Characterization of Human SLC4A10 as an Electroneutral Na/HCO3 Cotransporter (NBCn2) with Cl– Self-exchange Activity*

The SLC4A10 gene product, commonly known as NCBE, is highly expressed in rodent brain and has been characterized by others as a Na+-driven Cl-HCO3 exchanger. However, some of the earlier data are not consistent with Na+-driven Cl-HCO3 exchange activity. In the present study, northern blot analysis showed that, also in humans, NCBE transcripts are predominantly expressed in brain. In some human NCBE transcripts, splice cassettes A and/or B, originally reported in rats and mice, are spliced out. In brain cDNA, we found evidence of a unique partial splice of cassette B that is predicted to produce an NCBE protein with a novel C terminus containing a protein kinase C phosphorylation site. We used pH-sensitive microelectrodes to study the molecular physiology of human NCBE expressed in Xenopus oocytes. In agreement with others we found that NCBE mediates the 4,4′-diisothiocyanato-stilbene-2,2′-disulfonic acid-sensitive, Na+-dependent transport of \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HCO}_{3}^{-}\) \end{document}. For the first time, we demonstrated that this transport process is electroneutral. Using Cl–-sensitive microelectrodes positioned at the oocyte surface, we found that, unlike both human and squid Na+-driven Cl-HCO3 exchangers, human NCBE does not normally couple the net influx of \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HCO}_{3}^{-}\) \end{document} to a net efflux of Cl–. Moreover we found that that the 36Cl efflux from NCBE-expressing oocytes, interpreted by others to be coupled to the influx of Na+ and \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HCO}_{3}^{-}\) \end{document}, actually represents a \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{CO}_{2}{/}\mathrm{HCO}_{3}^{-}\) \end{document}-stimulated Cl– self-exchange not coupled to either Na+ or net \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HCO}_{3}^{-}\) \end{document} transport. We propose to rename NCBE as the second electroneutral Na/HCO3 cotransporter, NBCn2.

The mammalian SLC4 family of solute carriers encompasses 10 functionally diverse proteins including Cl-HCO 3 exchangers, both electrogenic and electroneutral sodium-coupled bicarbonate transporters (NCBTs), 7 and a Na ϩ /borate cotransporter (for reviews, see Refs. 1 and 2). Electroneutral NCBTs play critical roles in regulating the intracellular pH (pH i ) of neurons (3) and in secreting HCO 3 Ϫ across the choroid plexus (4,5). These activities modulate pH in the surrounding brain extracellular fluid (BECF). Changes in both pH i and pH BECF can have profound effects on ion channels (e.g. the acidsensing ion channel ASIC; see Ref. 6), neurotransmitter receptors (e.g. some ␥-aminobutyric acid type A receptors; see Ref. 7), and neurotransmitter transporters (e.g. the SLC18 family of vesicular amine transporters; for a review, see Ref. 8), thereby influencing neuronal excitability (9,10), synaptic transmission (for a review, see Ref. 11), and other parameters. The three mammalian electroneutral NCBTs are commonly known as NBCn1, NDCBE, and NCBE and are encoded, respectively, by the SLC4A7, SLC4A8, and SLC4A10 genes.
The characterization of the SLC4A7 gene product (12) as NBCn1 (also known as NBC3) provided the first definitive molecular identification of an electroneutral NCBT (13). NBCn1 functions as a Cl Ϫ -independent Na/HCO 3 cotransporter. NBCn1 additionally mediates a Na ϩ conductance that is independent of its Na/HCO 3 cotransport activity (13). The poor sensitivity of the transporter to DIDS (12,13) matches the pharmacological profile of stilbene-insensitive NCBT evident in the basolateral membrane of medullary thick ascending limb epithelia (14,15) where NBCn1 is known to be expressed.
The SLC4A8 gene product was characterized as a Na ϩ -driven Cl-bicarbonate exchanger (NDCBE), the first mammalian electroneutral NCBT demonstrated to be Cl Ϫ -dependent. The importance of Na ϩ -driven Cl-HCO 3 exchange activity for regulating pH i has long been established by physiological means in squid giant axons (16,17), snail neurons (18), giant barnacle muscle fibers (19), and mammalian neurons (3). NDCBE is markedly different from NBCn1 as it has an absolute dependence on Cl Ϫ countertransport for its Na/HCO 3 cotransport activity.
The first slc4a10 gene product was cloned from a mouse insulinoma cell line (20), and a deletion in the human SLC4A10 gene has been linked to autism (21). In mice, the knock-out of slc4a10 results in small brain ventricles and a decreased susceptibility to seizure activity (22). The presence or absence of two major splice cassettes in rodent slc4a10 (23) results in multiple splice variants (23). The SLC4A10 gene product has been named "NCBE" for Na ϩ -driven Cl-bicarbonate exchanger. However, the data presented in support of this nomenclature, 1) NCBE-mediated 22 Na and 36 Cl influx, 2) HCO 3 Ϫ -stimulated 36 Cl efflux, and 3) bath Cl Ϫ dependence of NCBE-mediated HCO 3 Ϫ influx (20), do not prove NDCBE activity. In the present study, we present evidence that the Na/HCO 3 cotransport activity of SLC4A10 under physiological conditions is independent of any Cl Ϫ countertransport and thus that NCBE in fact normally functions as an electroneutral Na/HCO 3 cotransporter. We recommend that the common name for the SLC4A10 gene product be changed to NBCn2. A preliminary report of this work has appeared in abstract form (24).

EXPERIMENTAL PROCEDURES cDNA Constructs
Human AE1 cDNA in a Xenopus expression vector (AE1⅐pBSXG1; see Ref. 25) was a kind gift from Dr. Ashley Toye, Bristol University, Bristol, UK. We have previously reported the construction and use of the clones (a) rat NBCn1-B⅐pGH19 (cDNA encoding NBCn1 in the Xenopus expression vector pGH19; Ref. 13), (b) human NDCBE⅐pGH19 (26), (c) human NBCe1-A-EGFP⅐pGH19 (NBCe1-A with a C-terminal enhanced green fluorescent protein tag; Ref. 27), and (d) squid NDCBE⅐TOPO (under the control of a T7 promoter; Ref. 28).
NCBE⅐TOPO-Primers were designed against the 5Ј-and 3Ј-untranslated regions of the human brain SLC4A10 cDNA sequence (GenBank TM DNA accession number AB040457): sense,5Ј-GCAAGGTGCTTATTCCAGAGGCGTTAC-3Ј(nucleotides (nts) 61-87 of AB040457 where the start codon is at nts 93-95); antisense, 5Ј-TGCTTTGGGGAATCAGCTTCT-AGAGTG-3Ј (nts 3361-3387 of AB040457 where the termination codon is at nts 3357-3359). Full-length NCBE cDNA products obtained by PCR from a human Marathon kidney cDNA library (Clontech) were subcloned into the TOPO2.1 vector (Invitrogen) according to the manufacturer's instructions. The fidelity of the clones was confirmed by automated DNA sequencing performed by the Keck Facility at Yale.
NCBE⅐pGH19-For expression in Xenopus oocytes, we subcloned NCBE into pGH19. NCBE cDNA (including 32 nts of 5Ј-untranslated region and 28 nts of 3Ј-untranslated region) was amplified by PCR from NCBE⅐TOPO to include 5Ј EcoRV and 3Ј HindIII restriction sites (underlined) using the forward PCR primer 5Ј-CGAAGGATATCGCAAGGTGCTTATTC-CAG-3Ј and the reverse PCR primer 5Ј-CGAAGAAGCTTATG-CTTTGGGGAATCAGC-3Ј. The EcoRV-NCBE-HindIII PCR product was ligated into complementary SmaI and HindIII restriction sites in the pGH19 vector. NCBE translation from Met 1 (underlined) was put under the control of a strong Kozak sequence, GCCACCATGG, replacing the natural CAAAA-CATGG context using the QuikChange mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's recommendations.
Rapid Amplification of cDNA Ends, Cloning, and Sequencing of NCBE Splice Variants-The methods used to determine the presence and distribution of human NCBE splice variants are outlined in the supplemental data.

Northern Blots
Using the BioPrime DNA labeling system (Invitrogen), we generated a randomly primed 405-nt [ 32 P]dCTP-labeled probe using a cDNA template corresponding to nts 1-405 of the NCBE coding region. Northern blots of mRNAs from multiple human tissues (Clontech "Human," "Human brain II," and "Human brain IV") were hybridized with our probe for 2 h at 68°C and washed according to the manufacturer's instructions. Results were visualized by autoradiography using Biomax MR film (Eastman Kodak Co.) after a 16-h exposure to the hybridized blots at Ϫ80°C.

cRNA Synthesis and Injection into Xenopus Oocytes
cRNA Synthesis-cDNA constructs in pGH19 were linearized with NotI. AE1⅐BSXG1 was linearized with HindIII, and squid NDCBE⅐TOPO was linearized with NheI. Linearized cDNA was purified using the QIAquick PCR purification kit (Qiagen). Capped cRNA was transcribed from the linearized cDNA constructs using the T7 mMessage mMachine kit (Ambion, Austin, TX) according to the manufacturer's recommendations. cRNA was purified and concentrated using the RNeasy MinElute RNA Cleanup kit (Qiagen).
Xenopus Oocyte Isolation-Ovaries were surgically removed from anesthetized frogs in house as described previously (30) or purchased predissected from Nasco (Fort Atkinson, WI). Oocytes were separated by collagenase treatment (30). Stage V-VI oocytes were manually sorted from the total population of separated oocytes and stored in OR3 medium, containing 500 units each of penicillin and streptomycin, at 18°C until use.
Microinjection of cRNAs-One day after isolation, oocytes were injected either with 25 ng of cRNA (50 nl of a 0.5 ng/nl cRNA solution) or 50 nl of sterile water. Oocytes were stored in OR3 medium at 18°C for 3-7 days prior to assay.

Solutions
Nominally CO 2 /HCO 3 Ϫ -free "ND96" solution contained 93.5 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , and 5 mM HEPES. The pH of the solution was titrated to 7.50 with freshly prepared NaOH so that the final [Na ϩ ] was 96 mM. Our "CO 2 / HCO 3 Ϫ " solution contained 33 mM NaHCO 3 in place of 33 mM NaCl and was equilibrated with a 5% CO 2 , 95% O 2 gas mixture. In our butyrate-containing solution, which was nominally free of CO 2 /HCO 3 Ϫ , 30 mM sodium butyrate replaced 30 mM NaCl. In our Na ϩ -free "0 Na" solutions, 96 mM N-methyl-D-glucamine (NMDG ϩ ) replaced 96 mM Na ϩ . In our Cl Ϫ -free "0 Cl" solutions gluconate replaced Cl Ϫ . Solutions containing intermediate concentrations of Cl Ϫ were made by mixing Cl Ϫ -containing and Cl Ϫfree solutions in the appropriate ratio. DIDS was supplied as a disodium salt by Sigma and Research Organics, Cleveland, OH.

Electrophysiological Measurements in Xenopus Oocytes
The oocyte chamber and solution flow arrangements have been described recently in detail (27). All experiments were performed at room temperature (ϳ22°C). Oocytes were superfused with our ND96 solution to allow oocyte pH i and membrane potential (V m ), as described below, to reach a steady state prior to the application of an experimental solution. All solutions flowed at 4 ml/min. pH i Measurements-We simultaneously monitored pH i (using Hydrogen Ionophore I-Mixture B, catalog number 95293, Fluka Chemical Corp., Ronkonkoma, NY) and V m of oocytes by impaling cells with two microelectrodes as also described recently in detail (27). A brief summary of the technique is provided in the supplemental material. Data were acquired using customized software.
Surface and Intracellular [Cl Ϫ ] Measurements-In some experiments we simultaneously monitored oocyte pH i , V m , and cell surface [Cl Ϫ ] ([Cl Ϫ ] S ). We measured [Cl Ϫ ] S using a borosilicate glass microelectrode. After preparing the micropipette as in the pH i experiments, we used a microforge to break and fire polish the tip to a diameter of 5-10 m. We then filled the tip with a Cl Ϫ -selective liquid membrane (Mixture A, catalog number 24899, Fluka Chemical Corp.) and backfilled with solution containing 150 mM NaCl and 5 mM HEPES, titrated to pH 7.0 with NaOH. Because of the imperfect Cl Ϫ :HCO 3 Ϫ selectivity of the liquid membrane mixture, we performed two full calibrations of the electrode before each experiment: one in a series of CO 2 /HCO 3 Ϫ -free ND96 solutions modified to contain [Cl Ϫ ] at 5.8, 10, 19, and 30 mM and the other in a series of our CO 2 / HCO 3 Ϫ -containing extracellular solutions modified to contain [Cl Ϫ ] at 5.8, 10, 19, and 30 mM. In [Cl Ϫ ] S experiments in which one of the extracellular solutions was Na ϩ -free, we also performed a full calibration in a series of Na ϩ -free CO 2 /HCO 3 Ϫfree ND96 and Na ϩ -free CO 2 /HCO 3 Ϫ -containing solutions. We used an ultrafine computer-controlled micromanipulator (model MPC-200 system, Sutter Instrument Co., Novato, CA) to position the tip of the electrode at the oocyte surface and then to further advance it (ϳ40 m) until we observed a slight dimple in the cell membrane. Periodically during the experiment, the electrode was withdrawn 200 -300 m from the oocyte to recalibrate it at the [Cl Ϫ ] of the bulk extracellular fluid. Because the electrode is sensitive to log 10  In some experiments we measured intracellular [Cl Ϫ ] and V m using an approach similar to that in the pH i experiments except that: 1) the ion-sensitive microelectrode contained the Cl Ϫ rather than the proton mixture, 2) the backfill solution was the one described above for large tipped Cl Ϫ -sensitive electrodes, and 3) we calibrated the electrode in a series of CO 2 / HCO 3 Ϫ -free ND96 solutions modified to contain [Cl Ϫ ] at 10, 25, 40, 60, and 101 mM.

Cl Flux Experiments
Isotope-H 36 Cl (GE Healthcare) was supplied as a 0.2 M stock with an activity of 134 Ci/ml. We prepared a working stock of isotope by diluting the H 36 Cl with an equal volume of 0.2 M NaOH to create 0.1 M Na 36 Cl. 36 Cl Efflux-Detailed assay methods and efflux calculations are described in the supplemental material. In brief, oocytes were preloaded with 36 Cl overnight in nominally CO 2 /HCO 3 Ϫfree "ND96" solution, and efflux was measured twice for each oocyte. The first 30-min efflux period was always in ND96; the second 30-min efflux period was in one of a diverse range of solutions.
Data Analysis-Statistical analyses were performed in Microsoft Excel 1997 (paired and unpaired t tests assuming equal variances) and Kaleidagraph version 4.0 (one-way ANOVA with Student-Newman-Keuls multiple comparison posthoc analysis).

RESULTS
Cloning Human NCBE-Using primers specific to the untranslated regions of a published SLC4A10 transcript from human brain (GenBank accession number AB040457, encoding a variant protein that we call NCBE-A; Fig. 1), we generated a PCR product from a human kidney cDNA library. We subsequently ligated this product into the TOPO vector. DNA sequencing of four independent TOPO clones revealed four full-length cDNAs identical to NCBE-A except for an additional 90 nts corresponding to splice cassette A (Fig. 1). We deposited the sequence of the new human transcript in Gen-Bank (nucleotide accession number AY376402); here we temporarily refer to this longer protein product as NCBE-B (Fig. 1). We performed all functional assays in the present work using NCBE-B.
Human NCBE mRNA Is Widely Distributed in the Brain-Northern blotting of mRNAs prepared from various human organs, using a probe common to NCBE-A and NCBE-B, demonstrated that NCBE transcripts are predominantly expressed in the brain ( Fig. 2A, lane 2). The major transcript is ϳ6.3 kb with an additional larger product of ϳ9.5 kb. We saw no additional bands even with greatly increased periods of film exposure (not shown). Northern blotting of mRNAs pre- Human NCBE Has Multiple Splice Variants-Here we determined which of the multiplicity of NCBE splice variants, known to be expressed in rodents (23), could also be detected in a human brain cDNA library. The results of this survey, including the description of a novel C-terminal splice variant with a consensus protein kinase C phosphorylation site, are presented in the supplemental data.
Like NBCn1 and Like NDCBE, NCBE Mediates Electroneutral Na ϩ -dependent HCO 3 Ϫ Influx into Xenopus Oocytes-Our laboratory has demonstrated previously that the pH i of oocytes expressing NCBTs is able to recover from a CO 2 -induced acid load only in the presence of extracellular HCO 3 Ϫ and Na ϩ (e.g. see Ref. 31). In the present study, we found that pH i recovery (Fig. 3A, upper panel, segment bc) from a CO 2 -induced acid load (ab) is a characteristic of Xenopus oocytes expressing NCBE (mean dpH i /dt ϭ 14 Ϯ 1 ϫ 10 Ϫ5 pH units/s, n ϭ 82) but not of oocytes injected with H 2 O (Fig. 3B, upper panel; mean dpH i /dt ϭ 3 Ϯ 1 ϫ 10 Ϫ5 pH units/s, n ϭ 41). Furthermore this pH i recovery of NCBE-expressing oocytes requires HCO 3 Ϫ and is not merely a response to the fall in pH i because oocytes that were acidified in our butyrate solution failed to exhibit a pH i recovery (Fig. 4, upper panel, segment ef; mean dpH i /dt ϭ 1 Ϯ 1 ϫ 10 Ϫ5 pH units/s, n ϭ 6). In some experiments on the same oocytes, we monitored pH i recovery from a CO 2 -induced acid load either prior to (as represented in Fig. 4, upper panel, segment bc) or subsequent to (not shown) butyrate exposure. In this way we verified functional expression of NCBE in these cells (mean dpH i /dt ϭ 18 Ϯ 3 ϫ 10 Ϫ5 pH units/s, n ϭ 4). Note that the application of CO 2 /HCO 3 Ϫ to NCBE-expressing oocytes did not elicit any rapid changes in V m (arrows in Fig. 3 and Fig. 4, lower panels), indicating that the HCO 3 Ϫ -dependent pH i recovery is electroneutral.
To investigate cation dependence of HCO 3 Ϫ transport mediated by NCBE, we acidified NCBE-expressing oocytes (Fig. 5, upper panel, ab) in a Na ϩ -free CO 2 /HCO 3 Ϫ solution in which NMDG ϩ entirely replaced Na ϩ . The pH i of NCBE-expressing oocytes was unable to recover from this acid load (bc; mean dpH i /dt ϭ Ϫ2 Ϯ 1 ϫ 10 Ϫ5 pH units/s, n ϭ 6) under these conditions. Replacing bath NMDG ϩ with Li ϩ did nothing further to permit HCO 3 Ϫ transport (cd; mean dpH i /dt ϭ Ϫ4 Ϯ 3 ϫ 10 Ϫ5 pH units/s, n ϭ 6). However, subsequently replacing bath Li ϩ with Na ϩ initiated a substantial pH i recovery (de; mean dpH i /dt ϭ 45 Ϯ 6 ϫ 10 Ϫ5 pH units/s, n ϭ 6). The removal of bath Na ϩ did not elicit a substantial acidification (which would have been consistent with a reversal in the direction of transport). Instead when Na ϩ was removed and Li ϩ restored, the  Ϫ solution (segment ab). pH i recovery from the acid load (bc) was observed in oocytes expressing NCBE. The application of CO 2 /HCO 3 Ϫ to NCBE-expressing oocytes did not elicit and rapid changes in V m (arrow in lower panel).

FIGURE 4. HCO 3
؊ dependence of pH i recovery mediated by NCBE. Shown are representative traces from a series of experiments in which we simultaneously recorded pH i and V m in NCBE-expressing oocytes that we acidified either in our CO 2 /HCO 3 Ϫ solution (segment ab) or in our butyrate solution (segment de). pH i recovery from the acid load was only observed when cells were acidified in the presence of extracellular HCO 3 Ϫ (bc versus ef). The unusually large depolarization in segment aЈcЈ of this experiment was typical for this particular set of experiments but not representative of NCBE expressing oocytes as a whole (see Fig. 3 and Figs. 5-8) and likely reflects an unusually tight membrane (as evidenced by the relatively negative resting V m ) in the presence of the usual, small, endogenous acidstimulated current.
The stability of pH i in segment ef in Fig. 5, upper panel, does not rule out the possibility that NCBE can operate in reverse inasmuch as Li ϩ might behave as a blocker. Therefore, we performed a separate series of experiments, typified by those presented in Fig. 6, in which we acid-loaded with CO 2 /HCO 3 Ϫ (ab), monitored the pH i recovery (bc), and then replaced bath Na ϩ with NMDG ϩ (cd). We found that the mean segment cd rate of acidification in NCBE-expressing oocytes was Ϫ7 Ϯ 1 ϫ 10 Ϫ5 pH units/s (n ϭ 46) compared with Ϫ4 Ϯ 1 ϫ 10 Ϫ5 pH units/s (n ϭ 24) for H 2 O-injected oocytes. Although the mean acidification rate was significantly greater in NCBE-expressing than in H 2 O-injected cells (p Ͻ 0.01, unpaired one-tailed t test assuming equal variances), the difference was not substantial. Other Na ϩ -coupled HCO 3 Ϫ transporters are also noted to reverse poorly (16,26).
Where indicated, we performed some of the subsequent experiments in this study on oocytes expressing NCBE-EGFP to confirm NCBE expression in individual oocytes prior to, or following, the electrophysiological experiment (27). To examine whether the addition of a C-terminal EGFP tag interferes with the HCO 3 Ϫ transport function of NCBE, we compared the rate of pH i recovery of oocytes expressing NCBE with that of oocytes expressing NCBE-EGFP in our CO 2 /HCO 3 Ϫ solution. In side-by-side experiments, we found that the untagged protein mediates oocyte pH i recovery (dpH i /dt ϭ 10 Ϯ 1 ϫ 10 Ϫ5 pH units/s, n ϭ 6; not shown) at a rate that is indistinguishable from that of NCBE-EGFP (not shown; dpH i /dt ϭ 11 Ϯ 2 ϫ 10 Ϫ5 pH units/s, n ϭ 6; p ϭ 0.72, unpaired two-tailed t test assuming equal variances).
To investigate whether the HCO 3 Ϫ transport mediated by NCBE in oocytes required extracellular Cl Ϫ , we exposed NCBE-EGFP-expressing oocytes to two serial pulses of CO 2 / HCO 3 Ϫ , either the first or second of which, in alternating order, lacked extracellular Cl Ϫ (68 mM sodium gluconate replaced 68 mM NaCl). We found that the rate of pH i recovery from the CO 2 -induced acid load is not changed (p ϭ 0.89, paired two- FIGURE 5. Na ؉ dependence of NCBE-mediated HCO 3 ؊ transport (Part I). Shown is a representative trace from a series of experiments in which we simultaneously recorded pH i and V m in NCBE-expressing oocytes that we acidified in a CO 2 /HCO 3 Ϫ solution (segment ab) in which NMDG ϩ fully replaced Na ϩ . pH i recovery from the acid load did not occur in this solution (bc) or in a Li ϩ -containing solution (cd). pH i recovery was only permitted when Na ϩ was restored to the bath (de). pH i recovery ceased again upon subsequent removal of bath Na ϩ (ef). The application of CO 2 /HCO 3 Ϫ to NCBEexpressing oocytes did not elicit and rapid changes in V m (arrow in lower panel). FIGURE 6. Na ؉ dependence of NCBE-mediated HCO 3 ؊ transport (Part II). Shown are representative traces from a series of experiments in which we simultaneously recorded pH i and V m in NCBE-expressing oocytes that we acidified in CO 2 /HCO 3 Ϫ -containing solutions (segments ab). A, NCBE-expressing oocyte. B, H 2 O-injected oocyte. In the presence of Na ϩ , pH i recovery from the acid load occurred in oocytes expressing NCBE (segment bc in A; 12 Ϯ 1 ϫ 10 Ϫ5 pH units/s, n ϭ 46) but not H 2 O-injected oocytes (segment bc in B; 4 Ϯ 1 ϫ 10 Ϫ5 pH units/s, n ϭ 24). In both panels, removal of bath Na ϩ caused pH i to fall slowly (cd) as summarized in the text.

FIGURE 7. Extracellular Cl ؊ independence of NCBE-mediated HCO 3
؊ transport. A representative trace from a series of experiments in which we simultaneously recorded pH i and V m in NCBE-expressing oocytes that we twice acidified in CO 2 /HCO 3 Ϫ solutions, once in the presence and once in the absence of bath Cl Ϫ . In half of the experiments and in the trace shown here, the oocyte was acidified initially in a Cl Ϫ -containing CO 2 /HCO 3 Ϫ solution (segment ab) and subsequently in a Cl Ϫ -free CO 2 /HCO 3 Ϫ solution (segment aЈbЈ). In the other half of the experiments, we reversed the order of the two CO 2 / HCO 3 Ϫ pulses (not shown); oocytes were acidified initially in a Cl Ϫ -free CO 2 / HCO 3 Ϫ solution and subsequently in a Cl Ϫ -containing CO 2 /HCO 3 Ϫ solution. The mean pH i recovery rate from the acid load was identical in both solutions (bc versus bЈcЈ). tailed t test) whether extracellular Cl Ϫ is present (Fig. 7, upper panel, bc; mean dpH i /dt ϭ 9 Ϯ 2 ϫ 10 Ϫ5 pH units/s, n ϭ 6; rate measured at average pH i ϭ 6.77 Ϯ 0.08) or absent (Fig. 7, upper panel, bЈcЈ; mean dpH i /dt ϭ 10 Ϯ 1 ϫ 10 Ϫ5 pH units/s, n ϭ 6; rate measured at average pH i ϭ 6.77 Ϯ 0.09).
Like NBCn1 and Like NDCBE, NCBE Is Blocked by DIDS-We serially exposed NCBE-expressing (or H 2 O-injected) oocytes to our CO 2 /HCO 3 Ϫ solution followed by the CO 2 / HCO 3 Ϫ solution containing 200 M DIDS. As we have shown, pH i in NCBE-expressing oocytes recovered from the CO 2 -induced acid load (Fig. 8A, upper panel, bc). The application of DIDS blocked this recovery (cd; mean dpH i /dt ϭ Ϫ8 Ϯ 1 ϫ 10 Ϫ5 pH units/s, n ϭ 14). Again as expected, pH i in H 2 O-injected oocytes failed to recover from the CO 2 -induced acidification (Fig. 8B, upper panel, bc). However, the subsequent application of DIDS caused pH i to fall (Fig. 8B, upper panel, cd; mean dpH i /dt ϭ Ϫ8 Ϯ 2 ϫ 10 Ϫ5 pH units/s, n ϭ 7). Note that the segment cd acidification rates were indistinguishable in NCBE-expressing versus H 2 O-injected cells (p ϭ 0.53, unpaired two-tailed t test assuming equal variances).
Like NBCn1 but Unlike NDCBE, NCBE Does Not Mediate a Net Efflux of Cl Ϫ -We expressed NCBE-EGFP in Xenopus oocytes. As positive controls, we used oocytes expressing AE1, human NDCBE-EGFP, and squid NDCBE. As negative controls, we used oocytes expressing NBCn1-EGFP as well as H 2 Oinjected oocytes. Using sharp Cl Ϫ -selective microelectrodes, we found that the mean intracellular [Cl Ϫ ] of all six populations of oocytes, measured with the oocytes exposed to the ND96 solution, is ϳ40 mM (Table 1, column 2). This value, which is similar to that reported by others for oocytes (32,33), is considerably higher than the equilibrium value of 23 mM (computed using the mean measured V m of Ϫ37 mV) 8 and indicates that oocytes have one or more active Cl Ϫ uptake mechanisms (e.g. Na/K/2Cl cotransport). For all oocyte populations, [Cl Ϫ ] i rapidly reached a new steady-state value after we exposed the cells to a HEPES-buffered 10 mM Cl Ϫ solution (  Fig. 9 shows a series of representative experiments in which we used a blunt tipped Cl Ϫ -selective microelectrode to monitor the [Cl Ϫ ] S in each of the six populations of oocytes. Each experiment began with the oocyte exposed to the ND96 solution. Because Cl Ϫ -sensitive electrodes are sensitive to the log 10 [Cl Ϫ ], we reduced the bulk extracellular Cl Ϫ concentration ([Cl Ϫ ] Bulk ) to 10 mM to enhance the sensitivity of the electrodes to changes in [Cl Ϫ ] S . This reduction in [Cl Ϫ ] Bulk presumably enhanced the net Cl Ϫ efflux from the oocytes. Indeed we found that the mean [Cl Ϫ ] S of oocytes exposed to a HEPES-buffered 10 mM Cl Ϫ solution was ϳ20 mM (see Table 1, column 5); the ϳ10 mM difference between [Cl Ϫ ] Bulk and [Cl Ϫ ] S is a semiquantitative index of the net Cl Ϫ efflux analogous to the outward [Cl Ϫ ] gradient that others have measured using self-referencing ion-selective electrodes (34).
We next replaced the HEPES-buffered 10 mM Cl Ϫ solution with a solution of identical [Cl Ϫ ] Bulk but now buffered with CO 2 /HCO 3 Ϫ . If the oocyte is expressing a membrane protein that exchanges extracellular HCO 3 Ϫ for intracellular Cl Ϫ , then this switch to CO 2 /HCO 3 Ϫ ought to enhance the net efflux of Cl Ϫ , which we would detect as a transient increase in [Cl Ϫ ] S . For the positive controls, AE1, human NDCBE-EGFP, and squid NDCBE, applying CO 2 /HCO 3 Ϫ indeed caused a transient increase in [Cl Ϫ ] S followed by a slow decrease as represented in Fig. 9, A-C, and summarized in Fig. 10A. For the negative controls, NBCn1-EGFP and H 2 O, applying CO 2 /HCO 3 Ϫ caused a decrease in [Cl Ϫ ] S as represented in Fig. 9, D and E, and summarized in Fig. 10A. This decrease implies that that CO 2 / HCO 3 Ϫ (perhaps by lowering pH i ) reduces the net efflux of Cl Ϫ . Finally for NCBE-EGFP, applying CO 2 /HCO 3 Ϫ caused a decrease in [Cl Ϫ ] S (Figs. 9F and 10A) as we observed for the  ͓Cl Ϫ ͔ i in column 2 is the steady-state value of oocytes bathed in ND96 solution (͓Cl Ϫ ͔ Bulk ϭ 101 mM). We continuously monitored ͓Cl Ϫ ͔ i while switching ͓Cl Ϫ ͔ Bulk from 101 to 10 mM. The ⌬͓Cl Ϫ ͔ i value in column 3 reflects the change in ͓Cl Ϫ ͔ i 5 min after the switch to the HEPES-buffered 10 mM Cl Ϫ solution (i.e. corresponding to the time just before the switch to CO 2 /HCO 3 Ϫ in Fig. 9). In the NCBE-expressing oocytes, we followed the 5-min exposure to 10 mM Cl Ϫ (summarized in column 3) with a 15-min exposure to 0 mM Cl Ϫ (⌬͓Cl͔ i ϭ Ϫ1 Ϯ 1 mM, n ϭ 4; not shown). Finally the ͓Cl Ϫ ͔ S data in column 4, obtained on the oocytes shown in Figs. 9 and 10, represent the surface ͓Cl Ϫ ͔ just before the switch from the HEPES-buffered 10 mM Cl Ϫ solution to the CO 2 /HCO 3 Ϫ -buffered 10 mM Cl Ϫ solution. hNDCBE, human NDCBE; sqNDCBE, squid NDCBE.
negative controls. Ϫ uptake to the net efflux of Cl Ϫ in the presence of 10 mM extracellular Cl Ϫ .
As a control, we verified that pretreating AE1 oocytes with DIDS for 1 h, a treatment known to irreversibly block about half of 36 Cl efflux through mouse AE1 (35), reduces both the size of the [Cl Ϫ ] S spike and the rate of pH i recovery (Fig. 10, black bars). Thus, our approach for monitoring [Cl Ϫ ] S appears to be appropriate for assessing net Cl Ϫ efflux coupled to HCO 3 Ϫ influx.
Finally using an experimental approach similar to that presented in Fig. 5, when we first exposed NDCBE-expressing oocytes to CO 2 /HCO 3 Ϫ in the continued absence of Na ϩ , we observed neither a [Cl Ϫ ] S spike nor a pH i recovery (Fig. 10, fourth bar). Thus, our approach for monitoring [Cl Ϫ ] S appears to be appropriate for assessing net Cl Ϫ efflux coupled to Na ϩ and HCO 3 Ϫ influx. Unlike NDCBE, NCBE Mediates a 36 Cl Efflux That Is Na ϩ -independent-Others have reported that, in oocytes, NCBE mediates the efflux of 36 Cl when Na ϩ and HCO 3 Ϫ are present in the extracellular fluid (20).
Comparing the 36 Cl efflux from oocytes bathed in ND96 with those bathed in CO 2 /HCO 3 Ϫ solution, we found that, in the presence of bath Na ϩ , CO 2 /HCO 3 Ϫ stimulates 36 Cl efflux from oocytes expressing NDCBE-EGFP (Fig. 11A, bar F versus bar B; p Ͻ 0.0001, one-way ANOVA with Student-Newman-Keuls posthoc analysis as are all statistics quoted in this section) or NCBE-EGFP (Fig. 11B, bar F versus bar B; p Ͻ 0.0001) but not from oocytes that were injected with H 2 O (Fig. 11C, bar F versus bar B; p ϭ 0.995). These results are in agreement with previous reports (20,26).
We found that the CO 2 /HCO 3 Ϫ -stimulated 36 Cl efflux from NDCBE-expressing oocytes requires bath Na ϩ (Fig. 11A, bar G   11A, bar J versus bar F; p Ͻ 0.0001) also in agreement with a similar assay for NDCBE activity (26). The CO 2 /HCO 3 Ϫ -stimulated 36 Cl efflux from NDCBE-expressing oocytes also required bath Cl Ϫ (Fig. 11A, bar H versus bar F; p Ͻ 0.0001). These data are the first direct demonstration that NDCBE engages in Cl-Cl exchange under near physiological conditions.
As is the case with NDCBE-expressing oocytes, we found that the CO 2 /HCO 3 Ϫ -stimulated 36 Cl efflux from NCBE-expressing oocytes is blocked by 200 M DIDS (Fig. 11B, bar J versus bar F; p Ͻ 0.0001) and requires bath Cl Ϫ (Fig. 11B, bar H versus bar F; p ϭ 0.0004). However, the CO 2 /HCO 3 Ϫ -stimulated 36 Cl efflux did not require bath Na ϩ in NCBE-expressing oocytes (Fig. 11B, bar G versus bar C (p Ͻ 0.0001) and bar G versus bar F (p ϭ 0.247)) in contrast to NDCBE-expressing oocytes. In the absence of bath Na ϩ , the additional removal of bath Cl Ϫ negated the CO 2 /HCO 3 Ϫ -stimulated 36 Cl efflux (Fig.  11B, bar I versus bar G; p Ͻ 0.0001) and demonstrates that, at least in the absence of bath Na ϩ , NCBE engages in CO 2 /HCO 3 Ϫstimulated Cl-Cl exchange. A curious observation in NCBEexpressing oocytes is that, in the absence of bath Cl Ϫ , adding bath Na ϩ now stimulated a CO 2 /HCO 3 Ϫ -dependent 36 Cl efflux (Fig. 11B, bar I versus bar H; p Ͻ 0.03). This flux, which is only about one-third as large as the 36 Cl flux that we observed in the presence of bath Cl Ϫ under near physiological conditions (Fig. 11B, bar F), is consistent with a small amount of Na ϩ -driven Cl-HCO 3 exchange activity but only in the nonphysiological state in which extracellular Cl Ϫ is totally absent.

DISCUSSION
The purpose of this study was to clarify the distribution and molecular mechanism of the human SLC4A10 gene product toward a better understanding of its physiological purpose. Prior to the present report, nearly all of the work has concentrated on rodent orthologs of the transporter.

NCBE mRNA Is Widely Distributed in Human Brain
Our northern blots of human RNA demonstrated that, among the sources tested, NCBE is predominantly expressed in the brain ( Fig.  2A). These data accord with rat northern blot data that show highest expression in the brain but lower expression in ileum, kidney, pituitary tissue (20), and spleen (23). Those authors and others (36) detected, by northern blot and in situ hybridization, a broad distribution of NCBE transcripts throughout the brains of adult rats (20,23) and embryonic mice (36). Our Northern blotting study showed a similarly widespread expression of NCBE transcripts throughout the adult human brain (Fig. 2B). To our knowledge, our study is the first to identify the presence, in any species, of NCBE transcripts in the amygdala, caudate nucleus, putamen, and substantia nigra. Furthermore our study is the first to identify the corpus callosum and spinal cord as being regions of particularly weak NCBE expression. On the other hand, Hubner et al. have detected NCBE transcripts, by in situ hybridization, in the spinal cords of embryonic mice (36). Although the reason for this disparity is unresolved, we propose differences in species, developmental stage, and/or RNA preparation to be relevant considerations. The mRNA of another electroneutral Na ϩ -coupled HCO 3 Ϫ transporter, NDCBE, is also notably absent from human spinal cord (26).

Multiple Splice Variants of NCBE Are Detected in Human cDNA Libraries
A discussion of the presence and distribution of multiple human NCBE splice variants is provided in the supplemental material.

NCBE Mediates Electroneutral, DIDS-sensitive Na ؉ -coupled HCO 3 ؊ Transport
Previous reports of NCBE describe the protein as being a Na ϩ -driven Cl-HCO 3 exchanger. This assignment was based In each panel, the gray"A" bar represents the normalized value for the first collection period in ND96 solution. The white bars represent data from a second collection period (bars B-H), normalized to the value in the corresponding first collection period. The horizontal dashed line represents the average estimated normalized background (i.e."zero" efflux) value for the second period. We calculated that, at the beginning of the first efflux period (i.e. before gray bar), NDCBE-expressing oocytes contained 553 Ϯ 30 cpm (n ϭ 66), NCBE-expressing oocytes contained 774 Ϯ 30 cpm (n ϭ 34), and H 2 O-injected oocytes contained 325 Ϯ 48 cpm (n ϭ 27). For a given panel, these initial cpm values did not differ significantly among oocytes subsequently assigned to groups B-J (one-way ANOVA, p Ͼ 0.10 in all cases). The fractional loss of 36 Cl during the first 30-min efflux period in ND96 (i.e. during gray bar) was 13 Ϯ 1% (n ϭ 66) for NDCBE-expressing oocytes, 9 Ϯ 1% (n ϭ 79) for NCBE-expressing oocytes, and 33 Ϯ 3% (n ϭ 38) for H 2 O-injected oocytes. For a given panel, these first period fractional losses did not differ significantly among oocytes subsequently assigned to groups B-J (one-way ANOVA, p Ͼ 0.08 in all cases). on the results of two types of functional assays: (a) 22 Na and 36 Cl flux data from oocytes expressing NCBE (20) and (b) pH i measurements on BCECF-loaded mammalian cells that were transfected with NCBE (20,23). Our results agree with those of others in three major areas as described in the following paragraphs.
NCBE Is a HCO 3 Ϫ Transporter-NCBE mediates a pH i recovery from an acid load only in the presence of HCO 3 Ϫ regardless of whether one studies NCBE heterologously expressed in HEK293 cells (20), 3T3 cells (23), or Xenopus oocytes (present study, Figs. 3 and Fig. 4).
NCBE Cotransports Na ϩ and HCO 3 Ϫ -The original report of NCBE demonstrated that oocytes expressing NCBE accumulate 22 Na only in the presence of extracellular HCO 3 Ϫ (20). The same authors demonstrated that NCBE-expressing HEK293 cells, in the presence of HCO 3 Ϫ , are incapable of recovering from an acid load until Na ϩ is restored to the bathing medium (20), a result replicated by others in NCBE-expressing 3T3 cells (23) and in the present study in NCBE-expressing oocytes (Fig. 5).
DIDS Blocks NCBE-Others found that NCBE expressed in oocytes mediates a 22 Na influx that is blocked by 300 M DIDS (20). Others also found that, whether expressed in HEK293 cells or 3T3 cells, NCBE mediates a pH i recovery that is prevented by the addition of 300 M DIDS to the bathing medium (20,23). In the present study we demonstrated that 200 M DIDS is sufficient to block the pH i recovery mediated by NCBE in oocytes (Fig. 8). The lysine-containing motif "EKLFE," located at the extracellular end of the fifth transmembrane span of human, mouse, and rat NCBE, is predicted to play an important role in this DIDS blockade as has been demonstrated for lysine residues at this position in AE1 (37) and NBCe1 (38).
Another important aspect of NCBE molecular physiology, although never disputed, has actually never been demonstrated, namely that NCBE is an electroneutral transporter. The present study is the first and only to provide direct evidence that HCO 3 Ϫ transport mediated by NCBE is electroneutral. If NCBE was electrogenic and working with a 2:1 HCO 3 Ϫ :Na ϩ stoichiometry, for an oocyte with a membrane resistance of 0.25 megaohm we calculated that NCBE should elicit a hyperpolarization of Ϫ26 mV. 9 However, the switch from a CO 2 / HCO 3 Ϫ -free to a CO 2 /HCO 3 Ϫ -containing solution (Fig. 3A) did not elicit any rapid changes in the V m of oocytes expressing NCBE.

Influx
As it is not practical to deplete oocytes of intracellular Cl Ϫ , our laboratory has previously assayed the presence versus the absence of Na ϩ -driven Cl-HCO 3 exchange activity by reversing the direction of Na ϩ -driven Cl-HCO 3 exchange (by removing extracellular Na ϩ ) and then asking whether the subsequent removal of bath Cl Ϫ blocks the Na ϩ -dependent export of HCO 3 Ϫ from the cell (13,26,28,39). In other words, does net HCO 3 Ϫ efflux require extracellular Cl Ϫ ? However, because we were unable to substantially reverse the direction of NCBE transport by removing bath Na ϩ (Fig. 6), this approach is impossible for NCBE-expressing oocytes.
In the present study, we used a blunt tipped Cl Ϫ -selective electrode at the oocyte surface to monitor changes in [Cl Ϫ ] S , produced by changes in net Cl Ϫ efflux, as we exposed the cell to a CO 2 /HCO 3 Ϫ -containing solution. We validated our experimental approach in four ways.
1) We demonstrated that the exposure to CO 2 /HCO 3 Ϫ produces a transient increase in [Cl Ϫ ] S for oocytes expressing three known Cl Ϫ -coupled HCO 3 Ϫ transporters: human AE1, human NDCBE-EGFP, and squid NDCBE (Figs. 9 and 10). Our positive control studies provide the first demonstration that both human and squid NDCBE couple net HCO 3 Ϫ influx to a net Cl Ϫ efflux (Figs. 9 and 10), that is, in the physiological direction of transport. The previous work showed that human NDCBE, when operating in the forward direction, mediates a CO 2 / HCO 3 Ϫ -stimulated 36 Cl efflux (26), and human and squid NDCBE, when operating in reverse, require extracellular Cl Ϫ (26,28). 2) We demonstrated the lack of a transient [Cl Ϫ ] S increase for oocytes expressing rat NBCn1-EGFP or oocytes that had been injected with H 2 O (Figs. 9 and 10). 3) We prevented the increase in [Cl Ϫ ] S for oocytes expressing AE1 by preincubation with DIDS (Fig. 10). 4) We prevented the transient increase of [Cl Ϫ ] S in oocytes expressing human NDCBE-EGFP by switching the cell to CO 2 /HCO 3 Ϫ in the absence of bath Na ϩ (Fig. 10). Our central observation was that exposure to CO 2 /HCO 3 Ϫ does not produce a transient increase in [Cl Ϫ ] S for oocytes expressing NCBE-EGFP (Figs. 9 and 10), thereby providing the first demonstration that NCBE does not couple net HCO 3 Ϫ influx to a net Cl Ϫ efflux.

NCBE-mediated Net HCO 3 ؊ Influx Does Not Require Extracellular Cl ؊
The only area in which our conclusions conflict with those of others concerns whether NCBE, like NDCBE, couples a net efflux of Cl Ϫ to the net influx of HCO 3 Ϫ . Others have presented two major arguments in support of the hypothesis that NCBE is a Na ϩ -driven Cl-HCO 3 exchanger: 1) NCBE transport is Cl Ϫdependent and 2) NCBE transports 36 Cl. We will address the first argument in this section and the second in the following section.
The first argument in support of the hypothesis that NCBE is a Na ϩ -driven Cl-HCO 3 exchanger is that NCBE transport is Cl Ϫ -dependent. In their original report, Wang et al. (20) found that removing extracellular Cl Ϫ reduces the NCBE-dependent 22 Na ϩ influx into oocytes, during a 15-min incubation period, by ϳ50% (their Fig. 2a, bar "C" versus bar "A"). In their supplemental data they further report that the 22 , n ϭ 4). Thus, the aforementioned 22 Na data do not prove that NCBE engages in physiological Na ϩ -driven Cl-HCO 3 exchange activity.
In further studies, Wang et al. (20) heterologously expressed NCBE in HEK293 cells and demonstrated that removing bath Cl Ϫ reduces the NCBE-mediated pH i recovery from an intracellular acid load by ϳ70%. Again these results are inconsistent with a direct effect (i.e. lowering [Cl Ϫ ] o per se) on Na ϩ -driven Cl-HCO 3 exchange activity. Wang et al. (20) suggest that the extracellular Cl Ϫ removal led to Cl Ϫ depletion of the cells. However, we note that Cl Ϫ is very difficult to wash out of glomerular mesangial cells (40) and neurons (3). Moreover Cl Ϫ reloads into neurons with great rapidity (3). Although Wang et al. (20) preincubated their cells for 1 h in a Cl Ϫ -free medium, they did not verify that [Cl Ϫ ] i actually fell. Moreover when they prepulsed their cells with NH 3 /NH 4 ϩ to impose the acid load (41), they reintroduced 40 mM Cl Ϫ (as NH 4 Cl) and thereby probably reloaded their cells with Cl Ϫ just before monitoring the pH i recovery mediated by NCBE. Thus, the aforementioned pH i recovery data do not prove that NCBE engages in physiological Na ϩ -driven Cl-HCO 3 exchange activity.
Others, working with 3T3 cells, also report that external Cl Ϫ removal reduces the NCBE-mediated pH i recovery from an acid load (23). They suggested that Cl Ϫ either directly or indirectly stimulates NCBE. In the present study, albeit in oocytes, we found that removing bath Cl Ϫ does not significantly affect the net HCO 3 Ϫ influx (all of which is Na ϩ -dependent) mediated by NCBE (Fig. 7).

NCBE Mediates DIDS-sensitive Cl ؊ Self-exchange
The second argument in support of the hypothesis that NCBE is a Na ϩ -driven Cl-HCO 3 exchanger is that NCBE transports 36 Cl. In their original report, Wang et al. (20) demonstrate that, in NCBE-expressing oocytes preloaded with 36 Cl, the isotope exits most rapidly in the simultaneous presence of extracellular Cl Ϫ , Na ϩ , and HCO 3 Ϫ (their Fig. 2c, bar "A"). Applying DIDS (their Fig. 2c, bar "A ϩ DIDS") or removing bath Na ϩ (their bar "B") each inhibits 36 Cl efflux by ϳ45% (they performed no 36 Cl efflux experiments on H 2 O-injected oocytes). These data would be consistent with Na ϩ -driven Cl-HCO 3 exchanger activity save for two observations. First, replacing CO 2 /HCO 3 Ϫ with butyrate (their Fig. 2c, bar "E") reduces 36 Cl efflux by ϳ80% (versus only ϳ45% for either DIDS addition or Na ϩ removal). Second, removing bath Cl Ϫ inhibits 36 Cl efflux by ϳ45% (their Fig. 2c, bar "C").
Wang et al. (20) interpret the last observation as follows. As discussed above, their supplemental data show that reducing [Cl Ϫ ] o inhibits the 22 Na influx mediated by NCBE. Thus, they suggest that the external Cl Ϫ dependence of 36 Cl efflux (their Fig. 2c, bar "C") reflects the dependence of net transport by NCBE on external Cl Ϫ . However, we found that reducing [Cl Ϫ ] o to 0 mM (see previous section and Fig. 7) has no discernable effect on net HCO 3 Ϫ transport, ruling out a significant modulatory role for external Cl Ϫ . Therefore, we suggest that the trans-side Cl Ϫ dependence of 36 Cl efflux in the experiments of Wang et al. (20) indicates that about half of the total 36 Cl efflux represents a Cl-Cl exchange. In light of our further obser-vation that NCBE does not couple net HCO 3 Ϫ influx to net Cl Ϫ efflux (Figs. 9 and 10), the most straightforward explanation for the 36 Cl efflux data of Wang et al. (20) is that although NCBE does not mediate net Cl Ϫ transport, it mediates a DIDS-sensitive Cl-Cl exchange. Indeed Wang et al. (20) note that 36 Cl movement is bidirectional inasmuch as NCBE-expressing oocytes also mediate Cl Ϫ influx when bathed in a Na ϩ -and HCO 3 Ϫ -containing solution (their Fig. 2b, bar "A"). Using a similar 36 Cl efflux assay, which we validated using oocytes expressing NDCBE as a positive control and H 2 O-injected oocytes as a negative control, we demonstrated that enhanced 36 Cl efflux from NCBE-expressing oocytes occurs in both our Na ϩ -containing (Fig. 11B, bar F) and our Na ϩ -free (Fig. 11B, bar G) CO 2 /HCO 3 Ϫ solutions. Note that the Na ϩ -free condition is one under which NCBE is inactive with regard to net HCO 3 Ϫ transport (Figs. 5 and 6). Moreover enhanced 36 Cl efflux from NCBE-expressing oocytes was negated in our Na ϩand Cl Ϫ -free solution (Fig. 11B, bar I). Thus, the trans-side dependence of 36 Cl efflux on bath Cl Ϫ allows us to conclude that, in the absence of bath Na ϩ , all CO 2 /HCO 3 Ϫ -stimulated 36 Cl efflux represents Cl-Cl exchange. Furthermore because our [Cl Ϫ ] S data in Figs. 9 and 10 show that NCBE does not mediate net Cl Ϫ efflux in the presence of bath Na ϩ and Cl Ϫ , we can conclude that, in the presence of bath Na ϩ and Cl Ϫ , all CO 2 /HCO 3 Ϫ -stimulated 36 Cl efflux represents Cl-Cl exchange as well.
The surprising observation was that, in the absence of bath Cl Ϫ , bath Na ϩ supports a small but significant CO 2 /HCO 3 Ϫstimulated 36 Cl efflux (Fig. 11B, bar I versus bar H). Note that this flux, which is consistent with Na ϩ -driven Cl-HCO 3 exchange activity, occurs only under highly nonphysiological conditions: the complete absence of extracellular Cl Ϫ .
From the CO 2 -induced pH i decreases in the present study, we computed a mean intrinsic intracellular buffering power (42) of 18 Ϯ 2 mM/pH unit (n ϭ 7 NCBE-expressing oocytes). Given a mean pH i of 6.8 during a pH i recovery, the computed open system CO 2 /HCO 3 Ϫ buffering power (42) is ϳ15 mM/pH unit for a total intracellular buffering power of ϳ33 mM/pH unit. Because the average pH i recovery rate for NCBE-expressing oocytes in CO 2 /HCO 3 Ϫ was 14 ϫ 10 Ϫ5 pH units/s, we estimated that the rate of net HCO 3 Ϫ influx into an NCBEexpressing oocyte is 4.62 M/s. Given that the [Cl] i of NCBEexpressing oocytes is 40 mM and that 21% of oocyte Cl Ϫ exits during a 30-min incubation in Cl Ϫ -free CO 2 /HCO 3 Ϫ solution versus 10% during a 30-min incubation in Cl Ϫ -free ND96, we estimated that the Cl Ϫ efflux is 2.44 M/s. The ratio of the HCO 3 Ϫ entry to Cl Ϫ exit is ϳ1.9 versus the predicted 2.0 for Na ϩ -driven Cl-HCO 3 exchange. Thus, under the highly nonphysiological conditions in which Cl Ϫ is absent from the bath, we can account for all of the Cl Ϫ efflux by Na ϩ -driven Cl-HCO 3 exchange.
How do we interpret bar F versus bar H in Fig. 11B? Under the conditions of bar F ([Cl Ϫ ] o ϭ 101 mM) all of the CO 2 / HCO 3 Ϫ -stimulated 36 Cl efflux is Cl-Cl exchange. Under the conditions of bar H ([Cl Ϫ ] o ϭ 0), all the Cl-Cl exchange halts and is replaced by a smaller degree of Na ϩ -driven Cl-HCO 3 exchange. We emphasize that removing bath Cl Ϫ (i.e. the transition from the condition of bar F to bar H in Fig. 11B) does not reduce net HCO 3 Ϫ uptake (Fig. 7). Regarding NDCBE, in their 2001 study, Grichtchenko et al. (26) concluded that more than 80% of the 36 Cl efflux mediated by NDCBE represents Cl-Cl exchange. Compared with the oocytes and/or conditions used in that previous study, we found that the present HCO 3 Ϫ efflux is greater, whereas the present 36 Cl flux is lesser so that the present computed HCO 3 Ϫ : Cl Ϫ flux ratio is almost exactly the 2:1 predicted for Na ϩ -driven Cl-HCO 3 exchange. In other words, for the oocytes and/or conditions in the present study, we can account for all of the 36 Cl efflux of NDCBE by Na ϩ -driven Cl-HCO 3 exchange under physiological conditions (i.e. Fig. 11A, bar F). Our surface [Cl Ϫ ] data showed that with bath [Cl Ϫ ] as low as 10 mM NDCBE still mediates a net Cl Ϫ efflux (Figs. 9 and 10). However, our 36 Cl data showed that the complete removal of bath Cl Ϫ reduces the 36 Cl efflux to background levels (Fig. 11A, bar H versus bar F). The most straightforward explanations for this observation are that 1) Na ϩ -driven Cl-HCO 3 exchange by NDCBE requires that Cl Ϫ bind to an extracellular Cl Ϫ modifier site (43) or 2) removal of bath Cl Ϫ causes NDCBE to shift from Na ϩ -driven Cl-HCO 3 exchange to electroneutral Na/HCO 3 cotransport.
It is interesting to note that the ␥-aminobutyric acid transporter GAT-1, like NCBE, engages in futile Cl-Cl exchange. In the absence of Cl Ϫ , GAT-1 cotransports 2 Na ϩ /1 ␥-aminobutyric acid, but in the presence of Cl Ϫ , the transporter imports 1 Na ϩ /1 Cl Ϫ /1 ␥-aminobutyric acid in exchange for 1 Cl Ϫ (44). The involvement of Cl Ϫ in the transport cycle increases the affinity of GAT-1 for Na ϩ (44). It is possible that extracellular Cl Ϫ also raises the affinity of NCBE for one of its substrates (i.e. Na ϩ or HCO 3 Ϫ ). However, we note that removing bath Cl Ϫ did not reduce net HCO 3 Ϫ uptake by NCBE (Fig. 7) under the conditions of our assay with the transporter, without its natural posttranslational modifications or binding partners, heterologously expressed in Xenopus oocytes.

Conclusion
We conclude that NCBE, unlike NDCBE, does not under physiological conditions perform the Na ϩ -driven Cl-HCO 3 exchange activity for which it was named. Instead NCBE, like NBCn1, mediates an electroneutral Na/HCO 3 cotransport activity with an associated function. In the case of NBCn1, this associated function is a Na ϩ conductance (13); in the case of NCBE, this associated function is likely a futile Cl-Cl exchange. We therefore propose that NCBE be renamed NBCn2.

Implications
If our results in oocytes translate to mammalian cells, the reassignment of the ionic mechanism of SLC4A10 may have important implications for cells expressing the transporter. First, by not engaging in net Cl Ϫ transport, NBCn2 could operate without altering the equilibrium potential for Cl Ϫ , which could be important for stabilizing inhibitory postsynaptic currents. Second, by transporting only one HCO 3 Ϫ for each Na ϩ , NBCn2 would be less energy-efficient than NDCBE, the price paid for 1) not altering [Cl Ϫ ] i and yet remaining electroneutral and 2) being theoretically capable of taking up HCO 3 Ϫ against a relatively steep gradient.