The Na⁺-dependent chloride-bicarbonate exchanger SLC4A8 mediates an electroneutral Na⁺ reabsorption process in the renal cortical collecting ducts of mice

Abstract

Regulation of sodium balance is a critical factor in the maintenance of euvolemia, and dysregulation of renal sodium excretion results in disorders of altered intravascular volume, such as hypertension. The amiloride-sensitive epithelial sodium channel (ENaC) is thought to be the only mechanism for sodium transport in the cortical collecting duct (CCD) of the kidney. However, it has been found that much of the sodium absorption in the CCD is actually amiloride insensitive and sensitive to thiazide diuretics, which also block the Na-Cl cotransporter (NCC) located in the distal convoluted tubule. In this study, we have demonstrated the presence of electroneutral, amiloride-resistant, thiazide-sensitive, transepithelial NaCl absorption in mouse CCDs, which persists even with genetic disruption of ENaC. Furthermore, hydrochlorothiazide (HCTZ) increased excretion of Na⁺ and Cl^– in mice devoid of the thiazide target NCC, suggesting that an additional mechanism might account for this effect. Studies on isolated CCDs suggested that the parallel action of the Na⁺-driven Cl^–/HCO₃^– exchanger (NDCBE/SLC4A8) and the Na⁺-independent Cl^–/HCO₃^– exchanger (pendrin/SLC26A4) accounted for the electroneutral thiazide-sensitive sodium transport. Furthermore, genetic ablation of SLC4A8 abolished thiazide-sensitive NaCl transport in the CCD. These studies establish what we believe to be a novel role for NDCBE in mediating substantial Na⁺ reabsorption in the CCD and suggest a role for this transporter in the regulation of fluid homeostasis in mice.

Introduction

Sodium chloride is the main extracellular osmotic constituent and thereby determines extracellular volume and blood pressure. To maintain a constant extracellular volume, the kidney has to match sodium excretion to dietary sodium intake. Abnormal retention of sodium by the kidney can ultimately lead to expansion of the extracellular volume and hypertension (1), the most common pathological state in humans. Since sodium is freely filtered by the glomerulus, most of it has to be reabsorbed as the filtrate flows along the nephron. This reabsorption is mediated by the Na⁺/H⁺ exchanger NHE3 in the proximal tubule (2), by the Na⁺/K⁺/2Cl^– cotransporter NKCC2 in the thick ascending limb of Henle’s loop (3), and by the NaCl cotransporter NCC in the distal convoluted tubule (DCT) (3, 4). Finally, the remaining fraction of filtered sodium enters the connecting tubule and the collecting duct. In these latter segments, aldosterone increases distal sodium reabsorption via the Na⁺ channel ENaC (5). Supporting the importance of renal Na⁺ handling in blood pressure regulation, inactivating mutations in the genes that code for renal sodium transporters are associated with low blood pressure (6–10), whereas inherited and acquired forms of hypertension can result from increased renal sodium reabsorption (11).

Drugs that selectively block the different aforementioned renal sodium transporters are the pharmacological basis of treatment of disease states characterized by abnormal renal sodium retention, such as edematous disorders and hypertension. Although discovered half a century ago (12), thiazides have been the cornerstone of therapy for mild and moderate hypertension in nearly all prospective therapeutic trials to date (13). Their efficacy in preventing hypertensive cardiovascular complications such as stroke and congestive heart failure has been verified in large clinical trials (14). Thiazides are believed to act exclusively by blocking sodium absorption via NCC, which represents only approximately 5% of the total amount of Na⁺ filtered by the glomerulus (15, 16). However, previous studies have shown that approximately 50% of Na⁺ absorption in the rat collecting duct is thiazide sensitive and amiloride insensitive (17–19), even though the expression of its canonical target, NCC, is restricted to the DCT.

Given its clinical relevance, we aimed to identify the transport system that accounts for this amiloride-insensitive, thiazide-sensitive Na⁺ absorption in the cortical collecting duct (CCD). With a combined functional and genetic approach, we show that the parallel action of the Na⁺-independent anion exchanger pendrin/Pds/SLC26A4 and the Na⁺-dependent anion exchanger NDCBE/SLC4A8 mediates thiazide-sensitive electroneutral NaCl reabsorption in the CCD. This finding may have important implications for the treatment of arterial hypertension and our understanding of the role of the CCD in the regulation of Na⁺ and K⁺ homeostasis.

Results

Electrogenic and electroneutral Na⁺ absorption pathways coexist in the mouse collecting duct.

To verify the presence of the previously reported thiazide-sensitive component of Na⁺ absorption, we simultaneously measured transepithelial Na⁺ (J_Na), K⁺ (J_K), and Cl^– (J_Cl) fluxes and transepithelial voltage (V_te) in isolated mouse CCDs microperfused in vitro. Because mouse CCDs do not absorb NaCl under basal conditions (see ref. 20 and control group in Figure 1A), we stimulated NaCl absorption by feeding the mice a Na⁺-depleted diet for 2 weeks before the experiments. CCDs from NaCl-restricted wild-type mice absorbed Na⁺ and Cl^–, secreted K⁺, and generated a lumen-negative transepithelial voltage (V_te), consistent with ENaC-mediated Na⁺ absorption (Figure 1A). Amiloride in the perfusate, at concentrations that fully inhibit ENaC (10^–5 M) (21), did not change Cl^– absorption (Figure 1A), although both V_te and K⁺ secretion were eliminated and Na⁺ absorption was reduced by 60% (Figure 1A). In contrast, luminal addition of 10^–4 M hydrochlorothiazide (HCTZ) abolished both J_Cl and the amiloride-insensitive component of J_Na, whereas J_K and V_te were not affected by HCTZ (Figure 1A). We next tested the effects of luminal addition of 10^–4 M HCTZ on J_Na, J_Cl, J_K, and V_te in the absence of amiloride. Figure 1B shows that HCTZ decreased Na⁺ absorption by approximately 45% and almost abolished Cl^– absorption. However, HCTZ did not affect either V_te or K⁺ secretion (Figure 1B). Taken together, these results strongly support the hypothesis that HCTZ inhibits a system that is different from ENaC and that mediates electroneutral NaCl absorption in the mouse collecting duct.

Figure 1. Pharmacological characterization of transepithelial transport of Na⁺, K⁺, and Cl^– in collecting ducts isolated from wild-type mice.

(A) Effects of amiloride (10^–5 M) and HCTZ (10^–4 M) on Na⁺, Cl^–, and K⁺ transepithelial fluxes and on transepithelial voltage in CCDs isolated from Na⁺-restricted mice. J_Na, rate of Na⁺ absorption; J_Cl, rate of Cl^– absorption; V_te, transepithelial voltage; J_K, rate of K⁺ secretion. Statistical significance was assessed by ANOVA; comparisons between groups were tested by Bonferroni’s post-hoc test. n = 5 in each group, *P < 0.05 versus control, ^#P < 0.05 versus low Na⁺; ^†P < 0.05 versus low Na⁺ amiloride (amil). (B) Effects of HCTZ alone (10^–4 M) on Na⁺, Cl^–, and K⁺ transepithelial fluxes and on transepithelial voltage in CCDs isolated from Na⁺-restricted mice. Statistical significance was assessed by 2-tailed unpaired Student’s t test. n = 5 in each group; *P < 0.05 versus control (low Na).

To further exclude a role for ENaC in the amiloride-insensitive, HCTZ-sensitive component of sodium reabsorption, we similarly studied isolated perfused CCDs from mice with a collecting duct–specific disruption of α-ENaC (22) in vitro. It has been shown previously in this genetic model that disruption of the Scnn1a locus in the CCD abolishes α-ENaC protein expression and prevents the apical membrane expression of β and γ subunits in CCD cells (22), resulting in the complete ablation of ENaC channel activity in the collecting duct (22). As shown in Figure 2, CCDs from control mice on a normal Na⁺ diet again had no significant Na⁺, Cl^–, or K⁺ transport. Importantly, as also shown in Figure 2, CCDs from collecting duct–specific ENaC–KO mice on a Na⁺-depleted diet absorbed Na⁺ and Cl^– but did not generate a lumen-negative transepithelial voltage and did not secrete K⁺. Our results confirm that the thiazide-sensitive component of Na⁺ absorption is ENaC independent and most likely occurs through an electroneutral mechanism that does not promote K⁺ secretion.

Figure 2. Analyses of J_Na, J_Cl, and J_K and V_te in CCDs isolated from collecting duct–specific ENaC–KO mice maintained on a Na⁺-depleted diet.

The control group consists of littermate mice floxed for α-ENaC but negative for the HoxB7-Cre transgene, as detailed elsewhere (22). The control group was kept on a normal Na⁺ diet to provide the zero baseline values for each variable. Statistical significance was assessed by 2-tailed unpaired Student’s t test. n = 5 in each group; *P < 0.05.

CCDs from NCC-deficient mice display thiazide-sensitive NaCl absorption.

To investigate whether the thiazide-sensitive component of Na⁺ absorption in the CCD might occur through NCC, we measured ion transport and V_te in CCDs from mice with a genetic disruption of Slc12a3, the gene encoding NCC (Ncc^–/– mice) (23). Western blot and immunofluorescence analyses confirmed the complete absence of the NCC protein in this mouse model (24, 25). CCDs from Ncc^–/– mice on a NaCl-replete diet absorbed Na⁺ and Cl^–, whereas CCDs from pair-fed wild-type mice did not (Figure 3). Thus, after genetic disruption of Slc12a3, NaCl absorption was increased in the CCD, presumably consequent to sodium depletion. NaCl absorption in CCDs from Ncc^–/– mice was amiloride insensitive but was fully inhibited by HCTZ (Figure 3). A lumen-negative V_te or K⁺ secretion in CCDs from Ncc^–/– mice was never observed in any of the experimental conditions studied (data not shown). Thus, under NaCl-replete conditions, CCDs from mice with a targeted disruption of Ncc have little ENaC-mediated Na⁺ absorption but have robust electroneutral, thiazide-sensitive NaCl reabsorption.

Figure 3. Effects of amiloride (10^–5 M) and HCTZ (10^–4 M) on Na⁺ and Cl^– transepithelial fluxes in CCDs isolated from Ncc^+/+ and Ncc^–/– mice fed a Na⁺-replete diet.

Statistical significance was assessed by ANOVA; comparisons between groups were tested by Bonferroni’s post-hoc test. n = 5 in each group; *P < 0.05 versus Ncc^+/+; ^#P < 0.05 versus Ncc^–/–; ^†P < 0.05 versus Ncc^–/– amil.

To assess the physiological relevance of this transport system, we investigated whether thiazides have a diuretic effect in vivo in the absence of NCC (Figure 4). Consistent with the in vitro studies, we observed significant HCTZ-induced natriuresis and chloriuresis in Ncc^–/– mice, although the response was smaller and delayed relative to that in Ncc^+/+ mice (Figure 4). We conclude that NCC-independent thiazide-sensitive sodium absorption participates in renal sodium absorption and regulation of sodium balance in vivo.

Figure 4. Effects of HCTZ on urinary excretion of Na⁺ and Cl^– in Ncc^+/+ and Ncc^–/– mice.

One single dose of HCTZ (50 mg/kg body weight) or vehicle was administered intraperitoneally to Ncc^+/+ and Ncc^–/– mice. Urine samples were collected from 0 to 6 and from 6 to 12 hours after injection to measure urinary Na⁺ (top panels) or Cl^– (bottom panels) excretion. Results are expressed as the ratio to urinary creatinine. Statistical significance was assessed by 2-tailed Student’s unpaired t test. n = 5 in WT groups and n = 8 in KO groups; *P < 0.05, **P < 0.01 versus vehicle.

Thiazide-sensitive NaCl absorption in the CCD is bicarbonate dependent and involves 2 ion transporters.

In many epithelia, NaCl transport occurs through a Cl^–/HCO₃^– and a Na⁺/H⁺ exchanger working in parallel. In the CCD, Cl^– absorption is eliminated with genetic ablation of Slc26a4 (20, 26), the gene encoding the Cl^–/HCO₃^– exchanger pendrin that is found in the apical regions of type B and non-A–non-B intercalated cells (27). Since Cl^– transport in the CCD occurs through pendrin, we hypothesized that HCTZ-sensitive NaCl transport results from the coupling of pendrin-mediated Cl^– reabsorption with H⁺/HCO₃^–-dependent Na⁺ transport. Supporting this hypothesis, removal of CO₂/HCO₃^– from the perfusion and bath solutions abolished both Na⁺ and Cl^– transport by perfused CCDs isolated from Ncc^–/– mice (Figure 5).

Figure 5. Effects of bicarbonate on transepithelial fluxes of Na⁺ and Cl^– in CCDs isolated from Ncc^–/– mice.

CCDs from Ncc^–/– mice were either perfused in CO₂/HCO₃^–-containing solutions or in CO₂/HCO₃^–-free (no bicarbonate [no bicarb]) buffer. Statistical significance was assessed by 2-tailed Student’s unpaired t test. n = 5 in each group; *P < 0.001 versus control.

To identify the apical sodium transporter, we measured changes in intracellular pH (pH_i) in response to luminal Na⁺ removal and then to luminal Na⁺ readdition. Experiments were performed in the absence of basolateral Na⁺ to silence basolateral Na⁺/H⁺ exchange. Intercalated cells were distinguished from principal cells by their fluorescein-conjugated peanut lectin labeling (28) and by their greater uptake of BCECF when the fluorophore was added to the perfusate (29). Whereas in CCDs isolated from wild-type mice fed a standard Na⁺-replete diet, pH_i in both intercalated and principal cells was insensitive to changes in luminal Na⁺ (data not shown), and intercalated cell pH_i in CCDs of Na⁺-depleted wild-type mice fell with removal of Na⁺ from the perfusate. This drop in pH_i was fully reversed when Na⁺ was reintroduced into the lumen (Figure 6A). These findings indicate the presence of a Na⁺-coupled acid-base transporter that is upregulated in the apical membrane of intercalated cells in response to a Na⁺-restricted diet. Furthermore, the activity of this transporter was abolished in the absence of Cl^– and greatly reduced in the nominal absence of CO₂/HCO₃^– (Figure 6A), indicating that sodium uptake is mediated by a Na⁺-driven Cl^–/HCO₃^– exchanger rather than by a Na⁺/H⁺ exchanger or a Na⁺-HCO₃^– cotransporter. Because luminal removal of Na⁺ did not elicit any detectable pH_i changes in principal cells in the CCD (data not shown), this Na⁺-driven Cl^–/HCO₃^– exchanger appears to be restricted to intercalated cells.

Figure 6. Detection and functional characterization of NDCBE in collecting ducts isolated from wild-type mice fed a Na⁺-depleted diet.

(A) Na⁺ dependence of pH_i changes in intercalated cells of CCDs isolated from mice fed a Na⁺-depleted diet. Traces are the average of pH_i changes recorded when luminal Na⁺ was removed from, and then readded to the perfusate. Upper-left panel: Both Cl^– and HCO₃^–/CO₂ were present in the extracellular fluid; mean starting pH_i (immediately before luminal Na⁺ removal) was 7.03 ± 0.04. Upper right: extracellular Cl^– was absent; mean starting pH_i was 7.03 ± 0.05. Lower left: extracellular HCO₃^–/CO₂ was absent; mean starting pH_i was 7.09 ± 0.11. Lower right: initial rates of pH_i changes during exposure to different solutions. Values are mean ± SEM, with number of tubules in parentheses. Statistical significance was tested by ANOVA followed by Bonferroni’s post-hoc test. **P < 0.01 versus all other groups. o, outside. (B) Western blot analyses of NDCBE in the mouse kidney and CCD. Western blot analyses of 50 μg membrane fraction proteins from renal cortex isolated from Ndcbe^–/– mice or wild-type mice or of proteins from 100 CCDs (corresponding to ~25 mm of tubule) isolated from wild-type mice. NDCBE was detected in the renal cortex or isolated tubules of wild-type but not of Ndcbe ^–/– mice. Note that the apparent difference in abundance of NDCBE between the lane loaded with total cortex and the one loaded with isolated CCD does not reflect an enrichment of the cortex versus the CCD, but actually only reflects the difference in the quantity of protein loaded. (C) Analyses of amiloride-resistant Na⁺ and Cl^– transepithelial fluxes in CCDs isolated from NDCBE-KO mice maintained on a Na⁺-depleted diet. Fluxes were measured in the presence of 10^–5 M amiloride in the perfusate to reflect only the amiloride-resistant, HCTZ-sensitive component of NaCl reabsorption. Values are mean ± SEM; statistical significance was assessed by unpaired Student’s t test with Welch’s correction, when appropriate. n = 5 in each group.

While many HCO₃^– transporters have been reported in the mammalian kidney, only NDCBE (encoded by Slc4a8) mediates Na⁺- and Cl^–-dependent HCO₃^– transport (30, 31). NDCBE promotes the electroneutral exchange of 1 intracellular Cl^– ion for 1 Na⁺ and 2 HCO₃^– ions. Although predominantly expressed in the brain and testis, NDCBE is also detected in the kidney, the digestive tract, the retina, the thyroid, the aorta, and the spinal cord (see Supplemental Figure 1A and Supplemental Figure 2; supplemental material available online with this article; doi: 10.1172/JCI40145DS1). Slc4a8 transcripts were confirmed in mouse CCDs by RT-PCR (data not shown). Since our preceding experiments suggested that NDCBE might be important for Na⁺ transport by intercalated cells, we genetically disrupted Slc4a8 in mice (Supplemental Figure 1, B and C). Ndcbe^–/– mice produced from heterozygous matings followed Mendelian ratios and had no obvious phenotypical abnormalities. NDCBE protein was detected by immunoblot in renal cortex and isolated CCDs of wild-type mice, but not in Ndcbe^–/– mice (Figure 6B).

To determine whether NDCBE participates in amiloride-resistant NaCl transport in mouse CCDs, we characterized Na⁺ and Cl^– transport in CCDs from Ndcbe^–/– mice. Whereas amiloride-resistant NaCl absorption was detectable in CCDs from Na⁺-depleted WT mice, NaCl absorption in the presence of luminal amiloride was not different from zero in CCDs from Na⁺ depleted Ndcbe^–/– mice, demonstrating that amiloride-resistant Na⁺ transport depends on NDCBE (Figure 6C).

To assess whether HCTZ inhibits amiloride-resistant NaCl absorption by blocking NDCBE, and/or pendrin, we next tested the effects of 10^–4 M HCTZ on Na⁺-dependent and Na⁺-independent Cl^–/HCO₃^– exchange activities. Na⁺-dependent pH changes, in the nominal presence of extracellular Cl^– and HCO₃^–, were present in CCDs of Na⁺-depleted Ndcbe^+/+ mice but completely abolished in CCDs of Na⁺-depleted Ndcbe^–/– mice (Figure 7A), confirming the role of NDCBE in mediating this process. In addition, Na⁺-dependent pH_i changes were abolished by luminal HCTZ (10^–4 M). Similarly, apical Cl^–/HCO₃^– exchange mediated by pendrin was abolished by luminal HCTZ (Figure 7B). However, when heterologously expressed in Xenopus oocytes, NDCBE activity was not significantly affected by HCTZ (0.25 mM), and pendrin activity was inhibited by HCTZ (Figure 7C), although inhibition was only partial and required higher (1 mM) HCTZ concentrations than those found to inhibit this process in isolated CCDs. This difference in sensitivity could be due to the very different experimental conditions (e.g., temperature of the assays). However, it is also possible that HCTZ has an additional indirect effect of inhibiting NDCBE or pendrin in native CCDs. As HCTZ does not block NCC exclusively, but also inhibits carbonic anhydrase (32), we next tested the effects of the carbonic anhydrase inhibitor acetazolamide (ACZ) on J_Na and J_Cl and on NDCBE and pendrin activities measured in isolated CCDs. While ACZ abolished Cl^– absorption (Figure 8A) as well as Na⁺-independent Cl^–/HCO₃^– exchange (i.e., pendrin) activity (Figure 8B), it had no effect on Na⁺ absorption (Figure 8C) or on NDCBE activity (Figure 8D), which excludes the possibility that HCTZ affects Na⁺ transport through its effect on carbonic anhydrase.

Figure 7. Effects of HCTZ (10^–4 M) on NDCBE or PDS activity in isolated collecting ducts or on recombinant NDCBE or PDS expressed in Xenopus oocytes.

(A) Effects of HCTZ and Ndcbe disruption on Na⁺-dependent pH_i changes measured in intercalated cells of CCD isolated from Na-depleted Ndcbe^–/– or Ndcbe^+/+ mice fed a low-Na⁺ diet. Traces are the average of pH_i changes recorded when luminal Na⁺ was removed and then readded, in the presence of extracellular Cl^– (122 mM) and HCO₃^– (25 mM). Intracellular Na⁺-dependent acidification was detected in Ndcbe^+/+ mice but absent in Ndcbe^–/– mice or when HCTZ 10^–4 M was present in the perfusate. In these 3 different experimental conditions, mean starting pH_i values were 7.10 ± 0.02, 6.93 ± 0.11, and 7.01 ± 0.04, respectively. (B) Effects of HCTZ on apical Cl^–/HCO₃^– exchange activity in intercalated cells of CCDs isolated from Na-depleted animals. Traces are the average of pH_i changes recorded when luminal Cl^– was removed and then readded, in the presence of extracellular HCO₃^– (25 mM) and in Na⁺-free solutions. Intracellular Cl^–-dependent alkalinization, reflecting apical Cl^–/HCO₃^– exchange, was completely abolished when 10^–4 M HCTZ was present in the perfusate. Mean starting pH_i values (immediately before Cl^– removal) were 6.91 ± 0.03 and 6.89 ± 0.08, in the absence and presence of HCTZ, respectively. (C) Effects of HCTZ on mNdcbe-mediated HCO₃^– influx. Oocytes had been injected with mNdcbe cRNA or H₂O and incubated with HCTZ (0.25 mM). As a control, NDCBE-expressing and H₂O-injected oocytes were incubated with vehicle (methanol). Values are mean ± SEM with 6–9 oocytes per group. **P < 0.01, ***P < 0.001 versus H₂O-injected oocyte. HCO₃^– flux was unaffected by the application of HCTZ (0.25 mM) compared with vehicle alone (P = 0.279). (D) Effects of HCTZ on Pds-mediated ³⁶Cl^– uptake. Pendrin-expressing oocytes (mPds) were incubated in ND96 containing 0.1 or 1 mM HCTZ during the uptake period (16 minutes). As a control, pendrin-expressing and H₂O-injected oocytes were incubated with vehicle (methanol). Values are mean ± SEM, with 6–16 oocytes per group. *P < 0.001 versus H₂O; ^†P < 0.001 versus mPds, HCTZ 0.1 mM.

Figure 8. Effects of ACZ (10^–4 M) on NDCBE- or PDS-dependent transport in isolated collecting ducts.

(A) Effects of 10^–4 M ACZ on Cl^– transepithelial transport in CCDs isolated from Ncc^–/– mice. CCDs were isolated from Ncc^–/– mice and bathed and perfused with CO₂/HCO₃^–-containing solutions. Statistical significance was assessed by 2-tailed Student’s unpaired t test. n = 5 in each group; *P < 0.001 versus control. (B) Effects of luminal 10^–4 M ACZ on pendrin activity in isolated CCDs. Tubules were isolated from wild-type mice. Pendrin activity was assessed by measuring changes in pH_i when Cl^– was removed and then readded from the perfusate. Both bath and perfusate solutions contained 25 mM HCO₃^– and were sodium-free. Traces represent the average of recordings from independent tubules. n = 4–5 independent tubules by group. Mean starting pH_i values were 6.91 ± 0.03 and 6.72 ± 0.05, in the absence and presence of ACZ, respectively. (C) Effects of luminal 10^–4 M ACZ on Na⁺ transepithelial transport in CCDs isolated from Ncc^–/– mice. Statistical significance was assessed by 2-tailed Student’s unpaired t test. n = 5 in each group. (D) Effects of ACZ 10^–4 M on NDCBE activity in isolated CCDs. Tubules were isolated from wild-type mice. NDCBE activity was assessed by measuring changes in pH_i of intercalated cells when Na⁺ was removed and then readded from the perfusate. Both bath and perfusate solutions contained 25 mM HCO₃^– and 122 mM Cl^–. The bath solution was sodium-free to silence basolateral Na⁺/H⁺ exchanger activity. Traces represent the average of recordings from independent tubules. n = 4–5 independent tubules by group. Mean starting pH_i values were 7.03 ± 0.04 and 6.99 ± 0.05, in the absence and presence of ACZ, respectively.

Taken together, these data confirm that HCTZ inhibits NDCBE and pendrin in the intact tubule and thereby amiloride-resistant electroneutral NaCl absorption in the CCD.

Discussion

Only two apical sodium transporters are established in the distal nephron where aldosterone modulates sodium, potassium, and acid-base homeostasis: the thiazide-sensitive cotransporter NCC (4), which mediates electroneutral NaCl cotransport, and the amiloride-sensitive sodium channel ENaC (5), which mediates electrogenic Na⁺ absorption (Recently, the Na⁺-HCO₃^– cotransporter NBCn1/Slc4a7 has also been shown to be present in the collecting duct [refs. 33, 34]. However, no evidence has been reported yet that NBCn1 participates in transepithelial Na⁺ absorption.) Except at the very end of the DCT (or DCT2), there is no overlap of expression of the 2 proteins, as NCC is restricted to the DCT and ENaC to principal cells of the connecting tubule and collecting duct, respectively (35). Here, we identified what we believe to be a new mechanism of apical NaCl uptake in the collecting duct that results from parallel operation of 2 bicarbonate transporters: the Na⁺ driven Cl^–/HCO₃^– exchanger NDCBE and the Na⁺-independent anion exchanger pendrin. In vivo and in isolated tubules, this mechanism mediates net electroneutral thiazide-sensitive NaCl reabsorption in the CCD, thereby exhibiting an “NCC-like” activity. Our findings explain why thiazides block 50% of sodium absorption in rat CCD (17–19), although NCC was repeatedly shown to be absent from this nephron segment in different species (36–39). Moreover, the demonstration of thiazide-sensitive NaCl absorption in mice with genetic ablation of NCC (Figure 3) definitively rules out a possible involvement of the latter in this process. It is likely that the Na⁺ reabsorption pathway we describe here plays a relevant role in the regulation of sodium balance, since it is stimulated in response to either dietary sodium restriction (Figure 1) or renal salt wasting upon disruption of NCC (Figure 3); moreover, its inhibition by HCTZ increases Na⁺ and Cl^– excretion in NCC-deficient mice (Figure 4). Thus, our data imply that the anti-hypertensive action of thiazides might be, at least partially, mediated by inhibition of sodium transport in the CCD and not exclusively in the DCT. The observations that mice with impaired expression of ENaC all along the distal nephron exhibit marked renal salt wasting, mimicking pseudohypoaldosteronism type I (PHA1) (40), whereas mice with collecting duct–specific ablation of ENaC maintain sodium balance even when challenged with a low-sodium diet (22), have suggested that the collecting duct does not importantly participate in the regulation of extracellular volume, since increased ENaC activity and expression in the late DCT and the CNT are able to fully compensate for ENaC deletion in the CCD (22, 41). However, the present finding that collecting ducts isolated from these mice exhibit net electroneutral NaCl absorption after sodium deprivation also supports the concept that ENaC is not the only important mechanism responsible for adaptive changes of sodium absorption in the collecting duct. Indeed, based on our results, it is possible that NDCBE/pendrin might also represent part of the compensatory mechanism. Nevertheless, the observation that complete ENaC deletion leads to PHA1 indicates that electroneutral NaCl absorption through the intercalated cells cannot fully replace ENaC-mediated Na⁺ absorption but rather plays a complementary role.

Within the distal nephron, sodium transport is important not only for sodium balance regulation but also for potassium and acid-base homeostasis. In fact, sodium absorption through ENaC in principal cells is electrogenic and generates a lumen-negative transepithelial voltage (Figure 1), which in turn stimulates K⁺ and H⁺ secretion. The importance of this mechanism is highlighted by the features of primary hyperaldosteronism, or Conn syndrome, in which the excess of aldosterone, the main hormone stimulating ENaC, promotes sodium retention and arterial hypertension together with renal hypokalemia and metabolic alkalosis. Conversely, blockade of ENaC, for example, by amiloride-related diuretics, leads to hyperkalemia and metabolic acidosis. However, under certain circumstances, such as dietary sodium restriction, which also stimulates aldosterone secretion, sodium balance is maintained by an increase in distal nephron sodium absorption without any alteration of K⁺ or H⁺ homeostasis (42). This phenomenon, known as the “aldosterone paradox,” implies that, depending on the needs of the organism, the kidney is able to increase distal nephron sodium absorption with or without promoting K⁺ or H⁺ secretion. Recently, it was shown that the balance between electroneutral NaCl absorption by NCC within the DCT and electrogenic Na⁺/K⁺(H⁺) “exchange” promoted by ENaC in the connecting tubule and the collecting duct is finely tuned by the WNK pathway (43, 44). Our finding that the collecting duct also performs electroneutral NaCl absorption suggests that this balance might also occur within the CCD: The relative contribution of ENaC and pendrin/NDCBE to NaCl absorption in the CCD determines the magnitude of the transepithelial voltage in the CCD and thereby modulates K⁺ and H⁺ secretion. In the different situations studied here, electroneutral NaCl accounted for 40%–100% of total transepithelial Na⁺ absorption. In Ncc^–/– mice, a model in which upregulation of ENaC is thought to promote K⁺ wasting (45), we observed that electroneutral NaCl transport was the dominant mechanism accounting for sodium absorption in the collecting duct. This suggests that Ncc^–/– mice, by favoring this electroneutral pathway, are able to maintain NaCl balance while minimizing K⁺ loss, as attested by the absence of overt hypokalemia in this model when dietary K⁺ intake is maintained at a relatively high level (23, 45).

In summary, we have demonstrated what we believe to be a novel role of NDCBE in mediating ENaC-independent, thiazide-sensitive, and electroneutral Na⁺ reabsorption in the CCD. This finding has important implications for understanding the action of thiazides on Na⁺ reabsorption and blood pressure. Furthermore, the finding challenges the current concept of a functional separation between principal cells for the regulation of sodium and potassium balance and intercalated cells for acid-base regulation.

Methods

Animals.

Ncc^–/– and collecting duct–specific ENaC–KO mice have been characterized previously (22, 23). The generation of Slc4a8^–/– mice is described below. Studies were performed in a pure C57BL/6 background for the Ncc strain and a mixed 129SV/C57BL/6 background for the other strains. The appropriate littermates were used as controls. All animal protocols were approved by the review board of the Centre de Recherche des Cordeliers, Paris, France.

Generation of Slc4a8-KO mice.

A clone isolated from a 129/SvJ mouse genomic λ library (Stratagene) was used to construct the targeting vector. An approximately 11-kb EcoRI/KasI fragment including exons 10–16 of the Slc4a8 gene was cloned into the pKO-V901 plasmid (Lexicon Genetics) with a phosphoglycerate kinase (pgk) promoter–driven diphtheria toxin A cassette. A pgk promoter–driven neomycin resistance cassette flanked by loxP sites was inserted into the MfeI site in intron 11. A third loxP site and an additional EcoRI site were inserted into the KpnI site in intron 12. The construct was electroporated into R1 mouse embryonic stem cells. Neomycin-resistant clones were analyzed by Southern blot using EcoRI and an external approximately 500-bp probe. Correctly targeted ES cells were transfected with a plasmid expressing Cre-recombinase to remove the neomycin cassette and exon 12. Correctly recombined clones were identified with an internal second probe by Southern blot analysis after EcoRI digestion. Two independent embryonic stem cell clones were injected into C57BL/6 blastocysts to generate chimeras that were backcrossed with C57BL/6 mice. Studies were performed in a mixed 129Sv/C57BL/6 background in the F₆ and F₇ generation. Genotypes were determined either by Southern blot or by PCR of tail biopsy DNA. For PCR genotyping, the sense primers F1 (5′-GGCTAGGCAGTTCTTATCTTTCCC-3′), F2 (5′-GAGCAGCCCAGATGTACACCAGC-3′) and the antisense primers R1 (5′-GGCAATCCCCGTCATGGACG-3′) were used in a single PCR mix. The primer pair F1/R1 amplified a 320-bp wild-type allele, and the primer pair F2/R1 a 423-bp KO allele. Southern and Northern blot analyses were performed as described in ref. 46.

Antibody generation and Western blot analysis.

The NDCBE antiserum was raised in rabbits against the epitope ALSINSGNTKEKSPFN (amino acids 1,074–1,089, accession number NP_067505) and affinity purified. For Western blot analyses, 10–60 μg of the membrane-enriched protein fraction was separated on reducing 7.5% SDS-polyacrylamide gels. Blots were probed with the rabbit NDCBE antibody at a dilution of 1:250. Detection was done with the chemiluminescence ECL kit (Amersham Biosciences).

In vitro microperfusion, transepithelial ion fluxes, and pH measurements.

CCD segments were isolated from corticomedullary rays under a dissecting microscope with a sharpened forceps. Because CCDs are highly heterogeneous, relatively short segments (0.45–0.6 mm) were dissected to maximize the reproducibility of the isolation procedure. In vitro microperfusion was performed as described by Burg et al. (47). Because CCDs from mice are frequently unstable and collapse rapidly, measurements were conducted during the first 90 minutes of perfusion. Usually, collections from 4 periods of 15 minutes were performed in which 15–20 nl of fluid were collected. The volume of the collections was determined under water-saturated mineral oil with calibrated volumetric pipettes. For [Na⁺], [K⁺], and [creatinine] measurements, 11–15 nl were required, while 2–3 nanoliters were used for [Cl^–] determinations. Transepithelial voltage (V_te) was measured continuously between Ag-AgCl electrodes connected to 0.15-M NaCl-agar bridges inserted in the perfusion pipette and bathing solutions. Values for each period were averaged.

Intracellular pH was monitored using the pH-sensitive dye BCECF (29). Intracellular dye was calibrated at the end of each experiment using the high-[K⁺]/nigericin technique (48). Briefly, tubules were perfused and bathed with a HEPES-buffered, 95-mM K⁺ solution containing 10 μM of the K⁺/H⁺ exchanger nigericin. Four different calibration solutions, titrated to 6.5, 6.9, 7.3, or 7.5, were used. V_te was measured continuously as described elsewhere (49). [Na⁺], [K⁺], and [creatinine] measurements were performed by HPLC (50). [Cl^–] was measured by microcoulometry (51). For each collection, ion flux (J) was calculated and reported to the length of the tubule: J_Na = [([Na]_perf × V_perf) – ([Na]_coll × V_coll)]/l; J_K = [([K]_perf × V_perf) – ([K]_coll × V_coll)]/l; and J_Cl = [([Cl]_perf × V_perf) – ([Cl]_coll × V_coll)]/l, where “perf” indicates perfusate and “coll” indicates collection fluid.Therefore, positive values indicate net absorption, whereas negative values indicate net secretion of the ion. For each tubule, the mean of the 4 collection periods was used.

Measurement of pendrin and NDCBE activities in Xenopus oocytes.

Full-length mouse pendrin and NDCBE cDNAs were cloned by PCR from isolated CCDs using the mouse pendrin-specific primers forward 5′-GTCATCCCTCGTCGCATC-3′ and reverse 5′-TCTCAGGAAGCAAGTCTACGC-3′; and the mouse NDCBE-specific primers forward 5′-CGCGGATCCGCCACCATGCCCGCCGGGAGCAA-3′ and reverse 5′-GCTCTAGATCAGTTGAAGGGGCTTTTTTCTTTTGTATTTCCGG-3′. Both PCR products were subsequently ligated into pGH19 (52) and linearized with XhoI. cRNAs were transcribed using T7 RNA polymerase (mMESSAGE mMACHINE), and their quality was assessed by spectroscopy and agarose gel electrophoresis.

Pendrin activity was assessed as pendrin-dependent ³⁶Cl uptake. Oocytes were prepared from Xenopus laevis as described previously (52). Briefly, stage V–VI oocytes were injected with 25 ng of pendrin cRNA and kept at 16–18°C in ND96 (in mM: 96.0 NaCl, 2.0 KCl, 1.0 CaCl₂, 1.8 MgCl₂, 5.0 HEPES, pH 7.4; supplemented with 5 mM sodium pyruvate and 50 U/ml penicillin/streptomycin) for 48 hours before measuring ³⁶Cl uptake. Oocytes were washed twice at room temperature in 1 ml chloride-free buffer (in mM: 98 potassium-gluconate, 1.8 hemi-calcium-gluconate, 1 hemi-magnesium-gluconate, 5 Tris-HEPES, pH 7.5) and then incubated in 500 μl uptake medium (in mM: 100 potassium-gluconate, 5 Tris, pH adjusted to 7.5 with HEPES) containing 1.74 mM ³⁶Cl for 16 minutes. The oocytes were then washed 3 times in ice-cold chloride-free buffer and subsequently lysed individually in 200 μl of 10% SDS. The radioisotope content of each individual oocyte was measured by scintillation spectrometry after adding 3 ml scintillation fluid (Opti-Fluor, Packard).

NDCBE activity was assessed as NDCBE-dependent bicarbonate influx as described previously (31). Oocytes were injected with either 25 ng of mNDCBE cRNA or with H₂O for controls and incubated for 6 days at 18°C in OR3 medium (Leibovitz’s L15 medium diluted to approximately 200 mOsm/kg H₂O) supplemented with penicillin and streptomycin. The day of assay, oocytes were incubated for approximately 5 hours in ND96 containing either 0.5% methanol (vehicle) or ND96 containing 0.25 mM HCTZ/0.5% methanol. The pH_i of the oocytes was monitored as the cells were perfused with a solution containing 5% CO₂/33 mM HCO₃^– in the continued presence of HCTZ and/or vehicle. The measurement of pH_i using a H⁺-selective microelectrode has been described in detail elsewhere (53). Briefly, each oocyte is placed in a plastic perfusion chamber and impaled with a H⁺-selective microelectrode and a KCl-filled reference electrode. The cell is first perfused with ND96 solution until a stable pH_i reading is obtained, then the perfusion system is switched to deliver a 5% CO₂/33 mM HCO₃^––containing solution. An initial CO₂-induced acid load was followed by a pH_i increase that was converted — using the calculated buffering power of each oocyte — into a measure of “HCO₃^– flux” into the cells. Data are acquired using an FD223 dual-channel differential electrometer and analyzed using in-house software. HCO₃^– influx data (rates of pH_i increase) are converted into HCO₃^– flux (mM/s) using the calculated buffering power of the oocyte (change in pH_i due to entry of CO₂ plus the open-system buffering power due to HCO₃^–).

Statistics.

Experimental results are summarized as mean ± SEM. All statistical comparisons were made by use of unpaired Student’s t test or by ANOVA followed by a Bonferroni’s post-hoc test when appropriate. A P value less than 0.05 was considered significant.