Alkaline-shifted pHo Sensitivity of AE2c1-mediated Anion Exchange Reveals Novel Regulatory Determinants in the AE2 N-terminal Cytoplasmic Domain*

The mouse anion exchanger AE2/SLC4A2 Cl–/HCO–3 exchanger is essential to post-weaning life. AE2 polypeptides regulate pHi, chloride concentration, cell volume, and transepithelial ion transport in many tissues. Although the AE2a isoform has been extensively studied, the function and regulation of the other AE2 N-terminal variant mRNAs of mouse (AE2b1, AE2b2, AE2c1, and AE2c2) have not been examined. We now present an extended analysis of AE2 variant mRNA tissue distribution and function. We show in Xenopus oocytes that all AE2 variant polypeptides except AE2c2 mediated Cl– transport are subject to inhibition by acidic pHi and to activation by hypertonicity and NH+4. However, AE2c1 differs from AE2a, AE2b1, and AE2b2 in its alkaline-shifted pHo(50) (7.70 ± 0.11 versus 6.80 ± 0.05), suggesting the presence of a novel AE2a pH-sensitive regulatory site between amino acids 99 and 198. Initial N-terminal deletion mutagenesis restricted this site to the region between amino acids 120 and 150. Further analysis identified AE2a residues 127–129, 130–134, and 145–149 as jointly responsible for the difference in pHo(50) between AE2c1 and the longer AE2a, AE2b1, and AE2b2 polypeptides. Thus, AE2c1 exhibits a unique pHo sensitivity among the murine AE2 variant polypeptides, in addition to a unique tissue distribution. Physiological coexpression of AE2c1 with other AE2 variant polypeptides in the same cell should extend the range over which changing pHo can regulate AE2 transport activity.

Na ϩ -independent anion exchangers play a crucial role in the maintenance of intracellular and extracellular pH, cell volume, and chloride concentration by mediating electroneutral exchange of Cl Ϫ for HCO 3 Ϫ (1-3). Cl Ϫ /HCO 3 Ϫ exchangers are encoded by subsets of the Slc4 and Slc26 gene superfamilies. The SLC4 anion exchangers include the polypeptide products of at least three homologous genes: anion exchanger (AE) 3 1/Slc4a1, AE2/Slc4a2, and AE3/Slc4a3. AE1 polypeptides are expressed in erythrocytes and in Type A intercalated cells of the renal collecting duct. AE1 mutations are associated with hereditary spherocytic anemia and other erythrocyte dyscrasias and with dominant and recessive forms of renal tubular acidosis. The AE1 Ϫ/Ϫ mouse exhibits both severe hemolytic anemia (4) and distal renal tubular acidosis (5). AE2 and, to a lesser extent, AE3 transcripts are widely expressed in various tissues, including kidney, but these genes remain unlinked to monogenic human disease. Transcription of the mouse AE2/Slc4a2 gene generates at least five variant mRNAs: AE2a from the 5Ј-most promoter, AE2b1 and AE2b2 from promoter sequences within intron 2, and AE2c1 and AE2c2 from promoter sequences within intron 5 (6 -9). These five murine transcripts encode five AE2 polypeptides characterized by distinct N-terminal amino acid sequences (see Fig. 1). Expression of these alternate transcripts shows developmental and cell-type specificity (7). The AE2 Ϫ/Ϫ mouse fails to survive weaning and exhibits gastric achlorhydria and dental defects (10). However, a mouse lacking expression of AE2a, AE2b1, and AE2b2, but predicted to maintain expression of AE2c1 and AE2c2, exhibits male infertility with defective spermiogenesis but is otherwise grossly normal (11).
We have shown previously that the anion transport activity of the AE2a polypeptide is acutely regulated by several factors, including pH i and pH o (12)(13)(14)(15), hypertonicity (16,17), and NH 4 ϩ (17,18). However, functional analysis of the four other known AE2 variant polypeptides has not been reported. In this study, we characterize and compare the function and acute regulatory properties of all AE2 variant polypeptides. Most notably, we show that AE2c1 has an alkaline-shifted pH o dependence unique among the AE2 polypeptides, and we define at least two new sites within the N-terminal cytoplasmic region of AE2 whose absence can explain this novel pH o sensitivity. We also define the expression pattern of AE2 variant transcripts in an extended range of mouse tissues.

MATERIALS AND METHODS
Construction of Mouse AE2 (mAE2) Variant cDNAs-Plasmid p⌬X (19) was used as AE2a cDNA. cDNAs encoding alternate AE2 N-terminal variants were amplified by reverse transcription (RT)-PCR from total RNA preparations of stomach (AE2b1 and AE2b2 in 30 cycles and AE2c1 in 33 cycles) and kidney (AE2c2 in 35 cycles) using the Expand high fidelity PCR system (Roche Diagnostics) and a manual hot start procedure (8). The forward primer for AE2b2 was 5Ј-TCCCCCTTCT-TCTAGGTTCACC-3Ј (nucleotides (nt) Ϫ31 to Ϫ10) (6). Other forward primers and a common reverse primer from exon 8 were described previously (8), yielding the following PCR products: AE2a, 1022 bp; AE2b1, 974 bp; AE2b2, 1008 bp; AE2c1, 482 bp, and AE2c2, 768 bp. PCR fragments were purified from 1% agarose gels with a QIAquick gel extraction kit (Qiagen Inc.) and cloned into the "T-vector" pCRII (Invitrogen). Integrity of the cloned PCR amplification products was verified by sequencing. All mouse AE2 variant 5Ј-fragments were excised as EcoRI (vector-derived)-AvrII fragments and reconstructed with the remaining AE2 coding sequence in the oocyte expression vector pXT7 (20).
Construction of Mutant Mouse AE2 cDNAs-cDNAs encoding mAE2a N-terminal truncation mutants were generated by the methods described previously (12,13) with modifications. All N-terminal truncation mutants were constructed with a 10-nt native AE2a Kozak sequence preceding the initiator Met. The hexa-Ala substitution mutant (Ala 6 )124 -129, the triple-Ala substitution mutant (Ala 3 )124 -126, the triple mutant P143A/P145A/P149A, and the AE2b1 mutant E121V (numbering as in AE2a) were constructed by a four-primer method (12,13,21). Integrity of all PCR fragments and ligation sites was confirmed by DNA sequencing of both strands. 4 Tissue-specific Expression of AE2 Variant mRNA Transcripts as Detected by RT-PCR-Total RNA from freshly resected mouse stomach, duodenum, ileum, colon, liver, choroid plexus, heart, lung, uterus, and embryonic stem cells was prepared using an RNeasy kit (Qiagen Inc.). A mouse tissue total RNA panel for brain, embryo, kidney, ovary, spleen, testis, and thymus was purchased from Ambion, Inc. (Austin, TX). Reverse transcription was performed with a first strand cDNA synthesis kit (Ambion, Inc.) using 1 g of total RNA. 5% of the reaction volume was used for PCR with HotStart DNA polymerase (Qiagen Inc.) in a total reaction volume of 50 l in the supplier's recommended buffer. The complete reaction mixtures were incubated at 95°C for 15 min and then cycled through 45 s of denaturation at 94°C, 2 min of annealing at 60°C, and 2.5 min of elongation at 72°C. A 7-min final extension at 72°C was terminated by rapid cooling to room temperature. PCR products were separated on 1% agarose gels and visualized with ethidium bromide (GelDoc, Bio-Rad). Selected RT-PCR products of each AE2 variant transcript were purified and sequenced to confirm their molecular identities. PCR cycle number was adjusted for each tissue and for each AE2 variant mRNA, balancing the demands of low abundance transcript detection with the need to remain within or near the loglinear range of amplification for higher abundance transcripts. The upper limit for amplification was chosen as 38 cycles. Control PCR experiments using linearized purified AE2 cDNA templates in isolation and in an equimolar mixture showed that 1) amplification efficiencies of the forward primers unique to AE2a, AE2b1, and AE2b2 are similar, producing with their common reverse primer equivalent amounts of PCR products; and 2) the primer pair amplifying both the shorter AE2c1 and longer AE2c2 products underestimated AE2c2 abundance when in the presence of AE2c1. 5 RNase Protection Assay-The RNase protection assay (RPA) was conducted with the indicated amounts of total RNA (see Fig. 2E) from stomach or kidney using a PRAIII kit (Ambion, Inc.) with this modified hybridization step: the initial denaturation for 4 min at 95°C was followed by 10 cycles of consecutive 10-min incubations at 65, 60, 56, 52, 48, 45, 48, 52, 56, 60, and 65°C, with a final overnight incubation at 56°C. Reaction products were electrophoresed on a denaturing 7 M urea and 8% polyacrylamide gel, transferred to blotting paper, dried, and subjected to autoradiography.
cRNA Expression in Xenopus Oocytes-Capped mAE2 cRNAs were transcribed from linearized plasmids with a T7 MEGAscript kit and purified with an RNeasy mini kit (Qiagen Inc.). cRNA integrity was tested by formaldehyde-agarose gel electrophoresis, and cRNA concentration was estimated by the absorbance at 260 nm. Ovarian segments were excised from Xenopus laevis females anesthetized with 0.17% tricaine. The tissue was minced and incubated at room temperature for 1 h in 2 mg/ml Type A collagenase (Roche Diagnostics) in ND96 buffer, pH 7.4, containing 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 5 mM HEPES, and 2.5 mM sodium pyruvate and supplemented with 5 mg of gentamycin/100 ml. Stage V-VI oocytes were selected, manually defolliculated, and injected with 50 nl of cRNA or with H 2 O. The amount of cRNA injected per AE2 variant or mutant was titrated to approximate the activity of wild-type AE2a, producing efflux rate constants between 0.015 and 0.080 min Ϫ1 at pH 7.4. Injected oocytes were maintained at 19°C in ND96 buffer in the continued presence of gentamycin and pyruvate until used for assays. 36 (22). Data were acquired and analyzed with MetaFluor software (Universal Imaging, Chester, PA). dpH i /dt was measured from least-squares linear fits of initial slopes. EcR-293 cells stably expressing AE2b1 and AE2c1 (23) were grown on coverslips, incubated overnight in the absence or presence of the inducer ponasterone (5 M), loaded for 30 min with 5 M BCECF ace-toxymethyl ester, and mounted on a microscope stage. Cl Ϫ /HCO 3 Ϫ exchange was measured by BCECF fluorescence ratio imaging of 30 -70 subconfluent cells imaged in single ϫ20 visual fields during Cl Ϫ removal and restoration in the presence of 5% CO 2 and 24 mM HCO 3 Ϫ , with gluconate as the substituting anion. The AE2b2 and AE2c1 cDNAs subcloned into pcDNA3 were transiently transfected into HEK-293 cells on glass coverslips using Lipofectamine 2000 (Invitrogen). After 48 h, the transfected cells on coverslips were incubated for 30 min with 5 M BCECF acetoxymethyl ester and then mounted on a microscope stage. Cl Ϫ /HCO 3 Ϫ exchange was measured as described above in single visual fields containing 60 -90 cells. (AE2b2 was studied in transient transfectants because our stable EcR-293 cell lines had been made prior to publication of the AE2b2 sequence (6).) Immunodetection of mAE2 Variant Polypeptides-Groups of 10 oocytes were solubilized at 4°C in 1% Triton oocyte lysis buffer (10 l/oocyte) containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, and Complete protease inhibitor (Roche Diagnostics). After 30 min of intermittent shaking, the extract was centrifuged for 10 min at 4°C in a microcentrifuge. Clarified lysate was fractionated by SDS-8% PAGE, and proteins were transferred to nitrocellulose. After incubation with affinity-purified rabbit polyclonal antibody to the mAE2 C-terminal amino acids (aa) 1224 -1237 shared by all AE2 variant polypeptides (8,24), immobilized AE2 polypeptide was visualized on Eastman Kodak X-Omat film by enhanced chemiluminescence (PerkinElmer Life Sciences).
Rabbit polyclonal antibody to mAE2b2 aa 1-11 was raised against the high pressure liquid chromatography-purified, N-terminally acetylated, C-terminally amidated peptide MDFLLRPQPEP(C). The immunizing antigen was peptide-coupled through its added C-terminal amidated Cys residue to keyhole limpet hemocyanin via 3-maleimidobenzoic acid N-hydroxysuccinimide ester. Immune serum was affinity-purified over a 4-(N-maleimidomethyl)cyclohexanecarboxylic acid N-hydroxysuccinimide methyl ester-coupled peptide antigen column. HEK-293 cells transiently transfected with AE2b2 and AE2c1 cDNAs as described above were fixed in 2% paraformaldehyde and immunostained with the affinity-purified anti-AE2 antibodies in the presence of 12 g/ml peptide antigen or irrelevant peptide as described previously (8,23,24). Images were acquired with a Bio-Rad MRC1024 laser scanning confocal microscope.

RESULTS
Tissue Distribution of mAE2 Variant mRNA Expression-The five mAE2 variant transcripts encode the five predicted AE2 polypeptides schematized in Fig. 1. We examined the tissue distribution of the transcripts encoding these five polypeptides. AE2a mRNA was detected in all 17 tissues screened by RT-PCR (Fig. 2, A-D). AE2b1 and AE2b2 were also detected in all tissues analyzed, with the single exception of spleen. Stomach mRNA levels of AE2b1 appeared to be approximately equivalent to those of AE2a. Colonic levels of AE2b1 appeared to be slightly less abundant than of those of AE2a. The apparent relative expression levels of AE2a, AEb1, and AE2b2 in mouse kidney ( Fig. 2A) paralleled  and AE2c2, 768 nt. Uniformity of the RT reaction was confirmed for each tissue sample by RT-PCR of ␤-actin (23 cycles, not shown). E, RPA was performed with the indicated micrograms of total RNA from mouse stomach and kidney. Note that the level of AE2c mRNA (AE2c1 ϩ AE2c2) relative to that of AE2a ϩ AE2b is higher than apparent from band intensities since the signal strength of the randomly labeled probe must be evaluated according to the length of the protected fragments. ES cells, embryonic stem cells.
those detected by RT-PCR in rat kidney, in which most of the AE2b1 was from the inner stripe of the outer medulla. 6 Stomach was the only tissue studied that expressed substantial levels of AE2c1 mRNA. Trace AE2c1 expression was detected in liver, colon, choroid plexus, and uterus after 38 cycles of PCR with overexposure of gels (data not shown). AE2c2 expression was more widespread, with PCR products detected in all tissues examined except spleen, duodenum, brain, heart, and testis. 7 The AE2c2 mRNA abundance relative to AE2c1 in stomach (Fig. 2C) is likely an underestimate of the real value, as noted in control amplifications with this single AE2c primer pair (see "Materials and Methods").
The ability to detect AE2c transcripts by RPA was tested in mouse stomach and kidney (Fig. 2E). The structures of the AE2 variant transcripts do not permit design of a single RPA probe capable of detecting all five individual transcripts. Thus, the RPA probe was chosen to discriminate a single upper band representing combined AE2a, AE2b1, and AE2b2 and a single lower band of combined AE2c1 and AE2c2. The proportional signals detected in 10, 3, and 1 g of stomach RNA demonstrate the quantitative nature of the RPA. The ratio of gastric AE2a/ AE2b1/AE2b2 to AE2c1/AE2c2 detected by RPA (Fig. 2E) seemed lower than evidenced by RT-PCR (in which the differences are exponential) (Fig. 2, A and C). This difference highlights the potential pitfalls of comparing RT-PCRs not employing identical primer pairs. However, the kidney AE2c2 transcript clearly detected by a 38-cycle RT-PCR amplification ( Fig. 2A) was virtually undetectable by RPA (Fig. 2E), emphasizing the role of RT-PCR in high sensitivity transcript detection.
Four of the Five mAE2 Variant Polypeptides Are Functional in Xenopus Oocytes and in Mammalian Cells-All AE2 variant polypeptides ( Fig. 1) were expressed in Xenopus oocytes previously injected with equivalent amounts of cRNA. The extensively studied AE2a polypeptide accumulated to lower levels than did the other AE2 variant polypeptides (Fig. 3A). 36 Cl Ϫ efflux assays revealed that AE2b1 and AE2b2 mediated the highest rates of Cl Ϫ /Cl Ϫ exchange (Fig. 3B). AE2a and AE2c1 exhibited lower, roughly equivalent rates of anion transport. The AE2c1 polypeptide accumulated to higher levels compared with AE2a or any other AE2 variant. In contrast, AE2c2 mediated no detectable 36 Cl Ϫ efflux (Fig. 3B), despite polypeptide accumulation comparable with that of the most active variants, AE2b1 and AE2b2 (Fig. 3A). Therefore, AE2c2 was not examined further in these studies. AE2c1 expressed in Xenopus oocytes also mediated Cl Ϫ /HCO 3 Ϫ exchange (Fig. 3C), as was shown earlier for AE2a (25). We previously described immunodetection of muristerone-induced AE2b1 and AE2c1 overexpression in stably transfected EcR-293 cells (23). Fig. 4A shows immunodetection of AE2b2 in transiently transfected HEK-293 cells by antibodies to the common AE2 C-terminal peptide (upper panels) and to the AE2b2-specific N-terminal peptide (lower panels). Transient transfection of HEK-293 cells with cDNAs encoding AE2b2 or AE2c1 led to increased Cl Ϫ /HCO 3 Ϫ exchange activity (Fig. 4, B and C). In addition, Cl Ϫ /HCO 3 Ϫ exchange activity was induced by the muristerone analog ponasterone in clonal EcR-293 cell lines stably transfected with AE2b1 or AE2c1 (Fig. 4, D and E), consistent with the previously described parallel induction of heterologous AE2b1 and AE2c1 polypeptides (23).

Regulation of AE2 Variant Polypeptides by NH 4 ϩ and Hypertoni-
city-Murine AE2a, AE2b1, AE2b2, and AE2c1 were acutely regulated by NH 4 ϩ (Fig. 5, A and C) and hypertonicity (Fig. 5, B and C). 20 (13), 36 Cl Ϫ efflux rate constants for all active AE2 variants were strongly and reversibly decreased by intracellular acidification with bath butyrate (Fig. 6, A-C). Rate constants in the presence of 40 mM butyrate normalized to rate constants in the absence of butyrate were similar for AE2a and all active variants (Fig. 6C).
Regulation of Cl Ϫ /Cl Ϫ exchange rates by pH o revealed a novel property of AE2c1. Although AE2a, AE2b1, and AE2b2 demonstrated similar pH o(50) values (Fig. 6, D-F   AE2a (n ϭ 25)). This finding was unexpected in view of the wild-type pH o(50) values for AE2a N-terminal deletion mutants ⌬ N 99 and ⌬ N 310 (12,13). Moreover, all previous N-terminal cytoplasmic domain mutations in AE2 that altered pH o sensitivity shifted the pH o(50) to more acidic rather than more alkaline values (13)(14)(15).

Regulation of N-terminal Truncation Mutants by Intracellular and Extracellular pH-Because
AE2a mutant ⌬ N 99 exhibited a wild-type pH o(50) and because AE2c1 is equivalent to the nominal AE2a mutant ⌬ N 198, we hypothesized that the region of AE2a between aa 99 and 198 would include residues responsible for the novel alkaline-shifted pH o(50) for AE2c1. AE2a N-terminal truncation mutants ⌬ N 120, ⌬ N 150, and ⌬ N 175 (Fig. 7A) each exhibited sufficient 36 Cl Ϫ efflux activity (Fig. 7B) to allow tests of pH i and pH o sensitivity. All mutants exhibited a wild-type phenotype for inhibition upon pH i acidification by bath butyrate addition and activation upon intracellular alkalinization by butyrate removal (Fig. 7, C-E). The pH o(50) for AE2a mutant ⌬ N 120 (6.95 Ϯ 0.07, n ϭ 17) did not differ significantly from that for wild-type AE2a (6.80 Ϯ 0.05, n ϭ 25). In contrast, AE2a mutants ⌬ N 150 and ⌬ N 175 both demonstrated alkaline-shifted pH o(50) values of 7.76 Ϯ 0.08 (n ϭ 24) and 7.61 Ϯ 0.09 (n ϭ 17), respectively (Fig. 7, F-H), comparable in magnitude to the alkaline-shifted pH o(50) for AE2c1.
The difference in pH o(50) values between AE2a truncation mutants ⌬ N 120 and ⌬ N 150 focused attention on the region between aa 121 and 150. We therefore generated and studied the sequential AE2a deletion mutants ⌬ N 125, ⌬ N 130, ⌬ N 135, ⌬ N 140, and ⌬ N 145 (Fig. 8A). All these deletion mutants were active (Fig. 8B), and inhibition of Cl Ϫ /Cl Ϫ exchange by bath butyrate-induced intracellular acidification was similar in all deletion mutants tested (Fig. 8, C-E). However, inhibition was less pronounced than in the physiological AE2 variants (Fig. 6C) and in deletion mutants ⌬ N 120, ⌬ N 150, and ⌬ N 175 (Fig. 7E) 16); p Ͻ 0.01), although less than for mutant ⌬ N 150 or AE2c1 (Fig. 8, F-H). Therefore, two small, nearby, but nonadjacent regions of the AE2a N-terminal cytoplasmic domain encompassing aa 131-135 and 145-150 appear to be responsible for the majority of the difference between the pH o(50) values for AE2a and AE2c1.

Roles of Acidic and Pro-rich Sequences and of a Polymorphic Variant in Regulation of AE2 pH o Sensitivity by Amino Acids 121-150-The
AE2a region between aa 125 and 130 overlaps with an acidic sequence highly conserved among AE2 and AE3 polypeptides (Fig. 9A). To test the importance of this sequence to the alkaline-shifted pH o(50) for AE2a mutant ⌬ N 135 (and to that for AE2c1), we constructed the hexa-Ala substitution mutant (Ala 6 )124 -129 and the triple-Ala substitution mutant (Ala 3 )124 -126. The pH i sensitivity of mutant activity was unchanged (Fig. 9, C-E). However, (Ala 6 )124 -129 exhibited an alkaline-shifted pH o(50) similar to that for AE2 ⌬ N 135, whereas the pH o sensitivity of (Ala 3 )124 -126 resembled that of wild-type AE2a (Fig. 9,  F-H). Thus, residues 127-129 appear to contribute to the alkalineshifted pH o sensitivity of AE2c1 and replicate the moderate alkalineshifted pH o(50) for deletion mutant ⌬ N 135.
AE2a aa 145-150 overlap with a highly conserved proline-rich sequence fitting the consensus for a glycogen synthase kinase-3 phosphorylation site at either Ser 144 or Thr 148 (Fig. 9A). The importance of this region was addressed by the construction and study of the AE2a triple mutant P143A/P145A/P149A. Fig. 9 (C-E) demonstrates that pH i regulation of these mutants did not differ from that of AE2a. Ala sub-stitution of all three Pro residues did not replicate either the moderate alkaline shift of AE2 mutant ⌬ N 135 or the greater alkaline shifts of mutant ⌬ N 150 and AE2c1 (Fig. 9H). These data do not support a required role for this proline-rich region in controlling AE2 pH o sensitivity.
The human AE2 polymorphism E122V corresponds to mAE2 E121V within the AE2a region of aa 121-150 implicated in regulation of pH o sensitivity. This non-conserved polymorphism replaces negatively ϩ -stimulated 36 Cl Ϫ efflux rate constants of oocytes expressing mAE2 N-terminal variants AE2a, AE2b1, AE2b2, and AE2c1; B, 36 Cl Ϫ efflux rate constants of oocytes expressing mAE2 N-terminal variant polypeptides exposed sequentially to isotonic (210 mosM) and hypertonic (362 mosM) baths; C, -fold stimulation of 36 Cl Ϫ efflux by NH 4 ϩ and hypertonicity as calculated from rate constants in A and B. The data are the means Ϯ S.E. for n oocytes.
charged Glu with uncharged Val. The functional consequence of this substitution was tested in the background of mAE2b1. Neither pH i sensitivity nor pH o sensitivity of AE2b1 was altered by the E121V substitution (Figs. 6, E and F; and 10, A-D), suggesting that this polymorphism likely does not contribute to the pH sensitivity of human AE2.

DISCUSSION
The widely expressed mAE2 gene is the source of at least five defined transcripts arising from alternative promoter usage. The AE2a transcript encoding the longest AE2 polypeptide (26,27) has until now been the only functionally characterized AE2 polypeptide (12-18, 28, 29). The discovery of AE2b1, AE2c1, and AE2c2 transcripts in rat (9) was followed by detection of AE2c2 in mouse (8) and the discovery of AE2b2 in human and mouse (6,30). The functional importance of AE2c1 and AE2c2 was suggested by two phenotypically very different AE2 knockout mice. Elimination of all AE2 gene expression leads to a severe growth retardation phenotype and peri-weaning death (10). In contrast, mice lacking AE2a, AEb1, and AEb2, but not expected to lack AE2c1 or AEc2, exhibit an apparently selective defect in spermiogenesis, leading to male infertility with an otherwise grossly normal phenotype (11). However, the anion transport function of the AE2 gene products transcribed from the promoters downstream of the AE2a promoter has remained until now uncharacterized. The current study has evaluated the expression, function, and regulation of mouse AE2b1, AE2b2, AE2c1, and AE2c2 and compared them with the well studied AE2a. The results of this comparison include the first reported difference in acute regulation among the polypeptide products of the five known variant transcripts of rodent AE2.
AE2c2 Function in Xenopus Oocytes Is below the Level of Detection-AE2a, AE2b1, AE2b2, and AE2c1 each mediated Cl Ϫ /Cl Ϫ exchange in Xenopus oocytes (Fig. 3). The level of transport function did not correlate with the level of polypeptide accumulation. AE2c1 was also a functional Cl Ϫ /HCO 3 Ϫ exchanger in both oocytes and mammalian cells, and AE2b1 and AE2b2 mediated Cl Ϫ /HCO 3 Ϫ exchange in mammalian cells. AE2c2 did not exhibit detectable Cl Ϫ /Cl Ϫ exchange function in oocytes previously injected with 10 ng of cRNA. This lack of function was not due to the absence of AE2c2 polypeptide accumulation because the abundance of AE2c2 was equivalent to that of the highly active AE2b1 and AE2b2. The oocyte surface expression of AE2c2 compared with that of the functional AE2 variant polypeptides remains to be examined. The minimal functional activity of mAE2c2 in oocytes is particularly interesting insofar as rat AE2c2 mRNA does not encode an AE2c2 N-terminally elongated polypeptide. Instead, rat AE2c2 and AE2c1 mRNAs both encode the single rat AE2c1 polypeptide (8,9,24). values indistinguishable from that for AE2a, the pH o(50) for AE2c1 was shifted to alkaline values by almost 1 pH unit (Fig. 6) (12) and ⌬ N 310 (13) were known to be indistinguishable from that for wild-type AE2a. Therefore, the AE2a sequence whose absence explains the distinct AE2c1 pH o(50) was sought between AE2a aa 100 and 198. The initial N-terminal deletion mutants studied confirmed the existence of such sequence between AE2a aa 121 and 150 (Fig. 7). Determinants of the Novel pH o Regulatory Structure Comprise at Least Two Noncontiguous but Adjacent Sequences-The difference in pH o sensitivity between AE2c1 and other functional AE2 polypeptides was investigated more intensively by study of deletion and substitution mutants between aa 121 and 150 of the AE2 N-terminal cytoplasmic domain. Functional analysis of deletion mutants suggested at least two regions of importance, between aa 125 and 135 and between aa 145 and 150 (Fig. 8). The former region was notable for the highly conserved polyacidic stretch of AE2a aa 124 -129. The ⌬ N 135 phenotype was replicated by substitution mutant (Ala 6 )124 -129, but not by mutant (Ala 3 )124 -126, confirming a contribution of residues 127-129 to AE2 regulation by pH o (Fig. 9). This acidic stretch might contribute to a predicted "moderate stringency" consensus phosphorylation site for casein kinase-1 or -2 at Ser 131 (available at www.scansite.mit.edu). The mAE2 E122V equivalent of the common human single-nucleotide coding polymorphism E121V was shown not to modify AE2 regulation by pH o or pH i (Fig. 10).
Near and within aa 145-150 are three highly conserved Pro residues that constitute part of a predicted consensus Src homology-3 domainbinding site for phospholipase-␥ and that contribute to a predicted moderate stringency consensus phosphorylation site for glycogen synthase kinase-3 at either Ser 144 or Thr 148 (available at www.scansite.mit. edu). However, the AE2 triple-Ala substitution mutant P143A/P145A/ P149A did not suffice to shift pH o(50) to a more alkaline value (Fig. 9). Thus, the mechanism by which residues 145-150 contribute to regulation of AE2 pH o sensitivity remains to be determined. The entire difference in pH o(50) values between AE2a/AE2b1/AE2b2 and AE2c1 can be plausibly explained by the additive (but not obligatorily interactive) contributions of AE2a aa 127-135 and 146 -150, perhaps with additional residues immediately adjacent.
The region of AE2a aa 121-150 has no correspondent within that portion of the AE1/SLC4A1 N-terminal cytoplasmic domain for which a 3.4-Å x-ray crystal structure is available. The secondary structure of aa 121-150 is predicted to include an ␣-helical stretch in its middle portion, but 4 of the 14 following residues are prolines. The degree of physiological interaction of the pH o regulatory residues within this region with the previously described pH o regulatory residues within aa 336 -347 (14) and in adjacent stretches (15) remains to be determined. Also to be determined is how regulation of pH o(50) by the novel sites identified in this work interacts with the independent inhibitory regulation by intracellular protons and the independent activating stimuli of NH 4 ϩ and of hypertonicity, each regulated by other residues of AE2.

Implications of the Alkaline-shifted pH o(50) of AE2c1 for Other Determinants of pH o Sensitivity within the Amino Acid Sequence Unique to
AE2a-Previous mutational structure-function studies of mAE2 have generated many mutants that exhibit an "acid-shifted pH o(50) value" for Cl Ϫ /Cl Ϫ exchange. The AE2c1 polypeptide represents the first instance of an alkaline-shifted pH o(50) value (compared with AE2a) in a physiological AE2 isoform (Fig. 6). This property was surprising in view of the unchanged pH o(50) values for the AE2a N-terminal truncation mutants ⌬ N 99 and ⌬ N 310 (12,13). Moreover, further truncation, as well as indi- vidual missense mutations beyond aa 310, also acid-shifted the pH o(50) for Cl Ϫ /Cl Ϫ exchange (13)(14)(15). These data suggest that investigation of AE2a residues between positions 199 and 310 will uncover groups of residues whose mutation or removal will revert the alkaline-shifted pH o(50) for AE2c1 back to that shared by wild-type AE2a and ⌬ N 310, as suggested by preliminary experiments. 8 The relationship between these regions and the highly conserved aa 336 -347, mutations of which acidshift the pH o(50) for AE2 (14), remains unknown. The properties of the multiple AE2 N-terminal cytoplasmic domain mutants and natural variants studied to date suggest a complex folded structure in which many mutations can alter presumed regulatory interaction with the transmembrane domain to produce opposing shifts in the pH o sensitiv-ity of anion transport. Although the first half of this 705-aa AE2 N-terminal cytoplasmic domain is of unknown structure, we have previously modeled the domain's second half (14,15) on the pH 4.8 crystal structure of the corresponding AE1 domain (31). However, this modeling should be interpreted cautiously, as recent lanthanide resonance energy transfer data suggest a considerably more elongated native structure of the AE1 N-terminal cytoplasmic domain than indicated by the crystal structure (32).
Physiological Role of AE2c1 Expression in Stomach-The restriction of AE2c1 expression to stomach is appropriate if AE2c1 is expressed in parietal cells, where it would be exposed to high [HCO 3 Ϫ ] in the basolateral interstitial space. In an elutriation-enriched fraction of rabbit parietal cells, AE2c transcripts are indeed present, and semiquantitative RT-PCR suggested that proportional AE2 mRNA levels are AE2b Ͼ 8 C. E. Kurschat, A. K. Stewart, and S. L. Alper, unpublished data.