Functional comparison of mouse slc26a6 anion exchanger with human SLC26A6 polypeptide variants: differences in anion selectivity, regulation, and electrogenicity.

The unusually low 78% amino acid identity between the orthologous human SLC26A6 and mouse slc26a6 polypeptides prompted systematic comparison of their anion transport functions in Xenopus oocytes. Multiple human SLC26A6 variant polypeptides were also functionally compared. Transport was studied as unidirectional fluxes of (36)Cl(-), [(14)C]oxalate, and [(35)S]sulfate; as net fluxes of HCO(3)(-) by fluorescence ratio measurement of intracellular pH; as current by two-electrode voltage clamp; and as net Cl(-) flux by fluorescence intensity measurement of relative changes in extracellular and intracellular [Cl(-)]. Four human SLC26A6 polypeptide variants each exhibited rates of bidirectional [(14)C]oxalate flux, Cl(-)/HCO(3)(-) exchange, and Cl(-)/OH(-) exchange nearly equivalent to those of mouse slc26a6. Cl(-)/HCO(3)(-) exchange by both orthologs was cAMP-sensitive, further enhanced by coexpressed wild type cystic fibrosis transmembrane regulator but inhibited by cystic fibrosis transmembrane regulator DeltaF508. However, the very low rates of (36)Cl(-) and [(35)S]sulfate transport by all active human SLC26A6 isoforms contrasted with the high rates of the mouse ortholog. Human and mouse orthologs also differed in patterns of acute regulation. Studies of human-mouse chimeras revealed cosegregation of the high (36)Cl(-) transport phenotype with the transmembrane domain of mouse slc26a6. Mouse slc26a6 and human SLC26A6 each mediated electroneutral Cl(-)/HCO(3)(-) and Cl(-)/OH(-) exchange. In contrast, whereas Cl(-)/oxalate exchange by mouse slc26a6 was electrogenic, that mediated by human SLC26A6 appeared electroneutral. The increased currents observed in oocytes expressing either mouse or human ortholog were pharmacologically distinct from the accompanying monovalent anion exchange activities. The human SLC26A6 polypeptide variants SLC26A6c and SLC26A6d were inactive as transporters of oxalate, sulfate, and chloride. Thus, the orthologous mouse and human SLC26A6 proteins differ in anion selectivity, transport mechanism, and acute regulation, but both mediate electroneutral Cl(-)/HCO(3)(-) exchange.

The unusually low 78% amino acid identity between the orthologous human SLC26A6 and mouse slc26a6 polypeptides prompted systematic comparison of their anion transport functions in Xenopus oocytes. Multiple human SLC26A6 variant polypeptides were also functionally compared. Transport was studied as unidirectional fluxes of 36 3 ؊ by fluorescence ratio measurement of intracellular pH; as current by two-electrode voltage clamp; and as net Cl ؊ flux by fluorescence intensity measurement of relative changes in extracellular and intracellular [Cl ؊ ]. Four human SLC26A6 polypeptide variants each exhibited rates of bidirectional [ 14 C]oxalate flux, Cl ؊ /HCO 3 ؊ exchange, and Cl ؊ /OH ؊ exchange nearly equivalent to those of mouse slc26a6. Cl ؊ /HCO 3 ؊ exchange by both orthologs was cAMP-sensitive, further enhanced by coexpressed wild type cystic fibrosis transmembrane regulator but inhibited by cystic fibrosis transmembrane regulator ⌬F508. However, the very low rates of 36 Cl ؊ and [ 35 S]sulfate transport by all active human SLC26A6 isoforms contrasted with the high rates of the mouse ortholog. Human and mouse orthologs also differed in patterns of acute regulation. Studies of human-mouse chimeras revealed cosegregation of the high 36 Cl ؊ transport phenotype with the transmembrane domain of mouse slc26a6. Mouse slc26a6 and human SLC26A6 each mediated electroneutral Cl ؊ /HCO 3 ؊ and Cl ؊ /OH ؊ exchange. In contrast, whereas Cl ؊ /oxalate exchange by mouse slc26a6 was electrogenic, that mediated by human SLC26A6 appeared electroneutral. The increased currents observed in oocytes expressing either mouse or human ortholog were pharmacologically distinct from the accompanying monovalent anion exchange activities. The human SLC26A6 polypeptide variants SLC26A6c and SLC26A6d were inactive as transporters of oxalate, sulfate, and chloride. Thus, the orthologous mouse and human SLC26A6 proteins differ in anion selectivity, transport mechanism, and acute regulation, but both mediate electroneutral Cl ؊ /HCO 3 ؊ exchange.
SLC26 polypeptides expressed in apical membranes of epithelial cells (among them SLC26A3 (34), SLC26A4 (19), and SLC26A6 (23)) have generated considerable interest by virtue of their co-localization with cystic fibrosis transmembrane reg-ulator (CFTR). 1 Indeed, cAMP-stimulated Cl Ϫ /HCO 3 Ϫ exchange associated with overexpression of any of these three apically localized SLC26 polypeptides is enhanced when co-expressed with wild type but not with mutant CFTR in HEK-293 cells (28) or (in the case of SLC26A3/DRA) in Xenopus oocytes (12). In addition, SLC26 polypeptides can interact with and potentiate activation of CFTR (28,35). These data, together with evidence that cAMP-dependent epithelial HCO 3 Ϫ secretion is impaired or absent in cystic fibrosis (CF) and in cystic fibrosis mouse models (28,36), have focused attention on SLC26 transporters as leading candidates to mediate the major apical HCO 3 Ϫ permeability pathways of HCO 3 Ϫsecreting epithelial cells.
FIG. 1. Schematic of the six variant human SLC26A6 polypeptides and the single mouse slc26a6 (CFEX) polypeptide studied in this paper. The human SLC26A6 variant polypeptides LϩQ and LϪQ (long N terminus with or without Gln 632 ), SϩQ and SϪQ (short N terminus with or without Gln 611 ), A6a (another designation (25) for SϩQ), A6c (25), and A6d (25) are shown. CFEX (23) is the short N-terminal variant of mouse slc26a6. (The codon corresponding to human SLC26A6 Q611 is absent from the mouse slc26a6 mRNA). ing slc26a6 were hyperpolarized by Cl Ϫ removal. Ko et al. concluded that these two polypeptides mediate Cl Ϫ /HCO 3 Ϫ exchange of opposite stoichiometries (28) and proposed that the full range of electrogenic and electroneutral HCO 3 Ϫ secretion observed in epithelia can be explained by varying proportional contributions of different SLC26 anion exchangers, with or without invocation of HCO 3 Ϫ -permeable secretory anion channels (28,36,37,63).
Unlike the high degree of amino acid identity between orthologous mouse and human SLC4 bicarbonate transporters, orthologous mouse and human SLC26 polypeptide sequences (with the notable exception of the cochlear outer hair cell mechanotransducer, SLC26A5/prestin) diverge to a much greater degree. Mouse slc26a6 and human SLC26A6 share only 78% aa identity. A single demonstration of unidirectional isotopic influx has been reported for the human SLC26A6 (25) 2 after an earlier report failed to detect evidence of function in Xenopus oocytes (38). Therefore, we examined systematically the functional properties of several human SLC26A6 polypeptide variants and compared them with the only functionally studied mouse slc26a6 polypeptide variant, also known as the chloride/ formate exchanger, CFEX (23).
We found that four active human SLC26A6 polypeptide variants exhibit similar functional properties in Xenopus oocytes. All differed from mouse slc26a6/CFEX most prominently in their very low rates of 36 Cl Ϫ transport in Cl Ϫ /Cl Ϫ exchange assays and in their very low rates of 35 SO 4 2Ϫ transport. However, these same polypeptides mediated bidirectional [ 14 C]oxalate fluxes as well as Cl Ϫ /HCO 3 Ϫ , Cl Ϫ /OH Ϫ , and oxalate/HCO 3 Ϫ exchange activities (measured as dpH i /dt) at rates approaching those of mouse slc26a6/CFEX. Wild-type CFTR, but not CFTR ⌬F508, stimulated Cl Ϫ /HCO 3 Ϫ and Cl Ϫ /OH Ϫ exchange by human SLC26A6 and enhanced its cAMP sensitivity. High rates of 36 Cl Ϫ transport cosegregated with the transmembrane domain of mouse slc26a6 in studies with chimeric proteins. The orthologous polypeptides of both species mediated electroneutral Cl Ϫ /HCO 3 Ϫ and Cl Ϫ /OH Ϫ exchanges, although expression was accompanied by oocyte current with pharmacological properties distinct from those of the corresponding anion exchange activities. In marked contrast, Cl Ϫ /oxalate exchange by mouse slc26a6/CFEX was clearly electrogenic, whereas that mediated by human SLC26A6 appeared electroneutral. Two unusual human variant SLC26A6 polypeptides previously reported to be functional (25) were nonfunctional as anion transporters. These data emphasize the complexity of SLC26A6-mediated anion exchange. They suggest an experimental path toward definition of amino acid residues important for Cl Ϫ binding and/or translocation, provide an unusual example of a single polypeptide able to exchange different substrates with either electroneutral or electrogenic mechanisms, and suggest reexamination of current hypotheses of epithelial HCO 3 Ϫ secretion. cDNA Clones-The cDNAs studied are presented in the schematic of Fig. 1. Human SLC26A6 LϪQ was obtained in pCMV-SPORT6 from the Mammalian Genome Collection (MGC_21068, IMAGE:4398446, BC017697) and was resequenced in entirety. In Fig. 1, Gln 632 (Q632) is numbered as in the long N-terminal (L) sequence. The corresponding short N-terminal (S) position is Gln 611 (Q611). The codon encoding this Gln can be retained (LϩQ, SϩQ) or excluded (LϪQ, SϪQ) from the 5Ј-end of its exon by elective utilization of an alternate 3Ј splice-acceptor splice site (3,38,39). The MGC_21068 human cDNA lacks the optional 13-aa insert (beginning at L aa 108 as in the human SLC26A6 variant, GenBank TM BAC56861) and contains in its C-terminal cytoplasmic domain the 30-aa insert (beginning at L aa 602-631) rather than the alternative 11-aa insert present in the human SLC26A6 variant, Gen-Bank TM BAC56861.
Human SLC26A6 LϪQ subcloned in the Xenopus ␤-globin oocyte expression vector pGEMHE was obtained from D. Mount. Human SLC26A6a (equivalent to SϩQ), SLC26A6c (which has the short N terminus and lacks Gln 611 ), and SLC26A6d (also with the short N terminus), originally studied as subclones of pcDNA3.1 (39, 25) (Fig. 1  Mammalian Genome Collection (MGC_25824, IMAGE:4165725, BC028856 from the FVB/N strain; it corresponds to human SϪQ, since the equivalent residue to human Gln 611 is absent from the mouse gene). This murine slc26a6 cDNA (also studied by Ko et al. (28)) 3 encodes the polymorphic amino acid residues Glu 2 and Arg 549 in place of the alternative Gly 2 (23) and Pro 549 (3). The genomic origins of the long (L) (38) and short (S) N termini (39) in humans and in mice were described by Xie et al. (3).
Mutagenesis-From the MGC_21068 human SLC26A6 LϪQ cDNA, we generated the LϩQ (38), SϩQ (39), and SϪQ (3) variants of human SLC26A (Fig. 1) in the same pCMV-SPORT6 vector, using a four-primer PCR method (12). Oligonucleotide primers were obtained from BioSynthesis (The Woodlands, TX); sequences are available upon request. Mutagenized PCR products were ligated into appropriately engineered host plasmids to reconstruct the desired open reading frame, and recombination was confirmed by diagnostic restriction digestion. Plasmid DNA sequences of the PCR-amplified regions and their ligation junctions were confirmed on both strands. The variable presence of Gln 611 / Gln 632 (Fig. 1) was noted previously (38,39).
Solutions-ND-96 (pH 7.40) consisted of 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 5 mM HEPES, 2.5 mM sodium pyruvate, and 100 mg/ml gentamicin. ND-96 flux medium and other flux media lacked sodium pyruvate and gentamicin. In Cl Ϫ -free solutions, 96 mM NaCl was replaced with 96 mM sodium isethionate. The Cl Ϫ salts of K ϩ , Ca 2ϩ , and Mg 2ϩ were substituted with the corresponding equimolar gluconate salts. Hepes-free CO 2 /HCO 3 Ϫ -buffered solutions of pH 7.4 were saturated with 5% CO 2 , 95% air at room temperature for ϳ1 h and differed from Cl Ϫ -free ND-96 in replacement of 24 mM sodium isethionate with 24 mM NaHCO 3 Ϫ . The pH of CO 2 /HCO 3 Ϫ -buffered solutions was verified prior to each experiment. The addition to flux media of the weak acid salt sodium butyrate (40 mM) or of NH 4 Cl (20 mM) was in equimolar substitution for NaCl.
cRNA Expression in Xenopus Oocytes-Capped cRNA was transcribed from linearized cDNA templates with SP6, T7, or T3 RNA polymerase (Ambion, Austin, TX). RNA integrity was confirmed by agarose gel electrophoresis in formaldehyde. Mature female Xenopus (NASCO, Madison, WI) were maintained and subjected to partial ovariectomy under Tricaine/hypothermia anesthesia as described (12), conforming to methods approved by the Institutional Animal Care and Use Committee of Beth Israel Deaconess Medical Center. Stage V-VI oocytes were manually defolliculated following incubation of ovarian fragments with 2 mg/ml collagenase A or collagenase B (Roche Applied Science) for 60 min in ND-96. Oocytes were injected on the same day or the following day with cRNA (10 ng in most experiments; as little as 50 pg when indicated) or with water in a volume of 50 nl. Injected oocytes were then maintained for 2-6 days at 19°C.
Immunoblot-Oocytes previously injected with water or with cRNAs 3 K. Ishibashi, personal communication. encoding the indicated human SLC26A6 isoforms or mouse slc26a6/ CFEX were solubilized by gentle homogenization in buffer containing 50 mM Tris-Cl, pH 7.4, 1 mM EDTA, 1% Triton X-100, and protease inhibitor mixture (Roche Molecular Biochemicals), followed by 30 min of incubation at 4°C. The clarified lysate was fractionated by 8% SDS-PAGE, followed by semidry electrophoretic transfer to nitrocellulose and immunodetection as described (12,40).
Confocal Immunofluorescence Microscopy-Oocytes previously injected with water or with cRNAs encoding the indicated human SLC26A6 isoforms were fixed, clarified, and immunostained as described previously (40). Single oocytes were imaged with a Bio-Rad MRC1012 confocal laser-scanning immunofluorescence microscope.
Isotopic Flux Studies-Unidirectional 36 Cl Ϫ influx studies were carried out in ND-96 containing 10 M bumetanide for 15-, 30-, or (rarely) 60-min periods as described previously (12,41 4 Cl (20 mM) also activates human SLC26A6(LϩQ) but not mouse slc26a6/CFEX. The normalized values represent activated efflux rate constants for each oocyte divided by the preactivation steady state efflux rate constant for the same oocyte. "Control" represents the unstimulated value. F, mouse slc26a6/CFEX-mediated 36 Cl Ϫ efflux is minimally stimulated by butyrate removal and insensitive to NH 4 Cl. The normalized values are as in E. Oocytes were injected with 10 ng of the indicated cRNA. Values in the bar graphs present means Ϯ S.E. for n oocytes. assay was initiated by transfer of individual oocytes to 6-ml borosilicate glass tubes, each containing 1 ml of efflux solution. At defined intervals, 0.95 ml of this efflux solution was removed for scintillation counting and replaced with an equal volume of fresh efflux solution. Following completion of the assay with a final efflux period in the absence of bath chloride (substituted by sodium isethionate and by gluconate salts of potassium, calcium, and magnesium), each oocyte was lysed in 100 ml of 2% SDS. Samples were counted for 3-5 min such that the magnitude of two standard deviations was Ͻ5% of the sample mean.
Efflux data were plotted as ln(percentage of cpm remaining in the oocyte) versus time. 36 Cl Ϫ efflux rate constants were measured from linear fits to data from the last three time points sampled for each experimental condition. Within each experiment, water-injected and cRNA-injected oocytes from the same frog were subjected to parallel measurements. On each experimental day, activity of the tested human SLC26A6 polypeptides was compared with that of mouse slc26a6 polypeptide tested in the same lot of oocytes on the same day. Each experimental condition and each cRNA was tested in oocytes from at least two frogs. To vary pH i , oocytes were preexposed to 40 mM sodium butyrate (substituting for NaCl) prior to initiation of an efflux experiment to produce intracellular acidification. Upon its removal from the bath (with restoration of chloride) during the course of the efflux experiment, pH i rapidly alkalinized, whereas pH o remained constant (42). Other oocyte groups were exposed to 20 mM NH 4 Cl during the course of efflux experiments. Drugs were added to the bath or were injected into oocytes either prior to or together with isotope as indicated. 35  Fluorescence Ratio Measurement of Oocyte pH i -Oocyte pH i was monitored during bath superfusion using BCECF fluorescence excitation ratio imaging, as described previously (12,44,45). Cl Ϫ /HCO 3 Ϫ and Cl Ϫ /OH Ϫ exchange activities were assayed by measurement of dpH i /dt during bath Cl Ϫ substitution with isethionate and during subsequent restoration of Cl Ϫ . Initial rates of intracellular alkalinization and acidification were computed by linear least squares fit to at least six consecutive fluorescence excitation ratio measurements. Data acquisition and analysis were with MetaFluor software (Universal Imaging, Chester, PA). dpH i /dt values were compared by two-tailed t test. Values of initial pH i in the presence of Cl Ϫ were indistinguishable among groups studied in the presence of CO 2 and were similarly indistinguishable among groups in the nominal absence of CO 2 (see Supplementary Table I).
Qualitative Estimate of Extracellular [Cl Ϫ ] by Measurement of Lucigenin Fluorescence Intensity-The impermeant chloride indicator dye lucigenin (10 M) was in vitro calibrated to [Cl Ϫ ] between 0 and 40 mM in solution balanced with sodium gluconate to achieve a final anion concentration of 110 mM. Using suboptimal filters ( Ex ϭ 380 nm; Em ϭ 510 nm), the Stern-Vollmer constant (K SV ) of 10 M lucigenin in free solution was measured by linear regression analysis (46,47) to be 221 M Ϫ1 (in contrast to 390 M Ϫ1 measured at optimal wavelengths). Perioocyte [Cl Ϫ ] in droplets was estimated by a modification of previous methods used to measure perioocyte pH (41,48). Individual oocytes previously injected with water or with cRNA encoding mouse slc26a6/ CFEX or human SLC26A6 LϩQ were placed in a sodium gluconate bath in a coverslip bottom chamber on the microscope stage. The medium was removed by suction and replaced with a ϳ1.2-l droplet of 5% CO 2 /24 mM NaHCO 3 -buffered Cl Ϫ -free flux medium containing 10 M lucigenin. The humidified chamber atmosphere was equilibrated with 5% CO 2 and sealed with a coverslip to maintain nominally constant droplet volume. Lucigenin fluorescence intensity (F) was monitored and recorded for at least 15 min. Initial droplet fluorescence (F o ) was quenched as droplet [Cl Ϫ ] increased (nominally due to Cl Ϫ exit from the oocyte across its plasma membrane) according to the Stern-Vollmer Qualitative Estimate of Intraoocyte [Cl Ϫ ] by Measurement of MQAE Fluorescence Intensity-Oocytes previously injected with water or with cRNA encoding mouse slc26a6/CFEX or human SLC26A6 SϪQ were Cl Ϫ -depleted and nitrate-loaded by overnight incubation in Cl Ϫ -free nitrate medium. Oocytes were then injected with 50 nl of stock solution containing 100 M MQAE to achieve an estimated final oocyte concentration of 5-10 M MQAE and allowed to recover for 5-10 min in nitrate medium. One pair of MQAE-injected oocytes was placed in a glass bottom perfusion chamber on the microscope stage. Fluorescence intensity was monitored in two separated regions of interest in each oocyte while the perfusion medium was changed from 96 mM sodium nitrate to NaCl (ND-96). Decreased MQAE fluorescence intensity was interpreted as elevation of intraoocyte [Cl Ϫ ] (49) secondary to Cl Ϫ entry across the oocyte plasma membrane. MQAE ( Ex ϭ 340 nm; Em ϭ 460 nm) was in vitro calibrated to [Cl Ϫ ] between 0 and 35 mM in balancing gluconate solutions with total anion concentration of 110 mM, yielding K SV ϭ 206 M Ϫ1 (n ϭ 2). However, attempts to calibrate intraoocyte MQAE fluorescence intensity were not successful due to low apparent efficacy in oocytes of the Cl Ϫ /OH Ϫ exchanger ionophore, tributyltin, even at bath concentrations up to 80 M.
Two-electrode Voltage Clamp Measurements of Oocyte Current-Microelectrodes from borosilicate glass made with a Narashige puller were filled with 3 M KCl and had resistances of 2-3 megaohms. Oocytes were placed in a 1-ml chamber (model RC-11; Warner Instruments, Hamden CT) on the stage of a dissecting microscope and impaled with microelectrodes under direct view. Steady-state currents achieved within 2-5 min following bath change or drug addition were measured with a Geneclamp 500 amplifier (Axon Instruments, Burlingame, CA) interfaced to an HP computer with a Digidata 1200 interface (Axon). Data acquisition and analysis utilized pCLAMP 8.0 software (Axon). The voltage pulse protocol generated with the Clampex subroutine consisted of 20-mV steps between Ϫ100 and ϩ40 mV, with durations of 738 ms separated by 30 ms at the holding potential of Ϫ30 mV. Bath resistance was minimized by the use of agar bridges filled with 3 M KCl, and a virtual ground circuit clamped bath potential to zero.

Four Human SLC26A6 Variant Polypeptides Are Expressed at or near the Surface of the Xenopus Oocyte-
In contrast, all four tested human SLC26A6 polypeptides mediate 35 SO 4 2Ϫ influx 10 times more slowly than mouse slc26a6/CFEX (Fig. 4D) and 36 Cl Ϫ influx 20 times more slowly than the mouse polypeptide (Fig.  4A). In numerous experiments performed by four individuals over a 2-year period, the maximum value observed for human SLC26A6-mediated influx of 36 Cl Ϫ was 0.9 nmol/30 min. Even more remarkable was that human SLC26A6-mediated 36 Cl Ϫ efflux was undetectable in most experiments (Fig. 4, B and C) and only rarely exceeded values observed with water-injected oocytes (see Fig. 9A).
Human SLC26A6 Variant Polypeptides Transport [ 14 C]Oxalate at Rates Almost as Fast as Those of Mouse slc26a6/CFEX but Differ in Regulation-The human SLC26A6 polypeptides were next examined for their ability to transport oxalate, a previously reported property of mouse slc26a6 (3,23,24). Consistent with the apparent surface expression of human SLC26A6 polypeptides (Fig. 3), [ Table I). C, nominal Cl Ϫ /OH Ϫ exchange by representative oocytes expressing mouse slc26a6/CFEX (triangles) or human SLC26A6(LϩQ) (squares). D, rates of nominal Cl Ϫ /HCO 3 Ϫ exchange (white bars) and Cl Ϫ /OH Ϫ exchange (black bars) in oocytes injected with water or expressing SLC26A6(LϩQ) or mouse slc26a6/CFEX. Values in the bar graphs are means Ϯ S.E. for n oocytes. *, p Ͻ 0.001 versus the same assay in water-injected oocytes; #, p Ͻ 0.005 versus Cl Ϫ /OH Ϫ exchange by the same polypeptide. Mean initial resting pH i values for all groups were indistinguishable in the presence of CO 2 /HCO 3 Ϫ . Mean initial resting pH i values in room air were similarly indistinguishable among groups (Supplemental Table I). exhibited by mouse slc26a6/CFEX (Fig. 5A). [ 14 C]Oxalate efflux rates into Cl Ϫ bath reached 80% of those exhibited by the orthologous mouse protein (Fig. 5, B and C).
Mouse slc26a6/CFEX-mediated 36 Cl Ϫ influx is insensitive to extracellular pH, but regulation of Cl Ϫ /Cl Ϫ exchange by intracellular pH has not been reported. The bath addition of the weak acid butyrate acidifies oocytes, and its removal from the bath produces intracellular alkalinization at constant extracellular pH (12,43,44). Such intracellular alkalinization activates SLC26A3/DRA and SLC4A2/AE2. In contrast, bath addition of NH 4 Cl activates anion exchange by SLC26A3/DRA (12) and by SLC4A2/AE2 (52) despite acidification of the oocyte's intracellular pH to levels that would be inhibitory if produced by other means (65). Mouse slc26a6/CFEX-mediated efflux of [ 14 C]oxalate and of 36 Cl Ϫ were minimally enhanced by butyrate removal-induced intracellular alkalinization and unaffected by NH 4 Cl (Fig. 5, E and F). In contrast, human SLC26A6 LϩQmediated [ 14 C]oxalate efflux was stimulated 6-fold by butyrate removal and 2.5-fold by NH 4 Cl (Fig. 5E). In parallel experi-ments performed at the same time with the same group of oocytes (not shown), NH 4 Cl also stimulated AE2/SLC4A2-mediated 36 Cl Ϫ efflux as previously reported (51).
Two Unusual SLC26A6 Polypeptide Variants Are Inactive-SLC26A6d encodes the N-terminal cytoplasmic tail and transmembrane domain of SLC26A6a ((SϪQ), but with a unique C-terminal cytoplasmic tail sequence replacing the evolutionarily conserved STAS domain. SLC26A6c is characterized by in-frame deletion of exons encoding two putative transmembrane domains, with retention of the STAS domain (Fig. 1). Xenopus oocytes injected with either of these variant cRNAs were reported to exhibit DIDS-sensitive uptake of 36 Cl Ϫ and of 35 SO 4 2Ϫ equivalent in magnitude to that of SLC26A6a (25). We found that oocytes expressing SLC26A6a mediated low rates of 36 Cl Ϫ influx and very low efflux rates, as did oocytes expressing the SLC26 variants shown in Fig. 4. Moreover, oocytes expressing SLC26A6c or SLC26A6d cRNAs exhibited no detectable 36 Cl Ϫ uptake or efflux (not shown).
Therefore, we compared the A6a, A6c, and A6d variants of human SLC26A6 as transporters of oxalate and sulfate. Con- FIG. 9. Nonchloride bath anions do not elicit increased 36 Cl ؊ efflux from oocytes expressing human SLC26A6, but elevated temperature does so. A, 36 Cl Ϫ efflux from oocytes previously injected with water or with 10 ng of cRNAs encoding human SLC26A6(LϩQ) or mouse slc26a6/CFEX into baths containing the indicated anions at 96 mM or into 24 mM HCO 3 Ϫ plus 5% CO 2 with balancing isethionate. 5 Results are presented as normalized rate constants (means Ϯ S.E.), where 100% represents the rate constant for 36 Cl Ϫ efflux by mouse slc26a6/CFEX into the Cl Ϫ bath (0.12 Ϯ 0.015 min Ϫ1 ). Inset, time course of 36 Cl Ϫ efflux from representative oocytes previously injected with water or expressing human SLC26A6(LϩQ) (two upper curves) or mouse slc26a6/CFEX (two lower curves) in exchange for bath Br Ϫ and for bath Cl Ϫ . B, elevated temperature activates 36 Cl Ϫ /Cl Ϫ exchange by representative oocytes previously injected with 10 ng of cRNA encoding human SLC26A6(LϩQ) or with 50 pg of cRNA encoding mouse slc26a6/CFEX. Injection with 10 ng of mouse slc26a6/CFEX cRNA led to efflux rates too rapid for analysis. C, summarized results of 36 Cl Ϫ efflux temperature dependence experiments performed as in B, with 10 ng of cRNA encoding human SLC26A6(LϩQ) or with 50 pg of cRNA encoding mouse slc26a6/CFEX. Values in bar graphs are means Ϯ S.E. for n oocytes. *, p Ͻ 0.01 versus rate constant at 20°C. focal immunofluorescence microscopy revealed SLC26A6c expression at or near the oocyte surface (not shown). The alternative C-terminal sequence of SLC26A6d rendered it undetectable with the C-terminal antibody employed. Fig. 6A shows that SLC26A6a mediated [ 14 C]oxalate uptake at rates comparable with those of SLC26A6(SϪQ) (Fig. 4), whereas the SLC26A6c and A6d variant polypeptides were both inactive. SLC26A6c and A6d were similarly inactive as sulfate transporters, whereas SLC26A6a exhibited activity comparable with that observed for all human SLC26A6 isoforms studied in Fig.  4 (Fig. 6B).
Human  (3,24,26,27,28). As shown in Fig. 7, the four human SLC26A6 polypeptides tested along with mouse slc26a6/CFEX exhibited robust, bidirectional activities of Cl Ϫ /HCO 3 Ϫ exchange (Fig. 7, A and B) and nominal Cl Ϫ /OH Ϫ exchange (Fig.  7, C and D). Cl Ϫ /base exchange rates of the human SLC26A6 polypeptides were only slightly lower than those of mouse slc26a6/CFEX (Fig. 7D). These relative rates paralleled those of [ 14 C]oxalate efflux (Fig. 5) but differed from the very low 36 Cl Ϫ fluxes mediated by human SLC26A6 relative to those of mouse slc26a6/CFEX (Fig. 4). All human polypeptides also displayed high initial rates of exchange of intracellular oxalate for bath HCO 3 Ϫ . These rates decreased greatly within 5 min (Fig. 7, A and B).
Human SLC26A6 Polypeptide Variants Mediate cAMP-stimulated Cl Ϫ /HCO 3 Ϫ Exchange Activity, Further Enhanced by CFTR Coexpression-We have shown in Xenopus oocytes that co-expressed CFTR enhances and confers cAMP sensitivity upon human DRA/SLC26A3-mediated Cl Ϫ /HCO 3 Ϫ exchange activity (12). Ko et al. (28). showed in HEK-293 cells that CFTR co-expression with mouse slc26a6/CFEX similarly stimulates cAMP-sensitive Cl Ϫ /base exchange activity. Fig. 8, A and C, now shows that oocytes expressing either human SLC26A6(LϩQ) or mouse slc26a6/CFEX mediate cAMP-stimulated Cl Ϫ /HCO 3 Ϫ exchange. This stimulation is further enhanced by co-expression of wild type CFTR but is prevented by co-expression of mutant CFTR ⌬F508 (Fig. 8, B and C), which traffics inefficiently to the oocyte plasma membrane (40).
Further Characterization of the Minimal 36 Cl Ϫ Flux in Oocytes Expressing Human SLC26A6 Polypeptides-We considered several possible explanations for the puzzling discrepancy (unprecedented in our experience) between the minimal or absent 36 Cl Ϫ transport exhibited by human SLC26A6 and its robust Cl Ϫ /base exchange activity. Extensive control experiments systematically ruled out inadvertent cDNA mutations, variations of influx assay conditions, and variations in 36 Cl Ϫ isotope preparations as causes. 4 The experiments of Fig. 9 tested several conditions and maneuvers for possible stimulation of 36 Cl Ϫ transport activity by human SLC26A6. In contrast to high exchange rates of intracellular 36 Cl Ϫ with 96 mM extracellular Br Ϫ , 96 mM nitrate, or (at considerably lower rates) 24 mM HCO 3 Ϫ exhibited by mouse slc26a6/CFEX, human SLC26A6 exhibited very low rates of these exchange activities (Fig. 9A). 5 Both human SLC26A6(LϩQ) and mouse slc26a6/CFEX exhibited substantial temperature sensitivity, but the former was not activated to a greater degree than was the latter (Fig. 9, B and C). Thus, differential temperature sensitivities of 36 Cl Ϫ /Cl Ϫ exchange did not explain the different 36 Cl Ϫ transport rates by the orthologous polypeptides.
We therefore tested the possibility that the reduced rate of 36 Cl Ϫ transport by human SLC26A6 compared with that of mouse slc26a6 would be reflected by fluorometric measures of Cl Ϫ transport. Cl Ϫ efflux from oocytes into a Cl Ϫ -free surrounding droplet containing gluconate/HCO 3 Ϫ was estimated from the rate and magnitude of quench of lucigenin fluorescence in the droplet (49) (Supplemental Fig. 1). When the fluorescence quench in droplets surrounding water-injected oocytes (n ϭ 5) was normalized to a value of 1, the relative quench in droplets around oocytes previously injected with 10 ng of mouse slc26a6/ CFEX cRNA was 4.3 Ϯ 0.92 (n ϭ 8, p Ͻ 0.001 versus water). Relative quench in the droplets surrounding oocytes injected with 10 ng of human SLC26A6(LϩQ) cRNA was 1.8 Ϯ 0.18 (n ϭ 8, p Ͻ 0.02 versus water), considerably lower than that of mouse slc26a6/CFEX (p Ͻ 0.003).
We measured net Cl Ϫ influx into Cl Ϫ -depleted, nitrate-containing oocytes as quench of intracellular MQAE fluorescence upon substitution of bath nitrate with bath Cl Ϫ (Supplemental Fig. 1, C and D). The relative quench was 3-fold greater in oocytes expressing mouse slc26a6/CFEX than in those expressing human SLC26A6(SϪQ) (n ϭ 4, p Ͻ 0.001), which in turn exceeded by 2-fold that measured in water-injected oocytes (n ϭ 4, p Ͻ 0.001) in a separate, paired experiment. Thus, both net Cl Ϫ efflux in exchange for HCO 3 Ϫ and net Cl Ϫ influx in exchange for nitrate as estimated by fluorescence quenching were 3-4-fold higher in oocytes expressing mouse slc26a6/ CFEX than in oocytes expressing human SLC26A6.
High Rates of 36 Cl Ϫ Flux Are Encoded by the Transmembrane Domain of Mouse slc26a6/CFEX in Chimeras-We reasoned, despite the similarity of measured Cl Ϫ /base exchange 4 The possibility of a mutation in the human SLC26A6 sequences was ruled out by complete resequencing in two independent laboratories of SLC26A6(LϪQ) cDNA purchased separately from MGC by the two laboratories and by complete resequencing of SLC26A6(SϪQ) cDNAs cloned in two independent laboratories. Injection into Xenopus oocytes of each of the above cRNAs produced indistinguishable flux results for 36 Cl Ϫ and for [ 14 C]oxalate. A contribution from the expression vector was unlikely, since SLC26A6(SϪQ) cRNA transcribed from the standard MGC vector pCMV-SPORT6 or from the oocyte expression vector pGEMHE yielded identical flux results. Differences among published influx assay conditions were not responsible, since reduction of extracellular [Cl Ϫ ] from our standard 100 mM to values of 8 mM (3) or 3 mM (23-25) as used by others did not change the relative rates of 36 Cl Ϫ influx exhibited by the human and mouse orthologs (not shown); nor were relative rates of 36 Cl Ϫ influx or efflux by the human and mouse orthologs altered in the Ca 2ϩ -and Mg 2ϩ -free conditions used for oxalate flux measurements (not shown). Lot-specific problems with the isotope were ruled out, since 36 Cl Ϫ flux results were identical with two different lots of Na 36 Cl from ICN, with Na 36 Cl from Amersham Biosciences, and with H 36 Cl from Amersham Biosciences titrated to neutrality with NaOH (not shown). Any potential chlorine isotope effect (60) should account for only ϳ1% difference in transport rate between 36 Cl Ϫ and nonradioactive 35/37 Cl Ϫ . An oxidizing contaminant potentially produced during 36 Cl Ϫ generation (such as hypochlorous acid) was not evident, insofar as oocyte injection of the powerful antioxidant pyrrolidine dithiocarbamate (61) to a final estimated intracellular concentration of 200 M neither rescued 36 Cl Ϫ transport by human SLC26A6 (n ϭ 6) nor altered 36 Cl Ϫ transport by mouse slc26a6 (n ϭ 6, not shown). Moreover, Cl Ϫ /HCO 3 Ϫ exchange rates of human SLC26A6(LϩQ) were statistically indistinguishable in BCECF-loaded oocytes previously injected (as for the standard assay of 36 Cl Ϫ efflux) with 36 Cl Ϫ (dpH i /dt ϭ 1.25 ϩ 0.26 min Ϫ1 , n ϭ 5) or with the same amount of nonradioactive Cl Ϫ (dpH i /dt ϭ 1.46 ϩ 0.13 min Ϫ1 , n ϭ 5, p ϭ 0.23). Indeed, a parallel 36 Cl Ϫ efflux assay with oocytes prepared from the same frog on the same day confirmed the results of Fig. 3A obtained previously. Thus, the use of radioisotope does not explain the apparent difference between rates of 36 Cl Ϫ flux and Cl Ϫ -dependent base transport (measured as dpH i /dt) mediated by human SLC26A6 polypeptides. 5 Knauf et al. (23) previously reported no detectable 36 Cl Ϫ efflux from oocytes expressing mouse slc26a6/CFEX into a gluconate bath containing 10 mM HCO 3 Ϫ (in the absence of CO 2 gassing). However, Jiang et al. (24) subsequently used pH-sensitive microelectrodes to detect Cl Ϫ / HCO 3 Ϫ exchange in oocytes expressing mouse slc26a6/CFEX, suggesting a possible difference in sensitivity of the two methods.
rates, that the remarkable species difference in rates of 36 Cl Ϫ / Cl Ϫ exchange and of 35 SO 4 2Ϫ transport by mouse slc26a6/CFEX and human SLC26A6 should be explained by the divergent amino acid sequences of these orthologs. We therefore compared unidirectional isotopic fluxes of wild type mouse slc26a6/ CFEX and wild type human SLC26A6(SϪQ) with those of chimeras consisting of the short N-terminal transmembrane domain from one species attached to the C-terminal cytoplasmic domain from the other. As shown in Fig. 10, low relative Cl Ϫ fluxes characteristic of human SLC26A6 (h) were preserved in the chimera with a human transmembrane domain and a mouse C-terminal cytoplasmic domain (hm). In contrast, the high Cl Ϫ fluxes characteristic of mouse slc26a6 (m) were relatively well preserved in the chimera with a transmembrane domain from mouse and a C-terminal cytoplasmic domain from human (mh). Thus, the transmembrane domain sequence differences account for most of the difference in 36 Cl Ϫ /Cl Ϫ exchange rates between SLC26A6 orthologs. Both chimeras exhibited the human oxalate flux phenotype, and the phenotype for sulfate flux was an intermediate one. Thus, distinct transport substrates interact with or are influenced by nonidentical sets of amino acid residues in the orthologous anion transporter proteins.
Expression in Xenopus Oocytes of Human SLC26A6 or of Mouse slc26a6/CFEX Is Associated with Increased Current, but This Current Does Not Represent Electrogenic Exchange of Monovalent Anions-The initial report of Cl Ϫ /HCO 3 Ϫ exchange by mouse slc26a3/DRA expressed in HEK-293 cells showed no functional consequence of cell depolarization with a high K ϩ bath (11). In addition, membrane potential of oocytes expressing human SLC26A3/DRA was no more sensitive to bath Cl Ϫ removal than was that of water-injected oocytes (12). These data suggested that DRA-mediated anion exchange is electroneutral. Similarly, Jiang et al. (24) observed that the modest hyperpolarization of water-injected oocytes and oocytes expressing mouse slc26a6 upon bath Cl Ϫ removal was indistinguishable. In contrast, Ko et al. (28) reported that both murine slc26a3/DRA and murine slc26a6/CFEX expressed in Xenopus oocytes generate voltage clamp currents with properties of electrogenic Cl Ϫ /base exchange. Whereas unclamped oocytes expressing mouse slc26a3 depolarized during bath Cl Ϫ removal (28), oocytes expressing mouse slc26a6 hyperpolarized in response to the same stimulus (3,28). Interestingly, however, nominal Cl Ϫ /OH Ϫ exchange mediated by mouse slc26a6 was not accompanied by consistent, bath Cl Ϫ removal-induced hyperpolarization (3).
These varied results prompted our systematic electrophysiological comparison of oocytes expressing mouse slc26a6/CFEX and human SLC26A6. Fig. 11A shows that expression of mouse slc26a6/CFEX modestly increased oocyte current measured at Ϫ100 mV to Ϫ282 Ϯ 41 nA (n ϭ 8 from two frogs and to Ϫ275 Ϯ 8 nA in 33 additional oocytes from five additional frogs (not shown) subjected to different solution change protocols). During this series of sequential bath changes from Cl Ϫ to gluconate and then to gluconate/HCO 3 Ϫ and finally into Cl Ϫ /HCO 3 Ϫ , the maximum change in reversal potential (⌬E rev ) was ϩ3.3 mV, and current magnitudes did not change. Mouse slc26a6/CFEXassociated currents in 18 additional oocytes from three more frogs subjected only to Cl Ϫ -gluconate bath shifts in the absence of CO 2 /HCO 3 Ϫ exhibited similar minimal change in ⌬E rev . Thus, slc26a6/CFEX-mediated monovalent anion exchange was not detectably electrogenic, whether in the absence or presence of CO 2 /HCO 3 Ϫ . 6 The lack of evidence supporting electrogenic exchange of monovalent anions by mouse slc26a6/CFEX was not attributable to inadequate functional expression, as shown in Fig. 11B. The mean change in E rev upon bath change from Cl Ϫ to gluconate was Ϫ5.9 Ϯ 1.1 mV (n ϭ 12), consistent with the results of Fig. 11A. However, upon subsequent bath addition of 1 mM oxalate, mean E rev shifted a further Ϫ23.9 Ϯ 2.5 mV (n ϭ 12). An additional group of seven oocytes (from a different frog) expressing mouse slc26a6/CFEX also exhibited a mean ⌬E rev of Ϫ35 mV in response to the oxalate addition to a gluconate bath (as a sole maneuver, not shown). This evidence for electrogenic oxalate/chloride exchange by mouse slc26a6/CFEX is consistent with a previous report of CFEX-mediated electrogenic oxalate/HCO 3 Ϫ exchange (24). The data further strengthen the conclusion from Fig. 11A that monovalent anion exchange by mouse slc26a6 is electroneutral. Fig. 11C reveals that oocytes expressing human SLC26A6 exhibited a similar lack of change in current magnitude and E rev in response to bath changes from Cl Ϫ to gluconate and then to HCO 3 Ϫ /gluconate and finally to HCO 3 Ϫ /Cl Ϫ . Human SLC26A6-associated currents in 21 additional oocytes from four more frogs subjected only to Cl Ϫ -gluconate bath shifts exhibited similar minimal changes in ⌬E rev (not shown). These data strongly suggest that monovalent anion exchange by human SLC26A6 is also electroneutral. Fig. 11D confirms the finding of Fig. 9A for mouse slc26a6/ CFEX that expression of human SLC26A6 LϩQ modestly increased oocyte inward current in Cl Ϫ bath (measured at Ϫ100 mV) from the control value of Ϫ91 ϩ 26 nA to Ϫ363 ϩ 108 nA (n ϭ 10 oocytes from two frogs). When these SLC26A6-expressing oocytes were subjected to sequential bath changes, the mean reversal potential (E rev ) changed Ϫ2.2 Ϯ 1.7 mV from Cl Ϫ into gluconate bath. However, upon the subsequent addition to the gluconate bath of 1 mM oxalate, E rev shifted only Ϫ2.8 Ϯ 2.1 mV (p Ͻ 0.003 compared with the oxalate-induced shift in E rev in oocytes expressing mouse slc26a6/CFEX). Thus, in contrast to mouse slc26a6, human SLC26A6-mediated oxalate/Cl Ϫ ex- 6 Ko et al. (28) also supported the claim of voltage-sensitive anion transport with their observation of asymmetric tail currents (holding potential was 0 mV, occupying 67% of clamp time). In contrast, oocytes expressing mouse slc26a6 in our experiments (with holding potential of Ϫ30 mV occupying 4% of clamp time) exhibited symmetrical tail currents (not shown whether in the absence or presence of CO 2 /HCO 3 Ϫ . The longer and more extreme depolarization of oocytes in the former protocol may account in part for this difference. Ϫ , 5% CO 2 (plus 72 mM gluconate) to 24 mM HCO 3 Ϫ , 5% CO 2 (plus 72 mM Cl Ϫ ). The I-V curve of water-injected oocytes (control, n ϭ 4) was recorded in Cl Ϫ bath. B, I-V curves of oocytes expressing mouse slc26a6/CFEX (n ϭ 12) recorded during bath change from Cl Ϫ to gluconate, followed by the addition of 1 mM oxalate. E rev hyperpolarizes Ϫ24 mV upon the addition of oxalate. C, I-V curves of oocytes expressing human SLC26A6(LϩQ) (n ϭ 6) recorded during sequential bath changes from Cl Ϫ to gluconate to 24 mM HCO 3 Ϫ plus 5% CO 2 (plus 72 mM gluconate) to 24 mM HCO 3 Ϫ plus 5% CO 2 (plus 72 mM Cl Ϫ ). D, I-V curves of oocytes expressing human SLC26A6(LϩQ) (n ϭ 10) recorded during bath change from Cl Ϫ to gluconate, followed by the addition of 1 mM oxalate. The I-V trace of water-injected oocytes (control, n ϭ 4) is reproduced from A. change did not exhibit evidence of electrogenicity. 7 The voltage clamp results supporting electroneutrality of monovalent anion exchange were corroborated by measurement of 36 Cl Ϫ efflux rate constants and of Cl Ϫ /HCO 3 Ϫ exchange rates in BCECF-loaded oocytes during depolarization with high K ϩ . The rate constant for 36 Cl Ϫ efflux into ND96 from oocytes expressing mouse slc26a6 was 0.044 Ϯ 0.005 min Ϫ1 , and after shift into K ϩ bath it was 0.049 Ϯ 0.005 min Ϫ1 (means Ϯ S.E., n ϭ 7). The 20-fold lower 36 Cl Ϫ efflux rate constants in oocytes expressing human SLC26A6(SϪQ) also did not change in response to high K ϩ depolarization (n ϭ 3, not shown). dpH i /dt upon bath Cl Ϫ removal in BCECF-loaded oocytes expressing mouse slc26a6/CFEX was 0.0185 Ϯ 0.00034 units/ min (S.E., n ϭ 9) in the presence of Na ϩ and 0.0187 Ϯ 0.00033 units/min (n ϭ 13) in the same oocytes subsequently exposed to high K ϩ (p Ͼ 0.7). Oocytes expressing human SLC26A6(SϪQ) exhibited dpH i /dt values of 0.0153 Ϯ 0.00033 and 0.0163 Ϯ 0.00040 during Cl Ϫ removal first in the presence of Na ϩ and then in high K ϩ bath (n ϭ 11, p Ͼ 0.05). ϪdpH i /dt values during bath Cl Ϫ restoration were also unchanged by bath Na ϩ substitution with K ϩ in all experiments (not shown). Initial values of pH i prior to Cl Ϫ removal and restoration were indistinguishable in Na ϩ and in K ϩ baths for all oocyte groups (Supplemental Table I). Thus, high K ϩ depolarization did not alter rates of Cl Ϫ /Cl Ϫ exchange or of Cl Ϫ /HCO 3 Ϫ exchange by oocytes expressing mouse slc26a6 or human SLC26A6.
SLC26A6-mediated Anion Exchange and Its Accompanying Currents in Oocytes Are Pharmacologically Distinct-Mouse slc26a3/DRA-associated current in oocytes was minimally inhibited by DIDS (28), as previously reported for DRA-mediated Cl Ϫ /HCO 3 Ϫ exchange (11,12) and 36 Cl Ϫ flux (12). However, the effect of anion transport inhibitors on the bath Cl Ϫ removalinduced hyperpolarization of mouse slc26a6-expressing oocytes was not reported (3,28). Fig. 12, A and B, shows that oocyte currents associated with expression of either mouse slc26a6/ CFEX or human SLC26A6(SϪQ) were not inhibited by 100 M NS3623, despite 90% inhibition of associated [ 14 C]oxalate flux (Fig. 5C). Moreover, the subsequent addition to these same oocytes of 500 M DIDS did not inhibit currents in either set of oocytes (not shown). However, Fig. 12C shows that the negative shift in E rev produced by the bath addition of 1 mM oxalate to oocytes expressing mouse slc26a6/CFEX (in this experiment only Ϫ10.5 ϩ 2.4 mV) was fully reversed by 500 M DIDS (with a subsequent positive shift in E rev of 11.2 Ϯ 2.5 mV). Fig. 12D confirms that DIDS completely inhibited Cl Ϫ /HCO 3 Ϫ exchange activities of both human SLC26A6(SϪQ) and mouse slc26a6/CFEX, as previously shown for mouse slc26a6/CFEX (24,26). The small 36 Cl Ϫ influx activity mediated by human SLC26A6(LϪQ) was also completely inhibited by DIDS (Fig.  12E), as previously reported for the large 36 Cl Ϫ influx activity of mouse slc26a6/CFEX (3,23). Moreover, DIDS was equally effective as an inhibitor of mouse slc26a6/CFEX-mediated 36 Cl Ϫ efflux and of [ 14 C]oxalate efflux mediated by both human and mouse orthologs (Fig. 12F). Figs. 5D, 11, and 12 together show that expression of either mouse slc26a6 or human SLC26A6 induces in Xenopus oocytes endogenous currents that differ pharmacologically from SLC26A6-mediated anion exchange. Therefore, the incremental currents and the (electroneutral) monovalent anion exchange measured in these oocytes are not likely to be mediated by the same protein. DISCUSSION Functional study of mouse slc26a6/CFEX has led to novel models of human pathophysiology, but anion exchange mechanisms and biological functions of SLC26 transporters remain controversial. The remarkable degree of sequence difference between human SLC26A6 and mouse slc26a6 and the lack of functional data on SLC26A6 variant polypeptides prompted us to perform the first comprehensive functional comparison of these two orthologous gene products. Our results differ in some important ways from those previously reported and encourage reevaluation of the pathophysiological models built upon those earlier results.
Orthologous Mouse and Human SLC26A6 Polypeptides Differ in Anion Selectivity-Four of the human SLC26A6 polypeptide variants revealed similar activity profiles, but these profiles for some functions differed dramatically from those of their murine ortholog (Figs. 4 and 5). Two additional human polypeptide variants previously reported to be functional were shown to be inactive (Fig. 6). Previous 36 Cl Ϫ flux studies of human SLC26A6 revealed low level activity (25) or no function (38). These studies were confirmed by our observations of low 36 Cl Ϫ influx activity and very low 36 Cl Ϫ efflux activity by human SLC26A6, despite robust bidirectional transport of 36 Cl Ϫ by mouse slc26a6 (Fig. 4) as previously reported (3,23). Sulfate influx exhibited a qualitatively similar pattern: low activity in oocytes expressing human SLC26A6 and high activity for mouse slc26a6 (Fig. 4). However, tests of oxalate transport confirmed that human SLC26A6 at or near the oocyte surface ( Fig. 3) was indeed highly functional. Human SLC26A6-mediated oxalate efflux was 40 -80% that of mouse slc26a6, and oxalate influx was nearly so (Fig. 5). Chimera studies attributed most of the large species difference in 36 Cl Ϫ transport rate to the transmembrane domain of SLC26A6, whereas the difference in sulfate transport appeared to reside to a greater degree also in the N-terminal cytoplasmic domain (Fig. 10). This species difference in anion selectivity will provide a useful route to definition of the specific amino acid residues in SLC26 polypeptides that govern or modulate binding, translocation, and release of substrate anions.
Differences in Acute Regulation between Mouse and Human SLC26A6 Orthologs-Human SLC26A6-mediated [ 14 C]oxalate efflux was increased acutely both by the alkalinizing stimulus of bath butyrate removal and by the acidifying stimulus of NH 4 ϩ exposure (Fig. 5). In contrast, mouse slc26a6-mediated transport of [ 14 C]oxalate or 36 Cl Ϫ was insensitive to NH 4 ϩ and only minimally stimulated by butyrate removal, consistent with previous reports that mouse slc26a6-mediated transport of 36 Cl Ϫ and of [ 14 C]oxalate are insensitive to acidic bath pH (3,24,23). Stimulation of human SLC26A6-mediated oxalate efflux by butyrate removal (increasing pH i from ϳ6.6 to ϳ7.1) was not due to increased concentration of the divalent oxalate anion, with pK values of 1.3 and 4.3.
Stimulation of human SLC26A6 by NH 4 ϩ was not likely to be secondary to depolarization, since NH 4 ϩ -sensitive human SLC26A-mediated oxalate/Cl Ϫ exchange was weakly if at all electrogenic, whereas NH 4 ϩ -insensitive mouse slc26a6-mediated oxalate/Cl Ϫ exchange was strongly electrogenic. Stimulation of human SLC26A6 by NH 4 ϩ was not secondary to intracellular acidification, given robust stimulation by the alkalinizing stimulus of butyrate removal. The mechanism by which NH 4 ϩ stimulates activity of SLC4A2/AE2 but not SLC4A1/AE1 in Xenopus oocytes also remains unclear. However, elevated NH 4 ϩ concentrations are physiologically important not only in the kidney and the portal vascular bed but also in any epithelial lumens colonized by urease-positive bacteria such as Helicobacter pylori and Pseudomonas aeruginosa (52).
Inactive Physiological Variants of Human SLC26A6 -Four of the six tested human SLC26A6 polypeptides were functionally similar. However, SLC26A6c and SLC26A6d polypeptides expressed no detectable isotopic fluxes (Fig. 6), in contrast to the findings of Lohi et al. (25). This lack of function, reinforced by our observation of SLC26A6a-mediated transport of [ 14 C]oxalate and of [ 35 S]sulfate at levels equivalent to those exhibited by the human SLC26A6(SϪQ), emphasizes the structural and functional importance of both an intact transmembrane domain (not preserved in the SLC26A6c polypeptide) and the STAS domain of the C-terminal cytoplasmic tail (absent from SLC26A6d). The possibility that one or more inactive SLC26A6 variants might serve as a physiological dominant negative remains to be more thoroughly investigated.
Reconciliation of this robust human SLC26A6-mediated Cl Ϫ / base exchange with very low rates of 36 Cl Ϫ flux poses a problem not presented by mouse slc26a6 or by anion exchange mediated by human SLC26A3/DRA (12) and human SLC26A4/pendrin (17,19,28). 8 In contrast, preliminary expression studies with human SLC26A7, SLC26A8, and SLC26A9 reveal similarly low or undetectable 36 Cl Ϫ flux activity in concert with substantial Cl Ϫ /HCO 3 Ϫ exchange activity. 9 Thus, the property of discordant rates of anion exchange measured by different techniques in different conditions is not unique to human SLC26A6. 8  Ϫ exchange mediated by mouse slc26a6 (n ϭ 4) and by human SLC26A6(SϪQ) (n ϭ 4) to levels observed in water-injected oocytes (see Fig. 7, B and D). dpH i /dt was measured in the absence and in the subsequent presence of DIDS for each oocyte (means Ϯ S.E.). Values in the presence of DIDS were at levels observed in water-injected oocytes (Fig. 6). **, p Ͻ 0.001; *, p ϭ 0.003 versus before DIDS; #, t ϭ 0.002 versus human SϪQ. Initial pH i values were statistically indistinguishable (Supplemental Table I). E, DIDS (100 M) similarly inhibits Cl Ϫ uptake by human SLC26A6(SϪQ). F, 100 M DIDS inhibits [ 14 C]oxalate efflux mediated by mouse slc26a6 and by human SLC26A6(SϪQ) and 36 Cl Ϫ efflux mediated by mouse slc26a6. The bars in E and F represent means Ϯ S.E. for n oocytes.

HCO 3
Ϫ flux (equivalent proton flux, J Hϩ ) by mouse slc26a6 was estimated from dpH i /dt ϫ total oocyte buffer capacity 10 as 6.1 M/s. The corresponding value for J Hϩ by human SLC26A6 was 5.2 M/s. Mouse slc26a6-mediated Cl Ϫ /Cl Ϫ exchange measured as 36 Cl Ϫ efflux into Cl Ϫ bath in room air was calculated from the efflux rate constant (0.12-0.3 min Ϫ1 ) and an estimated intraoocyte [Cl Ϫ ] of 43 mM (native ϳ30 mM plus the post- 36 Cl Ϫ injection increment of ϳ13 mM) to yield Cl Ϫ /Cl Ϫ exchange rates of at least 85 M/s, but often higher. Since the relative rate of 36 Cl Ϫ efflux into the HCO 3 Ϫ bath was only 10% of that into the Cl Ϫ bath (Fig. 7A), a low end estimate for Cl Ϫ /HCO 3 exchange rate derived from 36 Cl Ϫ efflux measurements (for which intraoocyte [Cl Ϫ ] is ϳ30 mM) is 8.5 M/s, in moderate agreement with the 6.1 M/s value derived from BCECF fluorescence ratio. Thus, the magnitude of 36 Cl Ϫ transport by mouse slc26a6 in the presence of HCO 3 Ϫ agrees moderately well with the rates of fluorometrically measured HCO 3 Ϫ transport in response to complete removal and restoration of bath Cl Ϫ .
The large gap between human SLC26A6-mediated HCO 3 Ϫ transport at rates ϳ85% those of mouse slc26a6 (Figs. 7 and 8) and 36 Cl Ϫ transport at rates 0 -10% those of mouse (Figs. 4 and 9) remains to be explained. Under the conditions tested, mouse slc26a6 mediates 36 Cl Ϫ /Cl Ϫ exchange at rates at least 10-fold greater than those of Cl Ϫ /HCO 3 Ϫ exchange measured as J Hϩ . In contrast, human SLC26A6 mediates 36 Cl Ϫ efflux into Cl Ϫ bath at rates at least 2.5-fold lower than those of fluorometrically measured Cl Ϫ /HCO 3 Ϫ exchange (5.2 M/s) and sometimes at undetectable rates. Cl Ϫ transport by human SLC26A6 may differ from that by mouse slc26a6 in a requirement for CO 2 / HCO 3 Ϫ for optimal function. The inability to detect such a CO 2 /HCO 3 Ϫ dependence for human SLC26A6 as a consistent stimulation of 36 Cl Ϫ efflux by bath HCO 3 Ϫ (Fig. 9) may reflect effective competition by intracellular HCO 3 Ϫ at the protein's internal Cl Ϫ binding site so as to lower 36 Cl Ϫ efflux toward or below the level of detection.
We therefore propose that human SLC26A6 polypeptides mediate very low rates of Cl Ϫ /Cl Ϫ homoexchange while maintaining moderate to high rates of Cl Ϫ /HCO 3 Ϫ exchange. This may reflect a species-specific CO 2 /HCO 3 Ϫ -dependent conformation secondary to altered binding affinities at transport or regulatory anion binding sites.
Human SLC26A6-mediated Cl Ϫ /HCO 3 Ϫ Exchange Is cAMPsensitive and Further Stimulated by CFTR-Cl Ϫ gradient-dependent HCO 3 Ϫ transport by mouse slc26a6 and human SLC26A6 proteins was stimulated by cAMP to similar degrees. This stimulation by cAMP was enhanced for both orthologs by coexpressed wild-type CFTR (Fig. 8), as shown previously for human (12) and mouse DRA (28) and for mouse slc26a6 (28). The stimulation of bidirectional Cl Ϫ /HCO 3 Ϫ exchange by CFTR probably reflects direct interaction between CFTR and SLC26 from either species (35). Mechanistic contributions to this stimulation from the abrupt switch in conductive anion selectivity of CFTR induced in oocytes by bath Cl Ϫ removal (37) or from extracellular [Cl Ϫ ] control of CFTR gating (64) were neither clearly evident nor possible to rule out. The cAMP sensitivity of both SLC26A6 orthologs is consistent with secretin stimulation of luminal Cl Ϫ /HCO 3 Ϫ exchange in the guinea pig pancreatic duct (53). In contrast, mutant CFTR coexpression abrogated cAMP-stimulation of Cl Ϫ /HCO 3 Ϫ exchange (Fig. 8), consistent with deficient pancreatic HCO 3 Ϫ secretion in cystic fibrosis.
Monovalent Anion Exchange by Mouse slc26a6 and by Human SLC26A6 Is Electroneutral-Although high K ϩ depolarization experiments have not supported an electrogenic Cl Ϫ / HCO 3 Ϫ exchange mechanism for SLC26 polypeptides (11,12), bath Cl Ϫ substitution has been reported to hyperpolarize unclamped oocytes expressing mouse slc26a6 (3,28) and to depolarize oocytes expressing mouse slc26a3 (28). Ko et al. (28) further reported bath anion-dependent shifts in reversal potential of opposite polarity for mouse slc26a3 and slc26a6 in voltage-clamped oocytes (28). These data were invoked to support the proposal of electrogenic Cl Ϫ /HCO 3 Ϫ exchange of opposite stoichiometries (Ն2 HCO 3 Ϫ /1 Cl Ϫ for slc26a6/CFEX versus Ն2 Cl Ϫ /1 HCO 3 Ϫ for slc26a3/DRA). Based on these and other findings, they suggested a model (28,35) for pancreatic ductal secretion of 140 mM HCO 3 Ϫ ; axially arrayed, secretin-stimulated, CFTR-enhanced, electrogenic Cl Ϫ /HCO 3 Ϫ exchange by SLC26A6 in the proximal pancreatic duct and by SLC26A3/DRA in the distal pancreatic duct could sustain HCO 3 Ϫ secretion in the face of high inward HCO 3 Ϫ gradients to achieve substitution of most luminal Cl Ϫ (originally secreted by the acinus and early duct) with HCO 3 Ϫ . They further noted that regulated expression or activity of multiple SLC26 gene products together with CFTR in the same membrane could explain the full range of electroneutral and/or electrogenic Cl Ϫ -dependent HCO 3 Ϫ secretion reported in intact epithelial tissues (28).
This proposal encouraged our further exploration of the electrogenicity of anion exchange by human and mouse orthologs. We found that expression of both mouse slc26a6 and of human SLC26A6 in Xenopus oocytes indeed generated currents in Cl Ϫ bath, albeit of lower magnitude than those reported by Ko et al. (28). However, these oocyte currents were not attributable to Cl Ϫ /OH Ϫ or Cl Ϫ /HCO 3 Ϫ exchange, as judged by the following multiple criteria. First, mouse slc26a6-mediated (water control-subtracted) clamp currents measured at Ϫ40 mV (the approximate membrane potential in unclamped water-injected oocytes undergoing flux assays in Cl Ϫ bath) were only Ϫ25 nA (n ϭ 33; see also Fig. 11A). This current could account for only 5-10% of slc26a6-mediated 36 Cl Ϫ influx (10 -20 nmol/h, or 230 -560 nA) if all exchange flux were electrogenic, or even less in view of the depolarized resting potential (Ϫ26 mV) of slc26a6-expressing oocytes (3). Second, E rev of the slc26a6/ SLC26A6-induced currents did not change significantly in response to bath substitution of monovalent anion transport substrates, whether in the absence or presence of CO 2 /HCO 3 Ϫ . This lack of change in E rev contrasted dramatically with the large E rev shift in oocytes expressing mouse slc26a6 upon the bath addition of divalent oxalate. Third, the induced currents in oocytes expressing either mouse slc26a6 or human SLC26A6 were insensitive to inhibition by several effective blockers of slc26a6-mediated anion flux. Fourth, oocyte depolarization in high K ϩ bath did not modify either the 36 Cl Ϫ efflux rate constant for mouse slc26a6 or the rate of Cl Ϫ /HCO 3 Ϫ exchange (ϩdpH i /dt) mediated by either mouse slc26a6 or by human SLC26A6. Finally, voltage-clamped currents in oocytes expressing either mouse slc26a3 or human SLC26A3 confirmed the electroneutrality of monovalent anion exchange in the absence and presence of CO 2 /HCO 3 Ϫ , exhibiting neither change in current magnitude nor E rev . Expression of mouse slc26a3 in HEK-293 cells similarly failed to produce anion exchange currents detectable above background (35).
Electroneutral Cl Ϫ /HCO 3 Ϫ Exchange and Pancreatic Ductal HCO 3 Ϫ Secretion-The current observations suggest that electrogenic anion exchange by SLC26A6 and SLC26A3/DRA should not be relied on to satisfy the thermodynamic requirements of models of human or mouse pancreatic bicarbonate 10 Equivalent proton flux rates were calculated as dpH i /dt (corrected for values recorded in water-injected oocytes) ϫ total buffer capacity (␤ T ϭ ␤ i ϩ ␤ CO 2 ). Intrinsic buffer capacity (␤ i ) was 19 mM/pH unit (43), and CO 2 buffer capacity (␤ CO 2 ) was calculated from the measured pH i by the Henderson-Hasselbalch equation.
secretion. The secretin-stimulated final [HCO 3 Ϫ ] of mouse pancreatic juice remains uncertain (the unlikely value of 30 mM is the only one yet reported (54)). Even if mouse pancreatic [HCO 3 Ϫ ] reaches levels comparable with the rat's 70 mM, electroneutral Cl Ϫ /HCO 3 Ϫ exchange could suffice to achieve this final concentration (55,56). However, maximally stimulated human pancreatic juice [HCO 3 Ϫ ] resembles more closely the guinea pig's 125-150 mM than the rat's 70 mM. The apical cell membrane potential of isolated guinea pig pancreatic interlobular duct depolarizes only 10 mV, from Ϫ60 to Ϫ50 mV, during stimulation by cAMP or secretin (57). The latter potential of Ϫ50 mV can sustain adequate driving force for electrogenic HCO 3 Ϫ secretion via CFTR or other anion channels without invocation of electrogenic anion exchange, even in the presence of a luminal [HCO 3 Ϫ ] of 125 mM. Moreover, depolarization and hyperpolarization of duct cell apical membrane by changing luminal [K ϩ ] in the presence of 125 mM luminal HCO 3 Ϫ were not dependent on the presence of luminal Cl Ϫ (58). This finding is consistent with previous observations that high extracellular [HCO 3 Ϫ ] inhibits apical Cl Ϫ /HCO 3 Ϫ exchange in the pancreatic duct (53,59) and suggests that conductive HCO 3 Ϫ secretion by the distal pancreatic duct in the presence of high luminal [HCO 3 Ϫ ] is not mediated by anion exchangers. Recent calculations for the guinea pig distal duct predict that electrogenic Cl Ϫ /HCO 3 Ϫ exchange would be of limited consequence (63). The luminal [HCO 3 Ϫ ] achievable by apical exchange of two HCO 3 Ϫ outward for one Cl Ϫ inward (as proposed for slc26a6 (28)) is only 6 mM higher than that achievable by electroneutral exchange. The stoichiometry of 1 HCO 3 Ϫ /2 Cl Ϫ (as proposed for slc26a3/DRA (28)) predicts a luminal [HCO 3 Ϫ ] 6 mM lower than for electroneutral exchange. The increase in relative HCO 3 Ϫ permeability of CFTR induced by low bath Cl Ϫ (37) and the inhibitory gating of CFTR by extracellular Cl Ϫ (62) should also be considered as contributors to the final composition of the pancreatic ductal secretion.
Human SLC26A6 and Mouse slc26a6 Differ in Electrogenicity of Cl Ϫ /Oxalate Exchange-Cl Ϫ /oxalate exchange by mouse slc26a6 was shown under voltage clamp conditions to be strongly electrogenic (Fig. 11B), and the oxalate-dependent currents resembled [ 14 C]oxalate fluxes in their DIDS sensitivity (Fig. 12). These observations correspond to those recorded in unclamped oocytes in both the absence and the presence of CO 2 /HCO 3 Ϫ (24). In the current work, both mouse slc26a6 and human SLC26A6 also mediated high initial rates of oxalate/ HCO 3 Ϫ exchange. However, this exchange slowed considerably within 5 min (Fig. 7), perhaps reflecting a block of HCO 3 Ϫ efflux by subplasmalemmal accumulation of oxalate, since the affinity of mouse slc26a6 for oxalate greatly exceeds that of HCO 3 Ϫ (24). The data together support a mechanism in which divalent oxalate can be exchanged for monovalent Cl Ϫ or HCO 3 Ϫ by orthologs of both species. Curiously, however, although human SLC26A6 exhibited [ 14 C]oxalate influx rates and rates of [ 14 C]oxalate efflux into Cl Ϫ bath that were 30 -80% those of mouse slc26a6, Cl Ϫ /oxalate exchange by human SLC26A6 was not detectably electrogenic (Fig. 11D).
Failure to detect electrogenicity of oxalate/Cl Ϫ exchange by human SLC26A6 may represent a mechanistic difference with mouse slc26a6 arising from amino acid sequence divergence. (A single amino acid change suffices to change both anion selectivity and electrogenicity of the SLC4A1/AE1 anion exchanger (41).) slc26a6 is expressed in mouse proximal tubule brush border (23), where it mediates most or all of the luminal oxalatedependent chloride reabsorption (66). The difference between electroneutral and electrogenic transport might predict a lower rate of proximal tubular oxalate secretion in humans than in mice, should SLC26A6 be similarly rate-limiting in human proximal tubule. Failure to detect electrogenicity of oxalate/Cl Ϫ exchange by human SLC26A6 might alternatively represent activation by human SLC26A6 (in a species-specific manner) of an endogenous oocyte cation current, obscuring detection of oxalate/Cl Ϫ exchange current. It seems unlikely that human SLC26A6 expression levels, if adequate for detection of unidirectional oxalate flux, could be insufficient for detection of oxalate-evoked exchange current.
Although the coincidence of low 36 Cl Ϫ transport rates with robust rates of oxalate flux and Cl Ϫ -dependent HCO 3 Ϫ transport in human SLC26A6 remains for now unexplained, the resolution of this contrast will elucidate at least part of the molecular basis of anion selectivity among and translocation by SLC26 anion exchangers. This insight should enhance understanding of human epithelial HCO 3 Ϫ secretion and its dysfunction in cystic fibrosis and other disorders. Study of SLC26A6-mediated oxalate transport may also improve understanding of familial oxalosis syndromes and of nephrolithiasis-associated hyperoxaluria.