Cloning and Characterization of a Na+-driven Anion Exchanger (NDAE1)

Regulation of intra- and extracellular ion activities (e.g. H+, Cl−, Na+) is key to normal function of the central nervous system, digestive tract, respiratory tract, and urinary system. With our cloning of an electrogenic Na+/HCO3 − cotransporter (NBC), we found that NBC and the anion exchangers form a bicarbonate transporter superfamily. Functionally three other HCO3 −transporters are known: a neutral Na+/ HCO3 − cotransporter, a K+/ HCO3 − cotransporter, and a Na+-dependent Cl−-HCO3 − exchanger. We report the cloning and characterization of a Na+-coupled Cl−-HCO3 − exchanger and a physiologically unique bicarbonate transporter superfamily member. ThisDrosophila cDNA encodes a 1030-amino acid membrane protein with both sequence homology and predicted topology similar to the anion exchangers and NBCs. The mRNA is expressed throughoutDrosophila development and is prominent in the central nervous system. When expressed in Xenopus oocytes, this membrane protein mediates the transport of Cl−, Na+, H+, and HCO3 − but does not require HCO3 −. Transport is blocked by the stilbene 4,4′-diisothiocyanodihydrostilbene- 2,2′-disulfonates and may not be strictly electroneutral. Our functional data suggest thisNa+ driven anionexchanger (NDAE1) is responsible for the Na+-dependent Cl−-HCO3 − exchange activity characterized in neurons, kidney, and fibroblasts. NDAE1 may be generally important for fly development, because disruption of this gene is apparently lethal to the Drosophila larva.

Ionic homeostasis is the key to normal function of most biological systems. This regulation is especially important for tissues with highly specialized functions, such as the central nervous system (CNS), 1 digestive tract, respiratory tract, and urinary system. Active transport of ions by ATPases (pumps) maintains ionic gradients and aid ion channels in "setting" the membrane potential. Secondary active transporters make use of one or more aspects of the membrane electrochemical gradient to specifically move ions and nutrients into and out of cellular compartments.
Physiologically two other HCO 3 Ϫ transporters are known, a K ϩ /HCO 3 Ϫ cotransporter (5) and a Na ϩ -dependent Cl Ϫ -HCO 3 Ϫ exchanger (6,7). Here we report the cloning and characterization of a cation-coupled Cl Ϫ -HCO 3 Ϫ exchanger and a physiologically unique BTS member from Drosophila. When expressed in Xenopus oocytes, this membrane protein mediates the transport of Cl Ϫ , Na ϩ , H ϩ , and HCO 3 Ϫ but does not require HCO 3 Ϫ . Transport is blocked by the stilbene DIDS and may not be strictly electroneutral. Our expression data suggest this Na ϩ driven anion exchanger (NDAE1) (GenBank TM accession number AF047468) is responsible for the Na ϩ -dependent Cl Ϫ -HCO 3 Ϫ exchange activity characterized in neurons (6 -8), kidney (9,10), and fibroblasts (11). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM 1 The abbreviations used are: CNS, central nervous system; NBC, electrogenic Na ϩ /HCO 3 Ϫ cotransporter (i.e. SLC4A4); BTS, bicarbonate transporter superfamily; DIDS, 4,4Ј-diisothiocyanodihydrostilbene-2,2Ј-disulfonate; NDAE1, Na ϩ driven anion exchanger; AE, anion exchanger; bp, base pair(s); RT, reverse transcriptase; PCR, polymerase chain reaction; TM, transmembrane span; pH i , intracellular pH; aCl i , intracellular Cl Ϫ activity; aNa i , intracellular Na ϩ activity; UTR, untranslated region. 2 The electrogenic NBC is currently designated by several nomenclatures in the literature: NBC1, kNBC, pNBC, hhNBC, and SLC4A4 (see Ref. 3 for a detailed explanation). SLC4A4 is the human gene designation indicating "solute carrier family 4A, member 4" by the human genome nomenclature. The clones that are currently referred to as NBC2, mNBC3, and NBC n 1 are likely splice variants of the same gene; however, this has not been explicitly demonstrated. Currently NBC2 is given a designation of SLC4A6 and mNBC3 as SLC4A7. Another apparently distinct human cDNA, SLC4A8, was deposited in GenBank TM (AF069512) but has not yet been functionally characterized.

EXPERIMENTAL PROCEDURES
Cloning-We identified a Drosophila expressed sequence tag (AA567741, deposited by the Berkeley Drosophila Genome Project) with similarity to both the AEs and NBCs. We obtained this Drosophila clone (Research Genetics, St. Louis, MO) and sequenced it (W. M. Keck Biotechnology Resource Laboratory, New Haven, CT). This 3225-base pair clone has an initial Met followed by a 3090-base pair open reading frame and a 3Ј-UTR (104 bp). We directionally subcloned the cDNA into a Xenopus expression plasmid as previously (2). Linearized cDNA was used to make capped cRNA with the SP6 mMessage mMachine kit (Ambion, Austin, TX) as described previously (2) for both Xenopus oocyte studies and in situ hybridization. The full cDNA sequence of Drosophila NDAE1 is GenBank TM accession number AF047468.
Northern Analysis and RT-PCR Protocol-We isolated poly(A) ϩ RNA from Drosophila developmental stages and body segments as described previously (1). We used 2 g of poly(A) ϩ RNA from these stages for denaturing electrophoresis and electroblotting. The NDAE1-cDNA was random primed and 32 P-labeled. Hybridization overnight at 60°C in ExpressHyb (CLONTECH) followed by low stringency washing (42°C with 2ϫ SSC) did not result in discrete hybridization. Reverse transcription was performed using SuperScript RT kit according to the manufacturer's directions (Life Technologies, Inc.) with Drosophila poly(A) ϩ RNA. Using Drosophila NDAE1-specific primers, ExTaq polymerase (Panvera, Madison, WI), and dNTPs, we performed PCR with 30 cycles of 94°C (30 s), 55°C (45 s), and 72°C (45 s). Products were verified with a 0.65% agarose/Tris borate EDTA gel. The gel was Southern blotted onto Zeta-probe (Bio-Rad) and detected using randomprimed, digoxigenin-labeled NDAE1 cDNA according to the manufacturer's instructions (Roche Molecular Biochemicals). Detection of digoxigenin-labeled DNA probe was performed using DIG Luminescent Detection (Roche Molecular Biochemicals), recorded on x-ray film, and digitized using Adobe PhotoShop (Fig. 2a).
Sequence Analysis-Multiple sequence alignments were performed using the Clustal method and the PAM250 residue weight table (DNA Star program, Lasergene, Madison, WI) with the percentage divergence and similarity calculated as previously reported (1) and the alignment shaded and annotated using GeneDoc © .
In Situ Hybridization-To determine NDAE1 mRNA cellular distribution in Drosophila, we made whole mounts of 0 -24 h Drosophila embryos. To make antisense cRNA, the complete Drosophila NDAE1 was directionally cloned into pSport 2 (Life Technologies, Inc.) at EcoRI and SalI. Digoxigenin-labeled antisense NDAE1-cRNA was synthesized using the SP6 promoter and mMessage mMachine (Ambion) as described above and reduced to a mean size of ϳ200 bp by alkaline hydrolysis (16). Embryos were permeabilized using proteinase K treatment. Digoxigenin label was visualized using an anti-digoxigenin antibody coupled to alkaline phosphatase (17). Embryo staining was documented on slide film and subsequently digitized. Hybridization was determined specific if (i) NDAE1 staining was evident in discrete cells making DNA hybridization unlikely and (ii) staining with a sense RNA probe was negative.
Chromosomal Localization-Chromosomes were prepared and hybridized by standard methods (18). Biotin-labeled probes were generated by random hexamer-priming with Biotin-HighPrime ® (Roche Molecular Biochemicals) and the entire NDAE1-cDNA (i.e. the EcoRI/ HindIII fragment of the pSport 2 construct), according to the manufacturer's instructions. Horseradish peroxidase-labeled anti-biotin antibodies were used for detection.
Oocyte Experimental Solutions-The CO 2 /HCO 3 Ϫ -free ND96 contained 96 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , and 5 mM HEPES (pH 7.5 and 195-200 mosM). In CO 2 /HCO 3 Ϫ -equilibrated solutions, 10 mM NaHCO 3 replaced 10 mM NaCl and was maintained by continuous bubbling with 1.5% CO 2 /98.5% O 2 . In O-Na ϩ solutions, choline replaced Na ϩ . In O-Cl Ϫ solutions, gluconate replaced Cl Ϫ . Non-HCO 3 Ϫ solutions were bubbled with 100% O 2 to remove trace CO 2 and HCO 3 Ϫ . Oocyte Electrophysiology-50 nl of water (control) or RNA solution (35 ng of NDAE-cRNA) was injected into stage V/VI Xenopus oocytes. Voltage electrodes, made from fiber-capillary borosilicate and filled with 3M KCl, had resistances of 1-10 M⍀ (1). Ion-selective electrodes (pH, Cl Ϫ , and Na ϩ ) were pulled similarly and silanized with bis-(dimethylamino)-dimethylsilane (Fluka Chemical Corp., Ronkonkoma, NY). pH electrodes tips were filled with hydrogen ionophore 1 mixture B (Fluka) and backfilled with phosphate buffer (pH 7.0). Cl Ϫ electrode tips were filled with a Cl Ϫ ionophore (Corning, Corning, NY) and backfilled with 0.5 M NaCl; Na ϩ electrode tips were filled with sodium ionophore 1 mixture B (Fluka) and backfilled with 0.15 M NaCl. Electrodes were connected to a high impedance electrometer (WPI-FD223 for intracellular pH (pH i ), intracellular Cl Ϫ activity (aCl i ), or intracellular Na ϩ activity (aNa i ) and V m experiments), and digitized output data were acquired by computer. All ion-selective microelectrodes had slopes of Ϫ54 to Ϫ57 mV/decade ion concentration (or activity). pH electrodes were calibrated at pH 6.0 and 8.0; Cl Ϫ and Na ϩ electrodes were calibrated with unbuffered 10 and 100 mM NaCl (ionic strength was not identical). Selectivity of Cl Ϫ was checked using unbuffered 100 mM NaHCO 3 and for Na ϩ using 100 mM KCl. Na ϩ electrodes were greater than 50-fold selective for Na ϩ (19) and Cl Ϫ electrodes were at least 10-fold selective versus HCO 3 Ϫ . For voltage-clamp experiments (Warner Inst. Co., Oocyte Clamp), electrodes were filled with 3 M KCl/agar and 3 M KCl and had resistances of 0.2-0.5 M⍀. Oocytes were clamped to Ϫ60 mV and stepped from Ϫ160 to ϩ60 mV in 20 mV steps; the resulting data were filtered at 5 kHz (8 pole Bessel filter, Frequency Devices) and sampled at 1 kHz as previously reported (19). Data were acquired and analyzed using Pulse and PulseFit (HEKA Instruments, Germany).
Statistical Analysis-Values quantitated are indicated as the mean Ϯ S.E. Ion activities between control and NDAE1 oocytes were shown by a two-tailed t test to have a significance of p Ͻ 0.016 or less.

Characterization of the NDAE1 cDNA and Predicted
Protein-DNA sequencing of our clone revealed a single, long open reading frame flanked by 5Ј-and 3Ј-UTRs (UTRs, 426 and 104 bp, respectively). This Drosophila cDNA encodes a 1030amino acid membrane protein with both sequence homology and predicted topology similar to both the AEs and NBCs. The predicted protein is 43% similar to the cloned NBCs and 32% similar to the AEs (Fig. 1, a and b). Although the NDAE1 hydropathy plot (Fig. 1b) is similar to those of the BTS members, it is most similar to the AEs. A dendrogram of the published BTS sequences (Fig. 1c) implies that NDAE1 forms a new branch of this superfamily. Our NDAE1 topology model ( Fig. 1d) predicts (i) intracellular NH 2 and COOH termini, (ii) 12 transmembrane spans (TMs), (iii) a central exofacial loop with putative N-glycosylation sites, and (iv) multiple putative phosphorylation sites.
Recently the complete sequence of the Drosophila genome was reported (20). Although a predicted gene product "CG4675" in two forms (AAF52496, alt 1 and 2 for proteins and AE003616 for the assembled genomic contig) encoding the ndae1 gene was identified, the sequence analysis is not completely accurate. The predicted protein sequences are missing thirteen NH 2terminal amino acids (MAEKNEYIELPWT) partly encoded from an additional 5Ј-intron. CG4675-alt 2 contains a 69-amino acid insertion (amino acids 32-100 of the NH 2 -truncated protein), which we have not found present in Drosophila mRNA. A second Drosophila gene and protein have homology to the BTS family (CG8177). Using a pileup analysis, gene product CG8177 (GenBank TM accession number AAF50207) is about 32% identical to NDAE1 and the NBCs, but about 34 -40% identical to the AEs. Though CG8177 apparently encodes a HCO 3 Ϫ transporter protein, future transport experiments will be needed to determine the actual function.
Expression Profile of NDAE1 mRNA-Next, we determined the location of NDAE1 mRNA in Drosophila. Using Northern blot analysis of poly(A) ϩ RNA, we were unable to detect NDAE1 mRNA in embryos, isolated adult heads, or body parts. However, by RT-PCR we could detect NDAE1 mRNA in heads as well as several embryonic stages (Fig. 2a). In situ hybridization to NDAE1 mRNA in whole mount Drosophila embryos (Fig. 2, b and c) illustrates that NDAE1 is present during embryogenesis. CNS staining is apparent throughout embryogenesis (Fig. 2, b and c). Staining of the gut primordium and mesoderm is evident in stage 6/7 (Fig. 2b). Staining of a specific subset of cells in the CNS is detectable by late embryogenesis (Fig. 2c) as is staining of the anal plate (not shown), i.e. the larval absorptive apparatus. The NDAE1-sense strand controls did not stain (Fig. 2, d and e). The difference between RT-PCR and in situ hybridization verses Northern detection of NDAE1 mRNA likely reflects sensitivity by amplification or individual cell mRNA abundance of the two former techniques.
Physiology of NDAE1 Expressed in Xenopus Oocytes-To evaluate the physiologic function of NDAE1, we expressed it in Xenopus oocytes. Fig. 3 is a model illustrating ion transport attributed to Na ϩ -dependent Cl-HCO 3 exchange activity. We tested this model with oocytes expressing NDAE1. Fig. 4a shows that removal and replacement of bath Na ϩ , Cl Ϫ , or both, with and without HCO 3 Ϫ does not alter pH i of a water-injected control cell. However, expression of NDAE1 elevates resting pH i by ϳ0.3 pH units (Fig. 4b), i.e. control ϭ 7.27 Ϯ 0.03 (n ϭ 9) and NDAE1 ϭ 7.54 Ϯ 0.03 (n ϭ 18). The acidification elicited by CO 2 /HCO 3 Ϫ (Fig. 4a) is markedly reduced in NDAE1 oocytes (Fig. 4b) and greatly increases intracellular [HCO 3 Ϫ ] 3 (control ϭ 3.1 Ϯ 0.2 mM, n ϭ 9; NDAE1 ϭ 7.4 Ϯ 0.4 mM, n ϭ 16). The higher resting pH i and elevated [HCO 3 Ϫ ] are consistent with NDAE1's role as an acid extruder, "forward" transport in Fig.  3a. Bath Na ϩ removal elicits a robust pH i decrease illustrating that NDAE1 is readily reversible (Fig. 3b). Subsequent removal of Cl Ϫ stops and slightly reverses the acidification, whereas readdition of Na ϩ in the sustained absence of Cl Ϫ triggers a rapid pH i recovery (Fig. 4b). A similar response is completely blocked by 200 M DIDS (Fig. 4g). Our results indicate that 3

[HCO 3
Ϫ ] is calculated using the pH i obtained just before CO 2 , steadystate pH i in the presence of CO 2 /HCO 3 Ϫ , and the Henderson-Hasselbalch equation (19,21).
N-glycosylation sites. c, dendrogram illustrating the percent divergence (1, 3) of sequences in a, other cloned NBCs, and AEs (AEs: human AE1, M27819; and human AE2, S21086). The NDAE1 protein is 43-47% and 32-33% identical to the NBCs and the AEs, respectively. d, putative membrane model of NDAE1 protein. Twelve TMs are predicted from the primary amino acid sequence using a Kyte-Doolittle algorithm (window size, 18 amino acids) and by comparison of similar NBC and AE areas. The NDAE1 topology is most easily fit by the proposed 12-14 TM models for the AEs (36 -39). A large exofacial loop is predicted at the TM 5 and 6 junction, containing two predicted N-linked glycosylation sites (Asn 600 and Asn 618 ). For the AEs, the last two hydrophobic regions are long enough to span the membrane twice (39), consistent with 12 TMs; others have suggested as many as 14 (39,40). Both the NH 2 and COOH termini are predicted to be intracellular as in the NBCs and AEs. Putative TMs are indicated by numbered rectangles. Predicted starts and stops of TMs are indicated by amino acid letter and number. A single predicted DIDS reaction motif is indicated as a diamond. This putative DIDS reaction site, illustrated as a molecular marker, is predicted intracellular, yet oocyte experiments with NDAE1 (Fig. 4, g-i) indicate that bath applied DIDS inhibits transport. Ser 205 , predicted intracellular, is the only consensus protein kinase A (PKA, triangle) site. Of the 13 consensus sites for protein kinase C (PKC, circles), 7 are predicted to be intracellular (Thr 89 , Ser 199 , Ser 200 , Thr 366 , Thr 410 , Thr 939 , and Thr 954 ), and 6 to be extracellular (Thr 466 , Thr 523 , Thr 692 , Ser 706 , Ser 797 , and Ser 859 ). Additionally, there is a leucine zipper motif (LeuZip, cylinder) in the NH 2 terminus at Leu 119 -Leu 140 .

FIG. 2. NDAE1 expression in Drosophila.
We designate the Drosophila gene ndae1 and the protein NDAE1. By in situ hybridization we localized ndae1 to region 28A on Drosophila polytene chromosome 2L (not shown). The predicted location inferred from the Drosophila genome sequence is 2L-27E6 (20). a, Southern blot illustrating RT-PCR of Drosophila tissues and rkNBC. NDAE1 gene-specific primers were used to amplify ϳ750-bp fragments from RT reactions of Drosophila stage and tissue poly(A) ϩ RNA, with NDAE1 and rkNBC included as positive and negative PCR controls, respectively. The male and female lanes are thorax without heads. The "control" lane is unrelated DNA and the "water" lane contained no template. Products are obvious in Drosophila embryos and tissues. Southern blotting showed that all of the ethidium bromide-stained NDAE1 bands hybridized authentic, biotin-labeled NDAE1 probes. b-c, in situ hybridization of NDAE1 RNA in Drosophila embryos. Drosophila embryos were fixed and probed for NDAE1 mRNA using a single-stranded, digoxigenin-labeled antisense NDAE1 RNA probe. In stage 6 embryos (b), staining was detectable in the cephalic furrow, gut primordium (arrow), all mesoderm, and some ectoderm. c, stage 16 -17 embryo. There was strong staining in a subset of CNS cells and possibly some peripheral nervous system. d and e, in situ hybridization using sense RNA probes at equivalent Drosophila stages. d (ϳstage 6) and e (ϳstage 16) revealed no staining.
FIG. 3. NDAE1 transport model. Schematic illustrating ion movements through NDAE1. This model does not imply paired binding or that the exact transport pathway via NDAE1 is known. a illustrates the direction of NDAE1 transport, indicated as forward, at steady-state in the normal oocyte Ringer, ND96. Our model indicates that in comparison to controls NDAE1 oocytes at steady-state should (i) have a higher pH i , (ii) have a higher aNa i , and (iii) have a lower aCl i . This forward transport is observed experimentally with addition of bath HCO 3 Ϫ or removal of bath Cl Ϫ . b shows the direction of the transported ions for bath removal, "reverse" transport, of either Na ϩ or HCO 3 Ϫ . This reverse transport should (i) decrease pH i , (ii) decrease aNa i , and (iii) increase aCl i . Even though "HCO 3 Ϫ " is shown in both models, NDAE1 does not require HCO 3 Ϫ to function (see Fig. 4d and legend, Fig. 5b, and "Results"). NDAE1 is indeed functionally unique in the BTS. These pH i changes are consistent with Na ϩ and HCO 3 Ϫ cotransport in exchange for Cl Ϫ and H ϩ as observed in snail neurons (7) and squid axons (6).
We further tested our transport model (Fig. 3) by meas-uring aCl i . Fig. 4c shows that a control oocyte has ϳ31 mM aCl i (37.0 Ϯ 1.6 mM, n ϭ 9), which only slightly changes with ion replacement Ϯ CO 2 /HCO 3 Ϫ . Fig. 4d illustrates that an oocyte expressing the NDAE1 transporter has ϳ22 mM aCl i (29.5 Ϯ 2.1 mM, n ϭ 6). NDAE1 oocytes show both rapid and robust responses to ion replacement and addition of CO 2 / HCO 3 Ϫ , i.e. changes of 3-8 mM activity (Fig. 4d). CO 2 /HCO 3 Ϫ supplied to the bath decreases aCl i , and Na ϩ removal reverses this response. With both Na ϩ and Cl Ϫ removed aCl i change stops, but readdition of Na ϩ elicits a large and rapid fall in aCl i . The removal of bath CO 2 /HCO 3 Ϫ brings aCl i back to resting levels (Fig. 4d). These alterations of aCl i are also blocked by 200 M DIDS (Fig. 4h). Moreover, the beginning of Fig. 4d illustrates that HCO 3 Ϫ is not required for ionic movements through the transporter (solutions bubbled with 100% O 2 ). This physiologic characteristic is reminiscent of the multiple transported anions (e.g. OH Ϫ , Br Ϫ , I Ϫ ), and HCO 3 Ϫ stimulated activity of the AEs.
To discriminate between Na ϩ dependence ("binding") versus Na ϩ driven (transport), we measured the effect of NDAE1 . Because the number of NDAE1 transporters in the oocyte membrane is unknown, it is unclear what percentage of NDAE1-mediated transport is accounted for by these currents. Nevertheless, as indicated in the voltage clamp section, the J(ion) for Cl Ϫ , HCO 3 Ϫ , and Na ϩ is at least 1000-fold greater than the J(current).
FIG. 6. P element insertion site is present in NDAE1 mRNA. a is a diagram of the NDAE1 mRNA, the UTRs, and the P element insertion. Forward primer "5ЈdrPE-f" is a sequence 5Ј to the P element insertion, whereas "3ЈdrPE-f" is a primer between the P element and the AUG start. "MAEK-r" is the reverse primer designed against the start of the NDAE1 open reading frame (the additional 5Ј-intron not predicted for cg4675). b is an ethidium bromide-stained agarose gel from a RT-PCR indicating the placement of the P element in the 5Ј-UTR of the NDAE1 mRNA; Ϫ addition, removal of Na ϩ , removal of Na ϩ and Cl Ϫ , and removal of Cl Ϫ . a, pH i of water injected (control) oocyte. Both Na ϩ and Cl Ϫ are removed Ϯ CO 2 /HCO 3 Ϫ . b, pH i of a NDAE1-injected oocyte. Similar experiment to a with a NDAE1-expressing oocyte. Starting pH i values for NDAE1-oocytes are ϳ0.3 pH units higher than controls as expected for a HCO 3 Ϫ influx transporter, i.e. an acid extruder. c, aCl i of a water-injected oocyte. Note that aCl i is minimally altered by bath solution manipulations. d, aCl i of a NDAE1-injected oocyte. Non-CO 2 /HCO 3 Ϫ solutions are bubbled with 100% O 2 , illustrating that NDAE1 does not require HCO 3 Ϫ to function. Starting aCl i s are ϳ10 mM less than control oocyte indicating basal Cl Ϫ extrusion from the NDAE1-oocytes. e, aNa i of a water-injected oocyte. The aNa i is unaltered by any of the bath solution manipulations. f, aNa i of a NDAE1-injected oocyte. The steady-state aNa i is elevated in comparison to the control oocyte. g-i illustrate DIDS inhibition of ion transport via NDAE1. g, DIDS inhibition of NDAE1-mediated pH i changes. The oocyte was exposed twice to CO 2 /HCO 3 Ϫ , first without DIDS (not shown) and second with 200 M DIDS. Exposure to DIDS appears to completely block NDAE1 activity, resulting in a response similar to control oocytes. h, DIDS inhibition of NDAE1-mediated aCl i changes, second pulse shown. i, using a double CO 2 /HCO 3 Ϫ protocol as in g, DIDS also blocks the aNa i changes. The hatched bar at the bottom right corner represents 10 min for that experiment. function on aNa i of oocytes. Fig. 4, e and f shows representative traces from control and NDAE1 oocyte experiments, respectively, using similar solution protocols as in Fig. 4, a-d. A control oocyte (Fig. 4e) has ϳ2.6 mM aNa i (3.1 Ϯ 0.5 mM, n ϭ 10), which does not change with bath ion substitutions. Fig.  4f shows that aNa i is increased to ϳ5 mM in NDAE1-expressing oocytes (4.6 Ϯ 0.3 mM, n ϭ 10). Na ϩ is transported by NDAE1 as evidenced by (i) increased aNa i with the addition of CO 2 / HCO 3 Ϫ , (ii) reduced aNa i with Na ϩ removal, and (iii) increased aNa i with Cl Ϫ removal. Na ϩ transport via NDAE1 is blocked by 200 M DIDS (Fig. 4i). Changes of aNa i are always in the opposite direction as aCl i changes indicating a Na ϩ for Cl Ϫ exchange. As shown for both the pH i and aCl i responses, Na ϩ transport was also observed in the complete absence of HCO 3 Ϫ (not shown). Thus, our data indicate that this Drosophila Na ϩdependent Cl-HCO 3 exchanger is more appropriately named a Na ϩ -driven anion exchanger or NDAE1.
We noted that Cl Ϫ removal or the addition of HCO 3 Ϫ resulted in significant depolarizations only in NDAE1 oocytes (Fig. 4b). Therefore, we voltage-clamped and used anion transport inhibitors (DIDS, diphenylamine carboxylic acid, and niflumic acid) to evaluate the electrical nature of NDAE1 (Fig. 5). In a voltage-clamped oocyte, this depolarization is measured as an inward (negative) current. A comparison of water-injected control (Fig. 5a) and NDAE1 oocytes (Fig. 5b) illustrates that both Cl Ϫ removal and HCO 3 Ϫ addition elicit current specific to NDAE1 expression. The reversal potential of both control (Fig. 5c) and NDAE1 oocytes (Fig. 5d) is about Ϫ20 mV. In the absence of Cl Ϫ , there is also a HCO 3 Ϫ -stimulated current only in NDAE1 oocytes (Fig. 5b). This current has a linear voltage dependence (Fig. 5d). DIDS, diphenylamine carboxylic acid, and niflumic acid block the depolarization (unclamped cell) because of Cl Ϫ removal (Fig. 5e). However, the measured currents in NDAE1 oocytes are small compared with the pH i , aCl i , and aNa i changes. The voltage deflections and associated currents are either endogenous to the oocyte uncovered by NDAE1 activity or more likely a "leak" current through the NDAE1 transporter. Present data imply that the NDAE1 current represents a leak current rather than NDAE1 being "electrogenic": (i) the J(ion)/J(current) ratio 4 for Cl Ϫ , HCO 3 Ϫ , and Na ϩ is Ͼ 1000; and (ii) the pH i changes are two to three times greater for NDAE1 than rkNBC, whereas the transport currents are at least 10-fold smaller (30 nA versus 300 -500 nA, respectively) (19). These transporter currents (or voltage changes) would not have been detectable in snail neurons (7) or squid axons (6).
Identification of a P Element Insertion in the NDAE1 Gene-To begin investigating the role of NDAE1 in vivo, we searched the Berkeley Drosophila genome Project (BDGP) data base with the NDAE1 sequence and identified a P element insertion mutation in the vicinity of ndae1. This insertion lies 408 bases 5Ј of the predicted NDAE1 initiation codon (Fig. 6a). By RT-PCR we determined that the insertion site of this P element mutation lies within the NDAE1 5Ј-untranslated sequence, using wild type poly(A) ϩ RNA and primers that flank the site of insertion (Fig. 6b). These data suggest that this P element disrupts ndae1. Because this P element insertion was isolated in a screen for lethal P element-induced mutations (12,22), our data further suggest that ndae1 may be essential for viability. Detailed phenotypic and physiological analysis of this mutant will be presented elsewhere. responsible for returning the food stream to a neutral or acidic pH (33). Thus, it is possible that NDAE1 could have a more widespread role in insect acid-base balance. From the current data, we hypothesize that NDAE1 may aid in maintaining the high gut pH of Drosophila and potentially mosquitoes. The gut and salivary gland pH of mosquitoes, in particular, is a factor contributing to transmission of disease such as encephalitis and malaria. That is, acidic environments (pH Ͻ 7) appear to promote cellular infection (34). The high pH of the gut likely protects the vector organism. Infectivity of Plasmodium berghei, present in mice, to mosquitoes is reduced with low pH and low blood HCO 3 Ϫ of the mice (35). A corollary hypothesis is that the mosquitoes' ability to spread disease could be controlled by altering the transport activity of NDAE1. With the cloning of NDAE1 these questions may now be addressed at the genetic, molecular, and physiologic levels.