), it is apparent that KCC3 is unique in supporting a slightly higher activity in the presence of extracellular Br (Fig. 11B) compared with control oocytes in the presence of extracellular Cl. A similar preference for Br has been reported for K+-Cl cotransport in sheep red blood cells (31, 33).
The KCC3 protein is clearly expressed in red blood cells (35), and the partial concordance between the transport properties of KCC3 and red cell K+-Cl cotransport suggests that KCC3 is the dominant KCC in red blood cells. In the absence of KCC3a- and KCC3b-specific antibodies, it is not as yet clear which isoform is predominant in mature red blood cells; however, both transcripts are detectable in human bone marrow (Fig. 2) and in differentiated human K562 erythroleukemia cells (data not shown). It is less clear how the kinetic and functional attributes of KCC3 reflect the role of this transporter in the nervous system and, in particular, how the inactivation of KCC3 causes ACCPN (25). It is however evident that KCC3 is coexpressed with KCC2 in a subset of large neurons (51), in addition to human NT2-N cells, a cellular model of immature neurons (55). Because KCC3 differs from KCC2 in not mediating K+-Cl cotransport under isotonic conditions, it is unlikely to play a significant role in regulating baseline neuronal Cl activity, a major function of KCC2 (49). Swelling-activated K+-Cl cotransport mediated by KCC3 may, however, play a role in the volume-regulatory response to neuronal excitation (20), such that loss of this function leads to neurodegeneration and the CNS features of ACCPN (7).
Similar to other KCCs, the KCC3a and KCC3b isoforms are sensitive to several inhibitors, including loop diuretics, DIOA and thiazide-type diuretics (see Fig. 11A and RESULTS). DIOA is particularly potent, with almost complete inhibition at a concentration of 100 μM. The weak but significant sensitivity to thiazides is interesting as members of the electroneutral cation Cl-coupled cotransporters have been defined, in part, due to their sensitivity to diuretics, whereas the Na+-K+-2Cl cotransporters are sensitive to loop diuretics and resistant to the thiazides, and the Na+-Cl cotransporter is inhibited by thiazides and resistant to loop diuretics (22). We previously reported similar modest sensitivity to thiazides in both KCC1 and KCC4 (40). Analogous observations have been reported by Harling et al. (21), who showed that a KCC ortholog from the tobacco plant can be inhibited by bumetanide, furosemide, and the thiazide-like diuretic metolazone.
The dose-response curves using furosemide revealed an apparent half-maximal inhibition (K0.5) for both KCC3a and KCC3b of 250 μM. This value is similar to that observed in KCC1 (200 μM) (40) but differs from K0.5 in KCC2 (50 μM) (55) and KCC4 (900 μM) (40). The affinity for loop diuretics is thus similar in the KCC1-KCC3 subgroup and divergent within the KCC2-KCC4 subfamily, in that KCC2 has the highest affinity and KCC4 the lowest affinity. In this regard, the Km values that we observed for extracellular Cl in KCC1 (17.2 mM) (40), KCC2 (6.8 mM) (55), KCC3 (14.3 mM; present study), and KCC4 (16.1 mM) (40) vary by at most 10 mM. However, the four KCC paralogs exhibit differences in the transport capacity in the presence of other anions. Thus, although Cl transport kinetics are similar among K+-Cl cotransporters, the loop diuretic affinities and anion series are quite different, suggesting that the amino acid residues and TM domains defining each of these properties are not identical.
It has been shown that inhibition of protein phosphatases prevents the swelling- and NEM-induced activation of red cell K+-Cl cotransport (5, 30, 32, 56). Heterologous expression studies of mammalian KCCs have revealed a similar dependence on protein dephosphorylation (36, 40, 55, 58, 59). Our data support this hypothesis, since calyculin A, an inhibitor of both PP1 and PP2A (12), completely abrogated the hypotonic activation of KCC3a and KCC3b. To discriminate between serine-threonine phosphatases, we also tested the effect of 1 nM okadaic acid, which