Cloning, Characterization, and Chromosomal Mapping of a Human Electroneutral Na+-driven Cl-HCO3Exchanger*

The electroneutral Na+-driven Cl-HCO3 exchanger is a key mechanism for regulating intracellular pH (pH i ) in neurons, glia, and other cells. Here we report the cloning, tissue distribution, chromosomal location, and functional characterization of the cDNA of such a transporter (NDCBE1) from human brain (GenBankTM accession number AF069512). NDCBE1, which encodes 1044 amino acids, is 34% identical to the mammalian anion exchanger (AE2); ∼50% to the electrogenic Na/HCO3 cotransporter (NBCe1) from salamander, rat, and humans; ∼73% to mammalian electroneutral Na/HCO3 cotransporters (NBCn1); 71% to mouse NCBE; and 47% to a Na+-driven anion exchanger (NDAE1) fromDrosophila. Northern blot analysis of NDCBE1 shows a robust ∼12-kilobase signal in all major regions of human brain and in testis, and weaker signals in kidney and ovary. This human gene (SLC4A8) maps to chromosome 12q13. When expressed inXenopus oocytes and running in the forward direction, NDCBE1 is electroneutral and mediates increases in both pH i and [Na+] i (monitored with microelectrodes) that require HCO 3 − and are blocked by 4,4′-diisothiocyanostilbene-2,2′-disulfonic acid (DIDS). The pH i increase also requires extracellular Na+. The Na+:HCO 3 − stoichiometry is 1:2. Forward-running NDCBE1 mediates a36Cl efflux that requires extracellular Na+ and HCO 3 − and is blocked by DIDS. Running in reverse, NDCBE1 requires extracellular Cl−. Thus, NDCBE1 encodes a human, electroneutral Na+-driven Cl-HCO3 exchanger.

The first transporter shown to be involved in the regulation of intracellular pH (pH i ) was the Na ϩ -driven Cl-HCO 3 exchanger, initially described in squid axons (1)(2)(3), snail neurons (4 -6), and barnacle muscle (7). This acid extruder (i.e. a transporter that behaves as if it mediates net H ϩ efflux) could function according to any of the four schemes (8) in Fig. 1A. In physiology experiments on mammalian cells, it is often extremely difficult to distinguish this transporter from either an electroneutral Na/HCO 3 cotransporter (NBCn1, Fig. 1B) (9,10) or an electrogenic Na/HCO 3 cotransporter (NBCe1, Fig. 1C) (11,12) because of problems depleting cells of Cl Ϫ or measuring very small electrical changes. In the absence of electrical data, one could not distinguish an electroneutral Na ϩ -driven Cl-HCO 3 exchanger from the scheme in Fig. 1D, which is a hybrid of those in Fig. 1, A-C. The Na ϩ -driven anion exchanger (NDAE1) recently cloned from Drosophila (13) does not require HCO 3 Ϫ and could function according to either of the top two schemes in Fig. 1A, but with OH Ϫ replacing HCO 3 Ϫ . In mammalian cells, increases in pH i that appear to depend on Na ϩ , Cl Ϫ , and HCO 3 Ϫ have been described in neurons (14 -16), astrocytes (17,18), renal mesangial cells (19), corneal endothelium (20), bile duct (21), aortic endothelium (22), spermatozoa (23), and various transformed cells (24 -26). However, given the difficulties noted above, definitively ascribing a cell phenotype to a transporter will require molecular tools. It is therefore extremely important not only to clone the genes encoding various Na ϩ -driven HCO 3 Ϫ transporters, but also to assign their function unambiguously to one of the schemes in Fig. 1.
Here we report the tissue distribution, chromosomal location, and functional characterization of a cDNA that we cloned from human brain (GenBank TM accession number AF069512 and NCBI accession number AAC82380). Our physiological analysis indicates that this cDNA encodes an electroneutral Na ϩ -driven Cl-HCO 3 exchanger (NDCBE1, Fig. 1A).

EXPERIMENTAL PROCEDURES
Cloning of NDCBE1-We cloned NDCBE1 in three parts. After performing a BLAST search, using the salamander NBCe1 cDNA sequence (GenBank TM accession number AF001958) as the query, of the Gen-Bank TM data base, we obtained the central part as a cDNA expressed sequence tag (EST) 1 clone AA775966 (catalogue number CDNA-1401, Genome System Inc., St. Louis, MO). We obtained the 5Ј-end by performing rapid amplification of cDNA ends (RACE). Using human brain poly(A) ϩ RNA (CLONTECH, Palo Alto, CA) as the template, we generated cDNA using an NDCBE1-specific primer corresponding to nucleotide sequence 598 -627 (numbered from first nucleotide of open reading frame). The downstream, NDCBE1-specific primers for RACE corresponded to nt 547-579 and nt 328 -358. We used the two upstream primers provided in the RACE kit (Life Technologies, Inc.). We obtained the 3Ј-end by performing a nested polymerase chain reaction (PCR), using a human brain ZAPII cDNA library (gift of Dr. Nancy Lynn Johnston, John Hopkins University) as the template. The upstream, * This work was supported by National Institutes of Health Grant NS18400. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF069512 (for the electroneutral Na ϩ -driven Cl-HCO 3  ¶ Supported by the American Heart Association. 1 The abbreviations used are: EST, expressed sequence tag; RACE, rapid amplification of cDNA ends; nt, nucleotide(s); PCR, polymerase chain reaction; UTR, untranslated region; FISH, fluorescence in situ hybridization; DIDS, 4,4Ј-diisothiocyanatostilbene-2,2Ј-disulfonic acid; bp, base pair(s); kb,kilobase(s); BAC, bacterial artificial chromosome. NDCBE1-specific primers corresponded to nt 1876 -1905 and nt 2014 -2043, and the downstream primer corresponded to a sequence near the polycloning site in the pBluescript vector. We verified that the three cDNA fragments represent a single transcript by performing PCR using an upstream primer corresponding to a region (nt Ϫ44 to Ϫ18) in the 5Ј-untranslated region (UTR) and a downstream primer corresponding to a region in the 3Ј-UTR (nt 3136 -3165). We obtained the consensus sequence by directly sequencing the full-length PCR product (Keck Sequencing Center, Yale University). We also subcloned the full-length PCR product into the oocyte expression vector pGH19 (27), sequenced the clone, and corrected PCR errors on the basis of the consensus sequence. The full-length sequence (GenBank TM accession number AF069512) was released in 1998.
FISH Mapping-Using a NDCBE1 cDNA as template, we generated a 304-bp cDNA probe, corresponding to a unique region (nt 54 -358). DNA clone 477 L 11 from the RPCI-11 human BAC library was identified by Research Genetics, Inc. (Huntsville, AL). The purified BAC DNA was labeled with biotin-dUTP by nick translation. DNA of a chromosome-12 painting library was labeled with Cy3-dUTP by PCR. A biotin-labeled BAC probe, alone or together with Cy3-labeled chromosome-12 painting probe, was hybridized to metaphase chromosome spreads in the presence of human Cot-1 DNA and salmon sperm DNA. The biotin-labeled probe was detected by avidin-fluorescein isothiocyanate. Fifty metaphase spreads were taken for analysis and measurements. Gray scale images were obtained using an Olympus epifluorescence microscope coupled to a cooled CCD camera (Photometrics Ltd., Tucson, AZ). Fractional length measurement and band assignment were established by analysis of ten chromosomes (28).
Northern Analysis-Northern blots from various human tissues (catalogue numbers 7760-1 and 7759-1) were obtained from CLONTECH. The [ 32 P]dCTP-labeled, randomly primed 671-bp cDNA probe was generated to the unique 5Ј-region of NDCBE1 (nt Ϫ44 to 627). Membranes were incubated overnight at 68°C in ExpressHyb™ hybridization buffer (CLONTECH) containing the 32 P-labeled probe. Subsequently, membranes were washed at room temperature in 2 ϫ SSC, 0.05% SDS for 40 min and then at 50°C in 0.1 ϫ SSC, 0.1% SDS for 1.5 h, before being exposed to Kodak X-Omat film at Ϫ80°C for 24 h for detection of high-intensity signals.
Oocytes-We transcribed NDCBE1 cDNA in vitro using an mMessage mMachine™ kit (Ambion, Austin, TX) with T7 RNA polymerase. Defolliculated Xenopus laevis oocytes (Stage V-VI) were prepared as described previously (29) and injected with 50 nl of NDCBE1 cRNA (1 g/l) or water and incubated in OR3 media. Injected oocytes were maintained for 3-7 days at 18°C before use. For experiments in which we reversed NDCBE1, the 50-nl injectate contained not only NDCBE1 cRNA (1 g/l), but also cRNA encoding the amiloride-sensitive epithelial Na ϩ channel (ENaC; 0.2 g/l, gift of Dr. Cecilia Canessa, Yale University). Immediately after this coinjection, we added 20 M amiloride to the oocyte culture media. One hour prior to the experiment, we transferred coinjected oocytes into amiloride-free HEPES solution.
Electrophysiology-The voltage, pH-and sodium-sensitive microelectrodes, were prepared as described previously (10,29,30). The pH electrode tip was filled with proton ionophore 1 mixture B (Fluka Chemical Corp., Ronkonkoma, NY) and back-filled with a pH 7 phosphate buffer (31). The Na ϩ electrode tip was filled with sodium ionophore 1 mixture A (Fluka Chemical Corp.) and back-filled with 10 mM NaCl. Electrodes were connected to high-impedance electrometers (model FD-223; World Precision Instruments, Inc., Sarasota, FL), which in turn were connected to the A-D converter of a computer.
In electrophysiological experiments, the CO 2 /HCO 3 Ϫ -free ND96 solution contained (in mM) 96 NaCl, 2 KCl, 1.8 CaCl 2 , 1 MgCl 2 , and 5 HEPES (pH 7.5); osmolality was 188 -200 mOsm/kg, 22°C. In solutions equilibrated with 1.5% CO 2 (pH 7.50), 5% CO 2 (pH 7.50), and 20% CO 2 (pH 6.9), we replaced 10, 33, and 33 mM NaCl, respectively, with an equivalent molarity of NaHCO 3 . Ether N-methyl-D-glucammonium (NMDG ϩ ) replaced Na ϩ in Na ϩ -free solutions, and gluconate replaced Cl Ϫ in Cl Ϫ -free solutions. In some solutions we replaced 16 mM NaCl with 16 mM of n-butyric acid sodium salt (B-5887, Sigma). 36 Cl Fluxes-Ten to twenty oocytes were incubated at room temperature for ϳ3 h in 250 l of 36 Cl "loading solution", which consisted of (in mM): 70 NaCl, 4 KCl, 1.8 CaCl 2 , 1 MgCl 2 , and 32 HEPES titrated with NaOH to pH 7.5; 36 Cl was present as 190 Ci/mmol of total Cl Ϫ . We then rapidly washed the oocytes five times with 0.5 ml of ice-cold HEPES flux solution (same as "loading solution," but without 36 Cl). The washed oocytes were transferred to one-half of a 1-ml equilibriumdialysis chamber (BelArts Products, Pequannock, NJ) containing ϳ0.5 ml of ice-cold HEPES flux solution. The other half of the dialysis chamber, modified to permit continuous inflow and outflow of solution, was placed open-side up. We added a small magnetic stirring flea, covered the opening with a nylon mesh membrane (which permits free exchange of solution between the two chamber halves), and lightly coated the open edges of the chamber half with silicon stopcock grease (High Vacuum Grease, Dow-Corning, Midland, MI), which acted as a gasket when the two chamber halves were joined and placed oocyte-side up on a magnetic stirring plate. We flowed ice-cold HEPES flux solution at 8 ml/min for 2 min to wash out extracellular 36 Cl and then flowed room temperature solution at 3 ml/min for 4 min before beginning to collect samples of the chamber effluent at 3 ml/min every 3 min. Experiments with dye indicated that the time to exchange 95% of the fluid in the upper (i.e. oocyte) chamber half was ϳ2 min. All samples were collected directly into plastic scintillation vials, to which we later added 9 ml of Ultima Gold™ liquid scintillation counting mixture (Packard Instrument Co., Meriden, CT). At the end of the experiment, the chamber was rapidly taken apart, the oocytes were transferred to 150 l of a 10% SDS solution in 1 N NaOH for digestion, and a 50-l aliquot of the digest was prepared for liquid scintillation counting. We calculated the initial 36 Cl content of the oocytes and the fractional rate of 36 Cl loss during each sampling period in the experiment. The CO 2 /HCO 3 Ϫ flux solution used in the 36 Cl flux experiments was the same as the HEPES flux solution, except that 33 mM HCO 3 Ϫ replaced 33 mM Cl Ϫ , and the solution was equilibrated with 5% CO 2 , 95% O 2 .
Statistics-Data are expressed as mean Ϯ S.E. Statistical significance was judged from unpaired Student's t tests.

Molecular Characterization
Cloning-Querying with the sequence of the cDNA encoding salamander NBCe1 (12), we searched the GenBank TM data base and found a human brain EST clone (accession number AA775966) that, at one end, was 53% identical to query. Sequencing this EST clone revealed a 2-kb open reading frame, representing the center of the full-length clone. We obtained the 5Ј-end by RACE on human brain RNA and the 3Ј-end by PCR on a human frontal-lobe cDNA library. We obtained the full-length clone, which encodes 1044 amino acids, by performing PCR on human brain cDNA, using primers designed to amplify the entire open reading frame as well as portions of the 5Ј-and 3Ј-UTRs.
Chromosomal Mapping-An NDCBE1 BAC clone produced clear FISH signals on a pair of chromosomes (not shown), which, on the basis of their size, morphology, and DAPI stainbanding pattern, we identified as chromosome 12. Cohybridization of this BAC clone with a chromosome-12 painting probe confirmed the identification (Fig. 2C). The BAC clone hybridized 22% of the distance from the centromere to the telomere of arm 12q, corresponding to band 12q13 (Fig. 2D). In contrast, human NBCe1 (SLC4A5) maps (35) to chromosome 4q21, and human NBCn1 (SLC4A7) maps (36) to 3p22.
Tissue Distribution of mRNA-A Northern blot analysis of multiple human tissues (Fig. 2E) revealed a ϳ12-kb transcript, with strong signals in brain and testis and a weaker signals in kidney Ͼ ovary. The weak ϳ9.5-kb bands (pancreas Ͼ kidney) may represent NBCe1 (35,37). The very weak ϳ7.5-kb band (testis) may represent the human ortholog of NBCn1 (38). The bands at ϳ6.3 kb (brain Ͼ testis Ͼ kidney), ϳ4.2 kb (testis), and ϳ3.3 kb (brain Ͼ testis) may represent alternative splicing of the NDCBE1 primary transcript or products of different but related genes. We found that the three bands that appear in the Northern blot of whole brain also are present in multiple brain regions (not shown), including cerebral cortex, cerebellum, medulla, thalamus, and hippocampus. However, NDCBE1 was notably absent from spinal cord.

Physiological Characterization
Na ϩ Dependence of pH i Recovery-To determine the function of NDCBE1, we injected cRNA into oocytes and used microelectrodes to monitor pH i and membrane potential (V m ). In oocytes expressing NDCBE1 (Fig. 3A), extracellular 1.5% CO 2 , 10 mM HCO 3 Ϫ elicited a rapid fall in pH i due to CO 2 influx (39), followed by an increase due to HCO 3 Ϫ uptake. Removing extracellular Na ϩ converted the pH i recovery to a very slow acidification, reflecting reversal of NDCBE1. In water-injected oocytes, 1.5% CO 2 , 10 mM HCO 3 Ϫ caused the usual pH i fall,
Electroneutrality-In oocytes expressing NDCBE1 (Fig. 3A), 1.5% CO 2 , 10 mM HCO 3 Ϫ caused a small, slow depolarization (arrow), and removing Na ϩ caused a slight hyperpolarization (arrowhead), as observed in H 2 O-injected oocytes (not shown). In contrast, with oocytes expressing electrogenic NBCs, applying CO 2 /HCO 3 Ϫ elicits a large and rapid hyperpolarization, whereas removing Na ϩ elicits a large and rapid depolarization (12,29,40). Thus, NDCBE1 is electroneutral. HCO 3 Ϫ Dependence of pH i Recovery-When we acidified ND-CBE1-expressing oocytes with butyric acid, rather than CO 2 , pH i failed to recover in the presence of Na ϩ (Fig. 3B). However, after we removed the butyric acid and applied CO 2 /HCO 3 Ϫ , pH i initially recovered rapidly, even though pH i was substantially higher than in the presence of butyric acid. Thus, NDCBE1 requires HCO 3 Ϫ . Inhibition of pH i Recovery by DIDS-Applying 0.5 mM DIDS almost completely blocks the pH i recovery (Fig. 3C). In six experiments, the inhibition averaged 95% Ϯ 10%. Thus, ND-CBE1 is DIDS sensitive. 36 Cl Efflux-When we introduced 5% CO 2 , 33 mM HCO 3 Ϫ (pH 7.50) to NDCBE1-expressing oocytes, the 36 Cl efflux increased more than 3-fold (Fig. 3D). The slow decline in 36 Cl efflux probably reflects a NDCBE1-mediated increase in pH i , [HCO 3 Ϫ ] i and [Na ϩ ] i . These changes were absent in oocytes injected with water, rather than NDCBE1 cRNA. Moreover, in NDCBE1-expressing oocytes, the transition from HEPES to CO 2 /HCO 3 Ϫ did not affect 36 Cl efflux in either the absence of Na ϩ or presence of 0.5 mM DIDS. Thus, while NDCBE1 is mediating Na ϩ -dependent HCO 3 Ϫ uptake, it also mediates a Cl Ϫ efflux with the properties expected of a Na ϩ -driven Cl-HCO 3 exchanger.

FIG. 3. Functional characterization of NDCBE1 expressed in
Xenopus oocytes and running in forward direction. In A-C, oocytes were impaled with pH and voltage microelectrodes. A, Na ϩ dependence and electroneutrality. Just before Na ϩ removal, pH i increased at a mean rate (dpH i /dt) of 9.9 Ϯ 2.5 ϫ 10 Ϫ5 pH units s Ϫ1 (n ϭ 12). After removal of extracellular Na ϩ (arrowheads), dpH i /dt was Ϫ4.4 Ϯ 2.6 ϫ 10 Ϫ5 pH units s Ϫ1 (n ϭ 6). In record of membrane voltage (V m ), the arrow indicates application of CO 2 /HCO 3 Ϫ , and arrowheads indicate the removal and return of Na ϩ . B, HCO 3 Ϫ dependence. "Butyric acid" solution contained 16 mM total butyrate at pH 7.5. During flat portion of butyric acid record, dpH i /dt averaged 0.69 Ϯ 0.13 ϫ 10 Ϫ5 pH units s Ϫ1 (mean pH i ϭ 6.94 Ϯ 0.04), whereas during initial rising phase of CO 2 /HCO 3 Ϫ record, dpH i /dt averaged 12.1 Ϯ 1.0 ϫ 10 Ϫ5 pH units s Ϫ1 (mean pH i ϭ 7.03 Ϯ 0.02; n ϭ 5; paired experiments with random order; p ϭ 0.011 for pH i comparison, and p Ͻ 0.00001 for dpH i /dt comparison, one tail). C, DIDS sensitivity. D, 36 Cl efflux, presented as normalized fractional loss of 36 Cl. We discarded the first three 3-min samples to allow 36 Cl flux to stabilize. We normalized data to unity at the first sample shown in figure. The average, initial rate constants were 0.00024 s Ϫ1 for "control", 0.00034 s Ϫ1 for "0-Na ϩ ", 0.00016 s Ϫ1 for "DIDS" oocytes, which all were expressing NDCBE1; the value was 0.00013 s Ϫ1 for water-injected oocytes. For control, the average rate constant of the last four samples in CO 2 /HCO 3 Ϫ buffer was 0.00040 s Ϫ1 ; for DIDS, the comparable value for the last five samples was 0.00015 s Ϫ1 ; thus, the DIDS-sensitive rate constant was 0.00025 s Ϫ1 . Error bars indicate S.E. and are absent when S.E. was smaller than size of symbol. E, [Na ϩ ] i increase. The oocyte was expressing NDCBE1. F, comparison of Na ϩ and HCO 3 Ϫ transport in presence of 5% CO 2 , 33 mM HCO 3 Ϫ . For Na ϩ , values are mean rates of [Na ϩ ] i increase. For HCO 3 Ϫ , values are "pseudo-fluxes", i.e. product of dpH i /dt and total buffering power (65); the latter is the sum of intrinsic buffering power (16 mM/pH unit, computed from CO 2 -induced pH i decrease in water-injected oocytes) and CO 2 buffering power (computed from pH i and [CO 2 ]). Hash marks indicate S.E.; number of observations are in parentheses.
ports Na ϩ , we used Na ϩ -sensitive microelectrodes to monitor [Na ϩ ] i . In an oocyte-expressing NDCBE1, extracellular 5% CO 2 , 33 mM HCO 3 Ϫ caused [Na ϩ ] i to increase (Fig. 3E). The mean rate of this [Na ϩ ] i increase was substantially higher than in the presence of DIDS or in water-injected oocytes (Fig. 3F). The average, DIDS-sensitive d[Na ϩ ] i /dt was 1.01 M s Ϫ1 . In parallel experiments (not shown), we determined dpH i /dt under identical conditions, and computed the HCO 3 Ϫ pseudo-flux (Fig. 3F), the DIDS-sensitive component of which averaged 2.19 M s Ϫ1 .
Cl Ϫ Dependence of Reversed NDCBE1-We already knew that the squid axon's Na ϩ -driven Cl-HCO 3 exchanger is very difficult to reverse (3), consistent with the slow pH i decrease in 0-Na ϩ in Fig. 4A. We took two steps in an attempt to speed the reversed NDCBE1. First, we coexpressed ENaC Na ϩ channels to increase [Na ϩ ] i . Second, we exposed the oocyte to 20% CO 2 to increase [HCO 3 Ϫ ] i . Removing Na ϩ reversed NDCBE1, causing pH i to decline. However, removing extracellular Cl Ϫ reversibly blocked this decline (Fig. 4A) and caused a small hyperpolarization, as in water-injected oocytes (not shown). In waterinjected oocytes (Fig. 4B), Na ϩ removal blocked a very slow pH i recovery (probably due to endogenous Na-H exchange at very low pH i ), but Cl Ϫ removal had no effect on the pH i trajectory. Thus, the reversed NDCBE1 requires external Cl Ϫ , as expected of a Na ϩ -driven Cl-HCO 3 exchanger.

DISCUSSION
Related Transporters-After we submitted our NDCBE1 sequence to GenBank TM , two other groups (41,42) published partial sequences of NDCBE1, one a 2-kb fragment referred to as "NBC-3" (42). After we submitted our paper, a paper appeared by Wang et al. (43), who cloned from mouse insulinoma cells a cDNA named NCBE, 71% identical on the amino acid level to NDCBE1. Mouse NCBE's 5.5-kb transcript, like the transcripts of human NDCBE1, is robustly expressed in cerebrum and cerebellum. However, mouse NCBE mRNA is only weakly present in testis. The function of mouse NCBE is unclear. It mediates a 22 Na influx that largely depends on extracellular [Cl Ϫ ]. In addition, oocytes expressing this mouse clone mediate a 36 Cl efflux that is only partially external Na ϩ -dependent or DIDS-sensitive. Moreover, because no 36 Cl-efflux data are available from water-injected oocytes, it is impossible to know whether the 36 Cl-efflux data represent NCBE activity. Finally, no electrical data are available.
Human NDCBE1 is functionally similar to Drosophila NDAE1 (13) in that both exchange extracellular Na ϩ and "base" for intracellular Cl Ϫ . However, human NDCBE1 is strictly HCO 3 Ϫ -dependent, whereas Drosophila NDAE1 apparently transports OH Ϫ in the absence of HCO 3 Ϫ . Although CO 2 / HCO 3 Ϫ enhances pH i changes, it is not clear that Drosophila NDAE1 can transport HCO 3 Ϫ . The CO 2 /HCO 3 Ϫ could act strictly as a buffer, dissipating OH Ϫ unstirred layers. Another difference is that expression of Drosophila NDAE1 in oocytes is associated with a Cl Ϫ current, as well as an inward current caused by applying CO 2 /HCO 3 Ϫ . On the other hand, in oocytes expressing human NDCBE1, the V m changes caused by altering [HCO 3 Ϫ ] o or [Cl Ϫ ] o are no different than in water-injected oocytes. Because Drosophila NDAE1 and human NDCBE1 come from distantly related phyla and not closely related in terms of deduced amino acid sequence (47% identity), one must keep open the possibility that, although they appear superficially similar in some respects, Drosophila NDAE1 and human NDCBE1 may function by different molecular mechanisms (Fig. 1A).
Stoichiometry-The ratio of net HCO 3 Ϫ and Na ϩ fluxes mediated by NDCBE1 was (2.19 M s Ϫ1 )/(1.01 M s Ϫ1 ), or 2.17, consistent with the 2:1 stoichiometry expected of a Na ϩ -driven Cl-HCO 3 exchanger. Because NDCBE1 is electroneutral, we presume that the net Cl Ϫ efflux is the same as the net Na ϩ influx (Fig. 1A). However, it was impractical to measure the net Cl Ϫ efflux directly with ion-sensitive microelectrodes, because [Cl Ϫ ] i is too high relative to NDBCE1 expression. However, we can calculate the unidirectional Cl Ϫ efflux from the DIDSsensitive component of the rate constant for 36 Cl efflux in NDCBE1-expressing oocytes and the resting [Cl Ϫ ] i of NDAEexpressing oocytes (13): 0.00025 s Ϫ1 ϫ 29.5 mM ϭ 7.4 M s Ϫ1 . This unidirectional flux is ϳ7.3-fold higher than the expected net flux, suggesting that NDCBE1 mediates substantial Cl-Cl exchange in parallel with Na ϩ -driven Cl-HCO 3 exchange. Indeed, the unidirectional Cl Ϫ efflux from barnacle muscle fibers is also much higher than the net HCO 3 Ϫ efflux mediated by the endogenous Na ϩ -driven Cl-HCO 3 exchanger (44).
Conclusions-We have now cloned the electroneutral Na ϩdriven Cl-HCO 3 exchanger, the first transporter shown to regulate pH i in any cell. The heavy expression of the NDCBE1 transcript in multiple brain regions, including hippocampus, suggests that NDCBE1 plays a major role in pH i regulation in human neurons. In the rat, functional data show that the Na ϩ -driven Cl-HCO 3 exchanger is a key pH i regulator in pyramidal neurons from the hippocampal CA1 region (14). pH i is critically important for neuronal function because pH i changes substantially modulate the activity of a variety of CNS channels (45)(46)(47)(48)(49)(50)(51)(52)(53)(54)(55)(56)(57). Low pH i inhibits (and/or high pH i stimulates) spontaneous firing in neurons (16,58,59), membrane excitability (60), and epileptiform activity (61). pH i is important for CNS processes other than excitability. For example, neurite formation (62) requires HCO 3 Ϫ . Thus, NDCBE1 is in a position to influence a wide range of neuronal behaviors.