Barttin, a gene product of BSND, is one of four genes responsible for Bartter syndrome. Coexpression of barttin with ClC-K chloride channels dramatically induces the expression of ClC-K current via insertion of ClC-K-barttin complexes into plasma membranes. We previously showed that stably expressed R8L barttin, a disease-causing missense mutant, is retained in the endoplasmic reticulum (ER) of Madin-Darby canine kidney (MDCK) cells, with the barttin β-subunit remaining bound to ClC-K α-subunits (Hayama A, Rai T, Sasaki S, Uchida S. Histochem Cell Biol 119: 485–493, 2003). However, transient expression of R8L barttin in MDCK cells was reported to impair ClC-K channel function without affecting its subcellular localization. To investigate the pathogenesis in vivo, we generated a knockin mouse model of Bartter syndrome that carries the R8L mutation. These mice display disease-like phenotypes (hypokalemia, metabolic alkalosis, and decreased NaCl reabsorption in distal tubules) under a low-salt diet. Immunofluorescence and immunoelectron microscopy revealed that the plasma membrane localization of both R8L barttin and the ClC-K channel was impaired in these mice, and transepithelial chloride transport in the thin ascending limb of Henle's loop (tAL) as well as thiazide-sensitive chloride clearance were significantly reduced. This reduction in transepithelial chloride transport in tAL, which is totally dependent on ClC-K1/barttin, correlated well with the reduction in the amount of R8L barttin localized to plasma membranes. These results suggest that the major cause of Bartter syndrome type IV caused by R8L barttin mutation is its aberrant intracellular localization.
- CLC chloride channel
- Bartter syndrome
patients with bartter syndrome type IV exhibit phenotypes such as fetal polyuria that causes maternal polyhydramnios, postnatal polyuria causing dehydration, hypokalemic metabolic alkalosis, and sensorineural deafness. The gene responsible for this disease is BSND, and barttin encoded by BSND is an accessory subunit of ClC-K chloride channels (2, 8). ClC-K1 and ClC-K2 are highly homologous members of the CLC chloride channel family. ClC-K1 is present in both the apical and basolateral plasma membranes of the thin ascending limb of Henle's loop (tAL) and is responsible for the high chloride permeability of tAL (26). Knockout of ClC-K1 results in nephrogenic diabetes insipidus in mice (15). On the other hand, ClC-K2 is present in the basal plasma membranes of the thick ascending limb of Henle's loop (TAL), distal convoluted tubules (DCT), connecting tubules (CNT), and intercalated cells of collecting ducts (13, 32). ClC-Kb, a human homolog of ClC-K2, has been identified as one of the genes responsible for Bartter syndrome (21). Barttin was initially reported to recruit these ClC-K channels to plasma membranes (8, 10, 29). Additionally, Fahlke's group (11, 20) reported that barttin regulates permeation and gating of ClC-K channels. We have previously shown that a disease-causing missense mutant (R8L) of barttin is localized to the endoplasmic reticulum (ER) and cannot recruit ClC-K to plasma membranes in mammalian cell lines (10). Hence, it is unclear whether the pathogenesis of Bartter syndrome type IV in vivo is the result of mislocalization of ClC-K channels or a functional disturbance of ClC-K channels, or both. To clarify this issue, we generated the BsndR8L/R8L knockin mice (R8L knockin mice) by gene targeting.
MATERIALS AND METHODS
Generation of R8L knockin mice.
To generate R8L knockin mice, the targeting vector was prepared by using PCR-amplified segments of Bsnd, in which entire sequences were verified by sequencing. The point mutation (R8L) was introduced into exon 1 of the targeting vector by site-directed mutagenesis (Stratagene). The targeting vector was then transfected into J1 embryonic stem (ES) cells by electroporation (22). After selection with 150 μg/ml G418 and 2 μM ganciclovir, targeted ES clones were chosen from PCR (primers: F1 5′-TCC CTT ATA GAG GAG CAC AGT AAG CCA GCT-3′, R1 5′-TCC TGA CTA GGG GAG GGG GAG GAG TAG AAG TG-3′), Southern blotting, and sequencing of the mutation site. ES clones were injected into C57Bl/6 blastocysts. Chimeric males were bred with C57Bl/6 females to produce BsndNeo(R8L)/+ mice, and the neo cassette was deleted by crossing the mice with Cre recombinase-expressing transgenic mice (18). Offspring were genotyped by PCR with sense primer f1 (5′-CTG GGG AAC ATC CTT TTG CT-3′) and f2 (5′-AAG AGC TTG GCG GCG AAT GG-3′) and antisense primer r1 (antisense: 5′-GTG TCT CTA ACT AGC TGT GT-3′) and r2 (5′-CTG AAA GGA CAG GGC TGT TA-3′). f1/r1 amplified the wild-type allele (541 bp), f1/r2 amplified the R8L knockin allele (252 bp), and f2/r2 amplified the Neo allele (600 bp). The Animal Care and Use Committee of Tokyo Medical and Dental University approved the experiment.
Total RNA from mouse kidneys was reverse-transcribed by using Omniscript reverse transcriptase (Qiagen). PCR was performed with primers for barttin [sense; 5′-GGG TAC ATT CCT TAT CAG CC-3′ (exon 1) and antisense; 5′-TTG GAA GTC AGA GTC TGC TG-3′ (exon 3)]. Quantitative RT-PCR was performed with FastStart DNA Master PLUS SYBR Green I kit and LightCycler (Roche) software (Ver3.5). The relative abundance of each transcript was normalized to GAPDH (primer was obtained from Roche).
Rabbits were immunized to generate an anti-barttin antibody that recognized the two regions (PLPDKELGFEPDIQG+PEQEEEDLYYGLPD) of mouse barttin [immunoblotting (IB) 1:500 and immunofluorescence (IF) 1:200]. Primary antibodies also used in this study were rabbit anti-ClC-K (IB and IF 1:100) (27); rabbit anti-rat Na-K-2Cl cotransporter (NKCC2; IB 1:500) (Alpha Diagnostic); guinea pig anti-rat NKCC2 (IF 1:100) (19); guinea pig anti-Na-Cl cotransporter (NCC; IB 1:500, IF 1:200); goat anti-aquaporin-2 (AQP2) C-17 (IF 1:50, Santa Cruz Biotechnology); rabbit anti-α-endothelial Na channel (ENaC; IB 1:100, Chemicon); rabbit anti-β-ENaC (IB 1:100, Alomone); rabbit anti-γ-ENaC [IB 1:500, kindly provided by Dr. M. Knepper, National Institutes of Health (NIH)]; rabbit anti-pendrin (IB 1:250) (23); and rabbit anti-β-actin (IB 1:500, Cytoskeleton). Alkaline phosphatase-conjugated anti-IgG antibodies (Promega) were used as secondary antibodies for immunoblotting, and Alexa 488- or 546-conjugated secondary antibodies (Invitrogen) were used for immunofluorescence.
Semiquantitative immunoblotting was performed as described previously (31). Crude membrane fractions (17,000 g) were used to assess relative expression levels of the proteins. The intensity of bands was analyzed by using Image J (NIH).
Kidney histology and immunofluorescence.
For light microscopy, mouse kidneys were fixed by immersion in formalin, embedded in paraffin, and stained with hematoxylin and eosin. For immunofluorescence, mouse kidneys were fixed with periodate lysine (0.2 M) and paraformaldehyde (2%) in PBS by perfusion through the left ventricle. Tissue samples were soaked for several hours in 20% sucrose in PBS, embedded in Tissue-Tek OCT Compound (Sakura Finetechnical), and snap frozen in liquid nitrogen. Immunofluorescence images were observed by LSM510 Meta (Carl Zeiss). For electron microscopy, mouse kidneys perfused with 4% paraformaldehyde in PBS were cut into pieces of superficial cortex, outer medulla, and inner medulla, postfixed in 2% glutaraldehyde in 0.1 M cacodylate buffer at pH 7.4, dehydrated with ethanol, and embedded in LR White resin (Okenshoji, Tokyo, Japan). Immunoelectron microscopy was performed as described previously (16). Briefly, we obtained photographs of medullary TAL (mTAL) and tAL in the sections of outer medulla and inner medulla, respectively, using an Hitachi electron microscope H7000 or H7100 (Hitachi High-Technologies). Photographs of cortical TAL (cTAL) and DCT were obtained in the sections of superficial cortex. To quantify the number of gold particles in mTAL or tAL under a comparable tissue sampling, three to five transverse sections of mTAL or tAL, of which their shapes were close to right circles and of which the diameter was about the same, were selected in the sections from both wild-type and the knockin mice, and photographs to cover the whole cells in the tubules were taken at the magnification of ×8,000–10,000 and ×20,000–30,000 for mTAL and tAL, respectively. The plasma membrane localization of barttin was assessed by the ratio of the number of gold particles on the plasma membranes to the total number of gold particles within each tubule. A person who did not know the genotypes took photographs and counted the gold particles. For all immmunolabeling studies, mice had been fed a normal diet at least for 1 wk before death.
Blood and urine analysis and blood pressure measurement.
Before each measurement, the mice were adapted for 10–14 days to a normal (0.4% NaCl, 0.8% K) or a low-salt diet (0.01% NaCl). All foods were obtained from Oriental Yeast. Blood was drawn from the retroorbital sinus or venous plexus near the mandible without anesthesia. Blood data were determined by i-STAT (Fuso). Serum aldosterone levels were measured by the SRL clinical laboratory service. Mice were kept in metabolic cages for urine collection. Urine samples were analyzed by DRI-CHEM (Fujifilm). Urine osmolality was measured by using a Fiske Osmometer (model 110, Fiske Associates). Blood pressure of restrained conscious mice was measured by a programmable tail-cuff sphygmomanometer (MK-2000A) after a 2-day period of acclimation to the instrument. Blood pressure data were calculated as the average of five sequential measurements.
In vitro microperfusion of tAL and cTAL.
tAL and cTAL were microdissected and microperfused in vitro on an inverted microscope as previously described (15). Each distal end of the tubule was sucked into a glass micropipette, and the lumens were cannulated. To estimate chloride permeability, the 5-nitro-2-(3-phenylpropylamino)-benzoate-sensitive chloride (NPPB) diffusion potential (Vd) or transepithelial voltage (Vt) was measured (15).
Thiazide and furosemide infusion test.
We prepared a 5 mg/ml furosemide solution and a 25 mg/ml hydrochlorothiazide (HCT) solution in 50% DMSO for intraperitoneal injection. Before injection of the drugs, mice were injected intraperitoneally with 70 μl/g body wt saline to facilitate spontaneous voiding. One hour after saline injection, HCT (25 mg/kg body wt) or furosemide (5 mg/kg body wt) was intraperitoneally injected. The dose of diuretics was determined based on previous studies (5, 31). A solution of 50% DMSO was injected into control mice. All injection volumes of furosemide, HCT, and vehicle were the same (1 μl/g body wt). Urine was collected every 30 min by spontaneous voiding or bladder massage, and chloride excretion and urine volume were measured. Total chloride excretion during the observation period was measured to estimate the effect of diuretics. The diuretic infusion test was performed according to the previously described thiazide test method for humans (6).
All values are expressed as means ± SE. Statistical analyses were performed by using the unpaired t-test. P values <0.05 were considered statistically significant.
Generation of R8L knockin mice.
R8L knockin mice were generated by using homologous recombination in ES cells to create a mutant allele. Exon 1 of the Bsnd gene was modified to contain the R8L mutation, and the following intron was replaced with a cassette encoding the selection marker neomycin transferase flanked by lox P sites (Fig. 1A). Correct gene targeting was confirmed by Southern blotting (Fig. 1B). Recombinant ES cell clones were injected into blastocysts derived from C57BL/6J mice, which were implanted into pseudopregnant females. Resulting chimeric mice were crossed with C57BL/6 mice to produce BsndNeo(R8L)/+ progeny. The neo cassette was deleted by crossing the BsndNeo(R8L)/+ mice with Cre recombinase-expressing transgenic mice (Fig. 1A). Each genotype was verified by PCR. To confirm the introduced mutation, we sequenced the RT-PCR product of exon 1 (Fig. 1C).
Phenotypes of BsndNeo(R8L)/Neo(R8L) and BsndR8L/R8L knockin mice.
We first checked the expression of barttin mRNA and protein. In mice containing the neomycin selection cassette [BsndNeo(R8L)/Neo(R8L)], both mRNA and protein levels of barttin were significantly decreased (Fig. 2, A and B), suggesting that insertion of the neo cassette into intron 1 decreased transcription of the Bsnd gene. Immunoblotting of barttin (Fig. 2B) revealed that these mice had a significantly reduced level of the barttin protein and hence represent a severe hypomorph. In the BsndR8L/R8L knockin mice (R8L knockin mice), expression of barttin mRNA was almost identical to that of Bsnd+/+ mice (wild-type mice) (Fig. 2A). However, protein levels of mutant barttin were slightly decreased (Fig. 2B). The BsndNeo(R8L)/Neo(R8L) mice (hypomorphic mice) were smaller than wild-type mice when they were young; however, the difference in body weight decreased as the mice grew older, and the hypomorphic mice survived to adulthood. Kidneys of the hypomorphic mice showed dilation of the renal pelvis as early as 10 days after birth (Fig. 3A). Blood and urine phenotypes of the hypomorphic mice described here were obtained in mice already showing hydronephrosis. These mice developed severe salt- and water-losing phenotypes as summarized in Table 1. Their urine volume was about three times that of wild-type mice, and their urine osmolality was one-third that of wild-type mice. Hypomorphic mice also showed hypokalemia, metabolic alkalosis, and hyperaldosteronism (Table 1 and Fig. 3B). As ClC-K1 knockout mice only show a decreased ability for urine concentration but no Bartter-like phenotypes (15), these results suggested that ClC-K2 function was additionally impaired in our hypomorphic mice. In fact, the expression of ClC-K (including both K1 and K2) in immunoblots was significantly decreased in 9-day-old hypomorphic pups, in which the inner medulla was preserved (data not shown). These data confirm that both ClC-K1 and ClC-K2 were unstable without barttin coexpression, as previously observed in the inner ear (17). Compared with these hypomorphic mice, R8L knockin mice displayed milder phenotypes. When kept under a normal diet, R8L knockin mice did not show any apparent phenotype in gross physical appearance, kidney morphology (Fig. 3A), or urine or blood laboratory data (Table 1). However, when they were fed a low-salt diet for 10–14 days, they showed mild polyuria, hyponatremia, hypokalemia, and metabolic alkalosis. The blood pressure of R8L knockin mice was similar to that of wild-type mice (107 ± 3 vs. 101 ± 6 mmHg, knockin: n = 11, wild-type: n = 9, P = 0.20). However, when mice were fed a low-salt diet, the blood pressure of R8L knockin mice was lower than that of wild-type mice (98 ± 2 vs. 106 ± 4 mmHg, knockin: n = 11, wild-type: n = 9, P = 0.035). The urine-concentrating ability of R8L knockin mice was normal [1,716 ± 150 to 3,421 ± 86 mosmol/kgH2O, before and after 24-h water deprivation, compared with wild-type mice (1,428 ± 239 to 3,120 ± 256 mosmol/kgH2O), n = 6, P = 0.13] in contrast to the decrease in maximum urine osmolality of the hypomorphic mice (864 ± 102 to 1,738 ± 110 mosmol/kgH2O, n = 6, P = 0.024 vs. wild-type mice).
Expression of renal sodium- and chloride-transporting proteins.
ClC-K chloride channel α-subunits are unstable without barttin (10, 17). We found that the ClC-K protein levels were slightly decreased in our R8L knockin mice, mirroring the decreased level of barttin protein (Fig. 4). In other mouse models of Bartter syndrome, the protein abundance of NCC and ENaC is reportedly increased to compensate for decreased sodium reabsorption in the TAL (28). Although such adaptive responses were not evident when R8L knockin mice were fed a normal diet, the protein levels of NCC and ENaC did increase when the mice were fed a low-salt diet. In addition, we observed increased proteolytic cleavage of α-ENaC (30 kDa) in R8L knockin mice, strongly suggesting that they had a NaCl-losing phenotype (Fig. 4). Pendrin is an apical chloride-HCO3− exchanger in intercalated cells of CCD. It plays roles in thiazide-sensitive chloride reabsorption and HCO3− secretion in CCD (3). In R8L knockin mice, pendrin expression was significantly increased (Fig. 4). This increase could be a compensatory mechanism for the defective basolateral chloride exit through ClC-K2/barttin within the intercalated cells or for the salt-losing phenotype of the knockin mice. The metabolic alkalosis may also be involved in the increased expression of pendrin.
Cellular localization of barttin and ClC-K.
Previous reports (10, 11, 20) demonstrate decreased plasma membrane localization of ClC-K/barttin channels when ClC-K was coexpressed with barttin subunits containing mutations found in human disease. We therefore investigated the cellular localization of ClC-K/barttin in our R8L knockin mice fed a normal diet. By immunofluorescence, NKCC2 and NCC appear to be expressed in the apical membranes of mTAL and DCT, respectively (Fig. 5, A and B). In these nephron segments, basolateral membranes are deeply invaginated along rich mitochondria. Accordingly, the immunofluorescence signals from barttin appear to have a cytosolic localization, even in wild-type mice (8). However, a closer analysis identifies clear barttin staining along the invaginated basal plasma membranes in wild-type mice. On the other hand, mutant barttin appears to be localized more on the apical side (indicated by arrowheads in Fig. 5, A and B). The staining near the basal lamina (indicated by arrows) was significantly decreased (Fig. 5, A and B). This difference was confirmed by immunoelectron microscopy. Although the gold particles were exclusively on the basolateral plasma membranes in wild-type mice (Fig. 5C), the number of gold particles on the plasma membranes near the basal lamina of mTAL and in the subapical cytoplasm was significantly reduced and increased, respectively, in R8L knockin mice (Fig. 5D). In the upper portion (near apical membranes) of invaginated basal plasma membranes, the plasma membrane localization of R8L was preserved. Approximately 50% of R8L barttin was not on the basolateral plasma membranes but present intracellularly (Fig. 5E).
The reduced plasma membrane localization of R8L was much more evident in CNT than in TAL (Fig. 6A). In this study, we designated AQP2- and barttin-positive cells in cortical labyrinth segments as CNT cells. Accordingly, all branched or labyrinth segments of collecting duct (initial collecting tubule) might be included in this CNT. In the CNT cells, AQP2 antibodies stained the apical membrane and the cytosol, while wild-type barttin was clearly detected in basolateral membranes with little intracellular staining (arrowheads in Fig. 6A, top). However, no such clear membrane staining but rather cytoplasmic staining of R8L barttin was observed in all CNT cells (arrowheads in Fig. 6A, bottom).
Barttin was previously shown to be present in intercalated cells (8). In contrast to the clear membrane staining of wild-type barttin in the intercalated cells in CNT (arrows in Fig. 6A, top), R8L barttin staining was mainly intracellular (arrows in Fig. 6A, bottom). Since expression profiles and physiological roles of ClC-K/barttin in different subtypes of intercalated cells in CNT and cortical collecting duct (CCD) have not yet been characterized, there is a limitation in this study to evaluate the difference in intracellular localization of wild-type and the mutant barttin in intercalated cells in any cortical segment. In outer medullary collecting ducts (OMCD), however, the intercalated cells consist of a single cell type, i.e., α-intercalated cells (24), where R8L barttin showed intracellular staining in contrast to the clear membrane staining of wild-type barttin (Fig. 6B).
The localization of ClC-K was then investigated. The ClC-K antibody that we used recognized both ClC-K1 and ClC-K2 proteins (27). Unfortunately, the sensitivity of our antibody was not high enough to clearly detect ClC-K in immunostaining. However, as is the case with barttin, a noticeable difference in the intracellular localization of ClC-K between wild-type and R8L knockin mice was observed in intercalated cells in the cortex (Fig. 7).
Barttin is coexpressed with ClC-K1 in tAL (8). Since this nephron segment is very thin, we were unable to evaluate the intracellular localization of mutant barttin by immunofluorescence. However, immunoelectron microscopy clearly showed that approximately half of the mutant barttin signals were present in the cytoplasm (Fig. 8, A and B). In contrast, the signals were exclusively on the apical and basolateral plasma membranes in the tAL of wild-type mice.
In vitro microperfusion of tAL and cortical cTAL.
Next, we performed functional analyses of chloride transport in tAL and cTAL by using an in vitro microperfusion technique. In tAL, Vd of chloride ions has been used to measure the activity of ClC-K1 (15). When the bath chloride concentration decreased from 145 to 35 mM by substitution of chloride with gluconate, both the steady-state Vd and NPPB-sensitive component of Vd in the R8L knockin mice was significantly lower than in wild-type mice (Fig. 9A). The function of cTAL was also evaluated by measuring Vt with or without NPPB treatment. As shown in Fig. 9B, Vt values of cTAL in the R8L knockin mice under both conditions were not statistically different from those in wild-type mice.
In vivo clearance tests using furosemide and thiazide.
In the in vitro microperfusion study, tAL from R8L knockin mice was functionally impaired, but cTAL was not impaired. ClC-K1 knockout mice, which totally lack chloride permeability in tAL but have normal chloride permeability in TAL, do not show Bartter syndrome-like phenotypes (15). Therefore, we hypothesized that the function of other nephron segments, such as DCT, might be impaired in R8L knockin mice. In vitro microperfusion of mouse DCT is technically very difficult; therefore, we performed clearance tests to estimate the NaCl-reabsorbing ability of TAL and DCT by measuring the responsiveness to furosemide and thiazide. These tests are used clinically to differentiate between Bartter and Gitelman syndromes (6). When furosemide (5 mg/kg), a potent inhibitor of NKCC2 in TAL, was intraperitoneally injected into mice, chloride excretion and urine volume were increased in both wild-type and R8L knockin mice, with no differences between the two types of mice (Fig. 10A). On the other hand, the response to HCT (25 mg/kg) was significantly blunted in the R8L knockin mice (Fig. 10B). This result suggests that NaCl reabsorption in TAL is not significantly affected but that NaCl reabsorption in DCT might be responsible for the NaCl-losing phenotype seen in R8L knockin mice.
In this study, we have generated and analyzed barttin knockin mice that carry a mutation identified in human Bartter syndrome type IV (2). The hypomorphic mice carrying the neo cassette showed severe phenotypes (salt and water loss, hypokalemia, metabolic alkalosis, hyperaldosteronism, and growth retardation). Protein and mRNA levels of R8L barttin in the hypomorphic mice were significantly decreased, because the neo cassette in intron 1 of Bsnd may have prevented its proper transcription. Barttin knockout mice die a few days after birth because of severe dehydration (17). Since the hypomorphic mice survived to adulthood, the residual amount of R8L barttin in these mice may have contributed to their survival. Therefore, these mice can be used as adult barttin knockout mice.
We generated the R8L knockin mice to clarify the pathogenic effect of the R8L mutation in vivo. The protein levels of R8L barttin were not as decreased in the knockin mice as they were in the hypomorphic mice. Accordingly, it was possible for us to evaluate the functional consequence of this mutation in the kidney in vivo. In our previous report, R8L barttin stably expressed in Madin-Darby canine kidney (MDCK) cells was retained in the ER and could not reach the plasma membranes. Since the binding ability of R8L barttin to ClC-K was preserved, ClC-K was also found to be retained in the ER (10). These data are consistent with the lack of surface expression of ClC-K in Xenopus laevis oocytes coexpressed with R8L barttin (8). On the other hand, Fahlke's group showed that when R8L barttin and ClC-K channels are transiently overexpressed in MDCK cells, they could localize to plasma membranes (11). The authors concluded that the major pathogenesis of the R8L mutation was not the sorting defect but the inability of R8L barttin to functionally activate ClC-K. In R8L knockin mice, we clearly observed an aberrant localization of R8L barttin by immunofluorescence and immunoelectron microscopy. Although wild-type barttin was exclusively present on plasma membranes, R8L barttin was observed in the cytoplasm, probably on intracellular organelles, as well as on plasma membranes. By immunoelectron microscopy, about half of R8L barttin was observed on intracellular membranes in tAL (Fig. 8A). The chloride Vd of tAL measured in R8L knockin mice was also decreased (about half of wild-type mice) (Fig. 9A). Thus the reduction in the amount of R8L barttin localized to membranes correlated well with the functional decrease in chloride transport in tAL, clearly suggesting that the functional defect in chloride transport might come mainly from aberrant intracellular localization. If R8L was a complete loss-of-function mutant unable to functionally activate ClC-K channels, further reduction of chloride transport in tAL would be shown, regardless of whether the plasma membrane localization of R8L was preserved.
Microperfusion studies and the chloride clearance test with furosemide did not reveal any decrease in chloride transport in TAL in R8L knockin mice, although basolateral R8L barttin was clearly decreased. In contrast to simple diffusional transepithelial chloride transport in tAL that is fully dependent on ClC-K1 on both sides of the plasma membrane, the transepithelial chloride transport system in TAL is composed of NKCC2 and ROMK on apical plasma membranes and ClC-K2 and Na-K-ATPase on basolateral plasma membranes. These results suggest that decreased chloride conductance on the basolateral plasma membranes of TAL in R8L knockin mice might still be high enough for chloride transport determined by NKCC2, ROMK, and Na-K-ATPase. A similar phenomenon was also observed in some patients with CLCNKB (human homolog of ClC-K2) mutations; these patients did not show typical Bartter-like symptoms but rather showed Gitelman-like symptoms, such as hypomagnesemia and hypocalciuria (12). Another possible explanation for the lack of functional defects in TAL could be compensation provided by other chloride channels in TAL, such as the CFTR-like channel (9, 25). Since the channel is not CFTR itself, but can be identified by a patch-clamp study, we must perform a patch-clamp study of TAL in future investigations to clarify whether the CFTR-like channel is increased to compensate for salt loss.
In the case of DCT, however, the basolateral chloride conductance via ClC-K2 might decrease below the level determined by NCC activity in R8L knockin mice, so that the basolateral chloride conductance might become a limiting factor for transepithelial chloride transport. This interpretation may be consistent with the fact that the NCC protein level in R8L knockin mice was increased compared with that of wild-type mice under a low-salt diet (Fig. 4), although R8L knockin mice showed less responsiveness to thiazide. This NCC increase may be a compensatory effect for the salt-losing phenotype in R8L knockin mice, or it may be induced by an unknown factor when the basolateral chloride exit is impaired by R8L mutation. It will be interesting to determine whether the intracellular chloride ion could be such an unknown factor that mediates cross talk between apical and basolateral transporters. Similarly, the level of pendrin, an apical chloride/HCO3− exchanger in the intercalated cells of CCD, was increased in R8L knockin mice. Since pendrin was recently found to constitute a thiazide-sensitive chloride reabsorption pathway in the kidney (14), the reduced responsiveness to thiazide in R8L knockin mice could be attributed to decreased chloride reabsorption via pendrin. As in the case of NCC, the increased pendrin expression may be another compensatory effect for salt loss, or it may be explained by metabolic alkalosis in R8L knockin mice since pendrin expression is stimulated by alkalosis (3).
In heterologous expression systems (8, 11, 29), R8L barttin was identified as a loss-of-function mutant in terms of eliciting ClC-K current. Nonetheless, the phenotypes of R8L knockin mice were milder than expected. The renal function impairment of patients carrying the R8L mutation was also reported to be less severe than the impairment caused by other mutations, such as the loss of start codon (1, 4, 11). These results indicate that the functional characterization of barttin mutants in heterologous overexpression systems may not necessarily predict the severity of disease in vivo. In this respect, in vivo analyses that use knockin mice are especially important. One reason for the discrepancy between in vitro and in vivo results could be that the level of perturbation in the plasma membrane localization of R8L barttin might be dependent on the types of cells in which the mutant barttin is expressed.
There have been several reports that disease-causing mutant membrane proteins such as ΔF508-CFTR and AQP2-T126M were retained in the ER and that this retention was partially restored by drugs such as heat shock protein 90 inhibitor and curcumin (7, 30). We are currently screening a chemical library by using R8L-expressing MDCK cells to find candidate drugs to restore plasma membrane localization of R8L barttin. The R8L knockin mouse generated in the present study will be a valuable tool for testing the efficacy of candidate drugs in vivo after screening.
This study was supported in part by Grants-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (17081009 to S. Uchida), Grants-in-Aid for Scientific Research (20249047 to S. Uchida) from the Japan Society for the Promotion of Science, the Salt Science Research Foundation (1026), and the Takeda Science Foundation.
No conflicts of interest, financial or otherwise, are declared by the authors.
- Copyright © 2011 the American Physiological Society