The cytoplasmic and transmembrane domains of AE2 both contribute to regulation of anion exchange by pH.

We have compared regulation by pH of AE1 (band 3)- and AE2-mediated Cl uptake into Xenopus oocytes. Cl influx was assayed at varying extracellular pH (pH) values between 9.0 and 5.0 under conditions in which corresponding intracellular pH (pH) values were at or near steady-state. Wild type (WT) AE1 displayed a broad convex pH versus activity curve, with peak activity at pH 7.0 and 63% of maximal activity at pH 5.0. In contrast, WT AE2 displayed a steep pH versus activity curve, with peak activity at pH9.0 and full suppression at pH 5.0. The structural basis of these differing pH sensitivities was examined by expression of cRNAs encoding chimeric and truncated proteins. Mutant polypeptides were expressed in oocytes and detected at the cell surface. The AE2/AE1 polypeptide displayed a broad pH versus activity curve similar to that of WT AE1. In contrast, the AE1/AE2 polypeptide displayed a steep pH versus activity curve, which was shifted toward acid pH values from that of WT AE2 by 0.69 ± 0.04 pH units. Moreover, whereas the pH versus activity curves of AE2 Δ99 and WT AE2 were indistinguishable, AE2 Δ510 exhibited a pH versus activity curve acid-shifted from that of WT AE2 by 0.66 ± 0.13 pH units (indistinguishable from that of AE1/AE2). The data suggest that a pH sensor resides within the transmembrane region of AE2. The affinity for protons of this pH sensor is influenced by a modifier site located between residues 99 and 510 of the N-terminal cytoplasmic domain of AE2. Acidification of oocytes with acetate suggested that pH accounted for some but not all of the measured pH dependence of AE2.

The band 3-related AE anion exchanger gene family comprises at least three genes, each of which encodes polypeptides that mediate anion exchange when expressed in heterologous expression systems (reviewed in Ref. 1). The prototype red cell band 3 (AE1) serves both as a chloride/HCO 3 exchange mechanism and as a major membrane anchor for the spectrin/ ankyrin/actin cytoskeleton (2). The polypeptide products of the AE2 and AE3 genes are postulated to serve similar roles (1). Their polytopic transmembrane domains of ϳ530 C-terminal amino acids suffice to mediate anion exchange (3)(4)(5). Their N-terminal cytoplasmic domains of ϳ700 amino acids in length may, by analogy with red cell AE1, bind to cytoskeleton proteins of nonerythroid cells (6). Chloride/HCO 3 exchange contributes widely to maintenance of cellular pH, to secondary active chloride loading, and to transepithelial movement of chloride and bicarbonate (1).
Whereas chloride/monovalent anion exchange by red cell AE1 displays a broad pH versus activity profile (7), which serves to maintain relatively constant activity across the pH range of the capillary from its beginning to the end of its course through CO 2 -and acid-generating tissues, chloride/HCO 3 exchange and 36 Cl Ϫ influxes measured in tissue culture cells display a steep pH dependence (8 -10). Tissue culture cells, as well as most nonerythroid cell types, generally express AE2 and/or AE3 ion exchangers rather than AE1 (1,(11)(12)(13)(14)(15)(16)(17)(18). The steep pH dependence of the nonerythroid anion exchangers has led to the proposal that they contribute not only to cellular defense against alkaline loads, but also in some tissues to the maintenance of resting pH i (1,19) as well as to maintenance of resting [Cl] i (20). However, when the pH dependence of AE2mediated chloride/HCO 3 exchange was tested in transiently transfected mammalian cells (21,22), in infected insect cells (23), or in intact gastric glands (24), activation by alkaline pH i (when noted) required alkalinization beyond 0.25 pH units. These findings suggested that AE2 might be inadequately sensitive to intracellular alkalinization to play a physiological role in the regulation of resting pH i . In contrast, AE2 expressed in Xenopus oocytes was activated by exposure to alkaline media and inhibited by exposure to acidic media (25,26). This pH dependence of AE2-mediated chloride/base exchange contrasted with AE1-mediated chloride transport which, in preliminary experiments, showed considerably less sensitivity to pH (46).
Heterologous expression of AE1 in mammalian cells has to date not allowed study of AE1 transport function at the cell surface (27). As with AE1, AE1/AE2 chimeric polypeptides also did not traffic to the cell surface. Study of microsomes from transfected cells (27) and of proteoliposomes prepared by reconstitution of these microsomes in the presence of exogenous phospholipid (28) has allowed comparison of sulfate fluxes mediated by heterologous AE1 and AE2. However, the sulfate/ anion exchange measured in these studies is activated by acid pH, consistent with the proton/sulfate cotransport earlier described in red cells (29). The alkaline activation (or acid suppression) of AE-mediated chloride/base exchange in these vesicle systems has not been studied with the requisite time resolution. In contrast, AE1 is expressed at the surface of the Xenopus oocyte (30,31) at least as efficiently as AE2 (25). Thus, the oocyte lends itself to the comparative study of the regulation of heterologous AE gene products in an intact cell system.
We have compared the differences in pH dependence of chlo-ride/base exchange mediated by AE1 and by AE2 in Xenopus oocytes. The transmembrane domains of AE1 and AE2 are 65% identical in amino acid sequence. In contrast, the overlapping regions of N-terminal cytoplasmic domain are only 33% identical, and the 250 N-terminal amino acids of AE2 correspond to none in AE1. Inspection of the sequences suggests the possibility that the greater differences in the N-terminal cytoplasmic domains of AE1 and AE2 might explain the associated differences in pH sensitivity of chloride transport. However, analysis of the Na ϩ /H ϩ exchanger, NHE1, provided evidence for a regulatory pH sensor in the transmembrane domain of the polypeptide (32). In addition, these authors localized a modulatory function to the cytoplasmic domain of NHE1. In order to define and localize structures in the AE2 polypeptide that mediate regulation of transport activity by pH, we expressed chimeric and truncated AE anion exchangers in Xenopus oocytes and assayed Cl Ϫ transport across an extracellular pH range from 5.0 to 9.0. Our data suggest a model in which the transmembrane domain of AE2 contains a pH sensor whose affinity for protons is modulated by a structure residing between amino acids 99 and 510 of the AE2 N-terminal cytoplasmic domain.

EXPERIMENTAL PROCEDURES
Materials-Female Xenopus were purchased from NASCO (Madison, WI) and maintained as described previously (25). 2Ј,7Ј-bis(2-carboxyethyl)-5(6)-carboxyfluorescein acetoxymethylester (BCECF-AM) 1 was from Molecular Probes. Bumetanide was the gift of P. Feit (Leo Pharmaceutical). Na 36 Cl was purchased from ICN (Irvine, CA). All other chemical reagents were of analytical grade and purchased from Sigma, Calbiochem, or Fluka. Restriction enzymes and T4 DNA ligase were purchased from New England BioLabs (Beverly, MA). Taq DNA polymerase and dNTPs were purchased from Promega (Madison, WI). Sure TM Escherichia coli strain was from Stratagene (La Jolla, CA). Oligonucleotides were synthesized on a Milligen Cyclone Plus DNA synthesizer.
Solutions-ND-96 medium consisted of (in mM) 96 NaCl, 2 KCl, 1.8 CaCl 2 , 1 MgCl 2 , 5 HEPES, and 2.5 sodium pyruvate. All flux media lacked pyruvate and contained 38 mM NaCl as Na 36 Cl. Some flux media substituted equimolar MES or bis-tris propane for HEPES. Flux media were titrated to the indicated pH values before use.
PCR Products and cDNA Construction-Murine AE1 encoded in plasmid pBL (26) and murine AE2 encoded in plasmid p⌬X (11,25,26) were used as templates for polymerase chain reaction (PCR) and for RNA transcription. Chimeric cDNAs were constructed by a four-primer PCR method employing two rounds of PCR amplification (33). The AE1 cyto /AE2 memb chimera was constructed as follows (Table I). AE1 served as the first-round template for primers yz5 and yz2. AE2 served as the first-round template for primers yz1 and yz8. The two first-round amplification products were coannealed to serve as second-round template for the bracketing primers yz5 and yz8. The second-round chimeric amplification product joined AE1 nucleotide 1395 to AE2 nucleotide 2292 in a fragment of 841 nucleotides in length. This PCR product was completely digested by NcoI and partially digested by StuI, and the central fragment of 543 base pairs was gel-purified. To reconstruct the full-length chimera, this fragment underwent three-way ligation with the 3.8-kb NcoI/HindIII fragment of pBL (AE1) and the 1.6-kb StuI/ HindIII fragment of p⌬X (AE2). The final AE1 cyto /AE2 memb cDNA encoded mouse AE1 cDNA from its translational initiation codon to nucleotide 1395 (corresponding to amino acids 1-422) fused to mouse AE2 cDNA from nucleotide 2292 to its 3Ј end (corresponding to amino acids 704-1237, the entire transmembrane (TM)-spanning region beginning at TM1).
The AE2 cyto /AE1 memb chimera was constructed as follows (Table I). The AE2 plasmid p⌬X served as first-round PCR template for primers yz7 and yz4, and the AE1 plasmid pBL served as first-round template for primers yz3 and yz9. The two first-round amplification products were coannealed to serve as second-round template for the bracketing primers yz7 and yz9. The second-round chimeric amplification product joined AE2 nucleotide 2291 to AE1 nucleotide 1396 in a fragment of 1.7 kb in length. This PCR product was digested by NaeI and SphI, and the central fragment of 1.4 kb was gel-purified. To reconstruct the fulllength chimera, a plasmid containing a tandem head-to-tail insertion of AE2 and AE1 underwent partial digestion with NaeI and complete digestion with SphI to yield a 6.3-kb NaeI/SphI fragment, which was also gel-purified. The 1.4-kb PCR fragment and the 6.3-kb cDNA fragment were ligated to construct AE2 cyto /AE1 memb . Sequence analysis of the amplified plasmid demonstrated a missense mutation in the AE1 portion of the PCR product. This mutant cDNA was partially digested with BsmI and completely digested with HindIII. The resultant 5-kb fragment was ligated with the 2.5-kb BsmI/HindIII fragment from the AE1 plasmid, pBL. The final AE2 cyto /AE1 memb cDNA encoded AE2 nucleotides 105-2291 (corresponding to amino acids 1-703) fused to the AE1 sequence from nucleotide 1396 to its 3Ј end (corresponding to amino acids 423-929, the entire transmembrane-spanning region beginning at TM1).
The AE2 truncation mutants ⌬99 and ⌬510 were constructed as follows ( Table I). The AE2 PCR product generated from primers yz13 and sz8 was digested with SacII and AvrII and ligated to the large SacII/AvrII fragment of the AE2 plasmid p⌬X. The reconstructed AE2⌬99 cDNA encoded an engineered ATG followed by amino acids 100-1237. The AE2 PCR product generated from primers yz11 and yz12 was digested with SacII and SmaI and ligated to the large SacII/SmaI digestion product of p⌬X. The reconstructed AE2⌬510 encoded an engineered ATG followed by amino acids 511-1237.
DNA Sequence Analysis-Sequencing reactions were carried out using the DyePrimer cycle sequencing kit (Applied Biosystems), and the reaction products were sequenced on the Applied Biosystems model 371 automated sequenator. PCR-amplified fragments and their ligation junctions in all reconstructed cDNAs were completely sequenced on both strands to verify absence of introduced mutations.
In Vitro Transcription and Translation-cDNAs encoding WT AE1 and AE2 cyto /AE1 memb were linearized with HindIII. All other cDNAs were linearized with ClaI. Capped cRNA transcripts were synthesized from these templates with the MEGAscript kit (Ambion, Austin, TX), and resuspended in diethylpyrocarbonate-treated water. RNA integrity was confirmed by formaldehyde gel electrophoresis, and concentration was determined by A 260 . In vitro translation of cRNAs in the absence and presence of dog pancreas microsomes was with rabbit reticulocyte lysate (Promega).
Antibodies-Affinity-purified polyclonal antibody to the 12 carboxylterminal amino acids of murine AE2 has been described (16,17). Murine monoclonal antibody to the 12 carboxyl-terminal amino acids, 1224 -1237, of murine AE2 2 was purified by fast protein liquid chromatography (Pharmacia Biotech Inc.) with protein A-agarose affinity chromatography. Polyclonal antiserum to the 12 carboxyl-terminal amino acids of murine AE1 was the gift of Dr. A.-M. Garcia.
Immunoprecipitation of Total and Surface AE Proteins from Xenopus Oocytes-Groups of 10 -12 oocytes, injected either with 20 ng of cRNA or with water, were incubated for 48 h at 19°C in 200 l of ND-96, pH 7.4, containing 1 mCi/ml (30 M) of 35 S-methionine. Metabolically labeled oocytes were washed in modified ND-96, pH 8.0, in which 48 mM NaCl was replaced with 100 mM sucrose, and then incubated for 3 h at room temperature in the same medium in the presence or absence of 5 mg/ml papain. Subsequent washing, homogenization in Triton X-100, immunoprecipitation, and SDS-polyacrylamide gel electrophoresis fluorography were as described previously (31). 36 Cl Ϫ Influx Measurements-Defolliculated oocytes were injected with cRNA or water, and incubated in ND-96 for 2-4 days at 19°C. Groups of 8 -12 injected oocytes were preincubated for 15 min in 1 ml of ND-96 at the indicated pH and then transferred into 150 l of the same medium containing in addition 10 M bumetanide and 2.8 -3.2 Ci of Na 36 Cl. Total NaCl concentration was constant at 96 mM (38 mM of which was from Na 36 Cl). Two 10-l aliquots of mixed influx medium were removed for later determinations of 36 Cl-specific activity in each influx group. 36 Cl Ϫ influx into oocytes was carried out for 15 or 60 min in the absence or presence, respectively, of 100 M amiloride and then terminated by three washes in 25 ml of isotope-free ND-96. Each construct was tested under both influx conditions, with similar results. Some oocytes were acidified by acetate exposure. The incubation medium for these experiments was a modified ND-96 in which 38 mM NaCl was substituted with 38 mM sodium isethionate. Oocytes were preincubated for 30 min in this modified medium or in solutions in which sodium acetate was substituted for equimolar sodium isethionate. 36 Cl Ϫ influx was then performed as above.
Individual oocytes underwent scintillation counting of associated 36 Cl Ϫ . AE-mediated 36 Cl Ϫ influx was calculated by subtracting values for Cl Ϫ influx into water-injected oocytes subjected to identical flux conditions as part of each experiment. Every experiment that tested chimeric constructs compared the chimeras with both WT AE1 and WT AE2 in the same experiment. Every experiment that tested AE2 truncation constructs compared the mutants with WT AE2 in the same experiment. Each construct was tested in oocytes from at least four frogs, using multiple RNA preparations.
AE-mediated Cl Ϫ influx values determined in flux media of pH o 5.0 -9.0 were plotted as a function of pH o . Mean values measured for WT AE2, AE1 cyto /AE2 memb , AE2⌬99, and AE2⌬510 in each individual experiment were fit to the following first-order logistic sigmoid equation  (34) and by two-tailed t tests (Microsoft Excel, version 5.0 for Macintosh). Results did not differ when influx assays were stratified according to performance of 15-or 60-min uptakes (n ϭ 7 and n ϭ 6, respec-tively, for WT AE2). Results also did not differ when influx assays were performed at 4 or 5 pH o values per experiment (n ϭ 4 and n ϭ 9, respectively for WT AE2). Therefore, pooled results were analyzed.
Mean values measured for WT AE1 and for AE2 cyto /AE1 memb were also pooled by normalization to V max for each individual experiment. The transport values for these two polypeptides were not well fit by linear or logistic sigmoid functions and were plotted by hand.
Measurement of Oocyte Intracellular pH (pH i )-Oocyte pH i was measured by BCECF fluorescence ratio imaging as described previously (25,26). Oocytes were loaded with 2-5 M BCECF-AM for 1 h and mounted in a customized superfusion chamber mounted on an inverted microscope stage (25). Oocytes were irradiated, and fluorescence ratios were acquired and recorded to hard disk with an Image 1 digital ratio imaging system (Universal Imaging, West Chester, PA). The plane of focus of ϳ20 m (ϫ10 objective, N.A. 0.4) was set under the vegetal pole surface (25) for the measurements presented or at different focal planes (26) with similar results. In situ calibration of the BCECF fluorescence ratio (35,36) was performed as described (25). RESULTS cDNA Constructs of AE Mutations-In order to localize those protein domains within AE2 and AE1 that are necessary for expression of their different sensitivities to pH, domain-swap chimeras and truncated AE2 cDNAs were generated as described under "Experimental Procedures" and in Table I. The constructs are presented in schematic form in Fig. 1. The AE1 cyto /AE2 memb cDNA encoded a chimeric polypeptide consisting of mouse AE1 cytoplasmic amino acids 1-422 fused to mouse AE2 TM domain amino acids 704-1237. The AE2 cyto AE1 memb cDNA encoded a chimeric polypeptide consisting of mouse AE2 cytoplasmic amino acids 1-703 fused to mouse AE1 TM domain amino acids 423-929. AE2⌬99 cDNA encoded an engineered ATG followed by amino acids 100-1237. AE2⌬510 encoded an engineered ATG followed by amino acids 511-1237.
Biosynthesis and Surface Expression of WT and Mutant AE Polypeptides-cRNAs transcribed from the above described cDNAs were used as templates for in vitro translation reactions in reticulocyte lysate. Polypeptides of predicted size were translated from each cRNA. Each polypeptide was inserted into microsomal membranes and underwent an increase in M r and bandwidth suggestive of core N-glycosylation (not shown). Thus, the mutant cRNAs encoded polypeptides of the predicted length that folded in such a way as to permit core glycosylation.
Polypeptide biosynthesis was next analyzed in Xenopus oocytes (Fig. 2). Oocytes injected with WT AE2 cRNA produced polypeptides of 135 (core-glycosylated) and 160 kDa (complexglycosylated) as shown previously (25), while WT AE1 cRNA produced the typical broad band at ϳ100 kDa, which contains within it both core-and complex-glycosylated species (30,31). All constructs that contained the AE2 TM domain exhibited the expected two polypeptide bands, whereas all those containing the AE1 TM domain displayed the expected single band. The truncated AE2 ⌬99 cRNA produced polypeptides of 125 and 150 kDa, and ⌬510 cRNA produced polypeptides of ϳ80 and 115 kDa. Chimeric AE1 cyto /AE2 memb cRNA produced polypeptides of 110 and ϳ135 kDa. Chimeric AE2 cyto /AE1 memb cRNA produced a single polypeptide of ϳ130 kDa, which accumulated to the lowest level among the AE mutants tested.
The presence of these heterologous AE polypeptides at the oocyte surface was assessed by incubation of oocytes in the absence or presence of 5 mg/ml papain for 3 h at room temperature. Papain digestion of surface-exposed WT AE2 in Xenopus oocytes produced a carboxyl-terminal AE2 fragment of 32 kDa (Fig. 2). Accumulation of the 32-kDa fragment was maximal at 3 h. The same fragment was produced by papain digestion of porcine basolateral gastric microsomes, 2 in which the amino acid sequence of the C-terminal fragment defined a 35-kDa fragment cleaved within the nonglycosylated ectoplasmic loop linking transmembrane spans 7 and 8. This papain treatment protocol did not increase 36 Cl Ϫ influx in water-injected oocytes (not shown). Thus, papain digestion provided biochemical evidence for the presence at the oocyte surface of polypeptides encoding WT AE2, AE2⌬99, AE2⌬510, and AE1 cyto /AE2 memb .
The amount of AE2 cyto /AE1 memb at the oocyte surface was below the threshold of detection by polyclonal antibody to the carboxyl-terminal 12 amino acids of murine AE1. The corresponding carboxyl-terminal ϳ35-kDa fragment of human AE1 produced by papain digestion of intact human red cell ghosts is barely detectable as a broad, very faint smear by Coomassie Blue stain (37). Although this fragment of WT human AE1 has been immunoprecipitated from metabolically labeled Xenopus oocytes with a particular monoclonal antibody (30), the comparable fragment from mouse AE1 has not yet been detected in Xenopus oocyte lysates. Nonetheless, expression of both WT mouse AE1 and AE2 cyto /AE1 memb lead to increased Cl Ϫ uptake, as shown below.
Functional Analysis of Expressed Mutant cDNAs in Xenopus Oocytes-Measurement of 36 Cl Ϫ influx in oocytes expressing AE2 cyto /AE1 memb polypeptide provided a functional demonstration of surface expression of the mutant. Fig. 3 shows that injection of increasing mass quantities of AE2 cyto /AE1 memb cRNA produced increasing AE-mediated 36 Cl Ϫ influx activity. However, the nmol of chloride uptake/ng of injected cRNA for this chimera, as also for AE2⌬510, was very low compared with those measured for WT AE2 and WT AE1. In contrast, AE2⌬99 and AE1 cyto /AE2 memb chimera were both considerably more active per ng of injected cRNA than either WT cRNA. Operationally defined specific activities of chloride transport among the cRNAs spanned a range of 150-fold, from 17.8 nmol/ oocyte⅐h⅐ng (for AE1 cyto /AE2 memb ) to 0.12 nmol/oocyte⅐h⅐ng (for AE2 cyto /AE1 memb and for AE2⌬510). These differences were replicated using cRNA from multiple transcription reactions and from multiple cDNA templates. The rank order of specific activities of the AE polypeptides tested correlated with the rank order of amounts of AE polypeptide detected at the surface by proteolytic cleavage (Fig. 2).
The considerable range of AE-mediated chloride uptake/ng of injected cRNA among WT and mutant AE polypeptides suggested that prior titration of functional activity would be im- portant for functional comparison of multiple cRNAs in different lots of oocytes. Dose-response curves such as those of Fig. 3 allowed selection of quantities of each cRNA for injection, which reliably yielded absolute transport rates differing by no more than 3-fold among the various cRNAs in individual experiments. In the experiments summarized in Figs. 5 and 6, maximal AE-mediated Cl Ϫ uptakes at alkaline pH o were routinely 5-7 nmol/oocyte and never below 2 nmol/oocyte. The Cl Ϫ uptakes measured for all AE polypeptide constructs in pH sensitivity experiments represented initial rates.
Determination of Flux Assay Conditions in Which Oocyte pH i Is Near Steady State-When oocytes are placed into media of varied pH o , pH i changes in response. When pH o is shifted from neutral to acidic values, oocyte pH i acidifies to a minimum and then recovers toward resting pH i (25). ϳ40% of this recovery is due to endogenous sodium/hydrogen exchange activity (26,36,38) of the oocyte. We exploited these data to devise flux assay conditions in which AE-mediated chloride transport could be measured at near-steady-state oocyte pH i values. pH i of individual oocytes expressing WT AE2 was monitored via BCECF fluorescence ratio imaging during the transition from ND-96, pH o 7.4, to preincubation medium of varying pH o for 15 min and then into flux medium of the same pH o containing in addition 100 M amiloride and 10 M bumetanide. As shown in Fig. 4, oocytes exposed to varying pH o under these conditions showed considerable pH i change within 15 min, achieved nearsteady-state pH i values within ϳ30 min, and displayed minimal recovery over the ensuing 40 -80 min ( Fig. 4 and data not shown). At acid pH o this steady state reflected the balance between ongoing proton entry and the previously described (38) amiloride-resistant acid extrusion system(s) of the oocyte. 3 Table II summarizes the near-steady-state pH i values attained in these conditions by oocytes exposed to extracellular media of the indicated pH o values. Steady-state pH i in the presence of amiloride changed 0.12 units/unit pH o , similar to the value of 0.13 for maximal ⌬pH i previously determined in the absence of amiloride (25). This buffering of oocyte pH i contrasts with the minimal buffering of human erythrocyte pH i (Refs. 39 and 40; Table II), where the pH-dependence of AE1 measured previously reflected simultaneous and nearly equivalent variation of pH i and pH o (summarized in Ref. 7).
Regulation by pH of WT AE-mediated 36 Cl Ϫ Uptake-The pH dependence of chloride transport mediated by WT AE1 and by WT AE2 was compared under the assay conditions described in Fig. 4. Transport activities were measured at pH o values from 5.0 to 9.0. Near-steady-state pH i values measured in different oocytes in the same conditions ranged from 7.13 to 7.60 (Fig. 5). pH o determined by glass pH electrode and by BCECF fluorometry did not change during oocyte incubations of 15 and 60 min (not shown). Nonetheless, since the pH o range chosen exceeded the useful pH range of HEPES buffer, identical experiments were also performed in media buffered by 5 mM MES and by 5 mM bis-tris propane. The results obtained did not differ from those measured in the presence of HEPES (not shown). In addition, all AE constructs were subjected to pH o versus activity assays in which 60-min 36 Cl Ϫ influx in the presence of 100 M amiloride was compared to 15-min 36 Cl Ϫ influx in the absence of amiloride. pH dependence of transport in the two assay conditions did not differ.
WT AE1 and WT AE2 differed significantly in their regulation by pH (Fig. 5). WT AE1 (closed triangles) displayed a broad pH o versus transport activity curve (n ϭ 6). Activity was maximal at pH o 7.0, decreased to 62.7 Ϯ 6.9% of maximal value at pH o 5.0 (p Ͻ 0.005), and decreased to 82.7 Ϯ 5.3% of maximal value at pH o 9.0 (p ϭ 0.16). In contrast, WT AE2 activity (open triangles) displayed a steep dependence on pH o . Activity was maximal at pH o 9.0 and decreased as pH o decreased, with a pH o at which activity was half-maximal (pH o(50) ) of 7.03 Ϯ 0.08 (n ϭ 13). AE2 activity was completely suppressed at pH o 5.0, approximately corresponding to pH i 7.13 (see Table II). The data relating pH o and AE2-mediated 36 Cl Ϫ influx were well fit by a first order logistic sigmoid equation (Ultrafit goodness of fit index ϭ 0.93 Ϯ 0.03, n ϭ 13).
Regulation by pH of Chimeric AE-mediated 36 Cl Ϫ Uptake-This major difference between pH o versus activity curves of WT AE2 and WT AE1 allowed us to test the hypothesis that a single region of AE2 distinct in amino acid sequence from AE1 is responsible for the difference. Therefore, the pair of domainswap chimeras pictured in Fig. 1 was examined for pH sensitivity of 36  N-terminal cytoplasmic domain of AE2 was necessary and sufficient for the increased pH sensitivity of AE2-mediated 36 Cl Ϫ uptake predicted in its simplest form that the AE2 cyto /AE1 memb chimera would resemble AE2 in its regulation by pH. The converse hypothesis that the C-terminal transmembrane domain of AE2 was necessary and sufficient for the increased pH sensitivity of AE2-mediated 36 Cl Ϫ uptake predicted in its simplest form that the AE1 cyto /AE2 memb chimera would resemble AE2 in its regulation by pH. The experimental data partially supported this second hypothesis. The C-terminal transmembrane domain of AE2 proved necessary, but not sufficient, to reproduce the pH versus activity phenotype of WT AE2.
As shown in Fig. 5, AE2 cyto /AE1 memb (closed circles) displayed peak activity at pH o 8.0 and 9.0 and retained 61.9 Ϯ 4.7% maximal activity at pH o 5.0. (n ϭ 4). This pH dependence of AE2 cyto /AE1 memb resembled that of WT AE1 to a much greater degree than that of WT AE2. Therefore, the AE2 cytoplasmic domain was not sufficient to produce an AE2 pHregulatory phenotype in this assay. In addition, the presence of the AE1 transmembrane domain sufficed to preserve most of the AE1 pH-regulatory phenotype.
pH dependence of chloride transport activity of the converse chimera, AE1 cyto /AE2 memb , was also examined (open circles, Fig. 5). AE1 cyto /AE2 memb almost precisely resembled WT AE2 in its pH o versus activity curve, except for a shift to acid pH values. AE1 cyto /AE2 memb was also maximal at pH o 8.0, but the pH o 50 value was 6.34 Ϯ 0.06 (n ϭ 13). The mean of ⌬pH o50 (the difference between pH o 50 for WT AE2 and pH o 50 for AE1 cyto / AE2 memb ) values measured in each of 13 individual experiments was 0.69 Ϯ 0.04 units (p Ͻ 0.001). This value and its statistical significance were not different, whether calculated only from the nine experiments in which transport was assayed at all five pH o values or (as presented) from these nine plus four additional experiments in which transport was assayed at only four pH o values. Results were also indistinguishable whether calculated as the mean of ⌬pH o(50) values calculated from individual experiments or as the ⌬pH o(50) calculated as the difference between mean pH o(50) values for WT AE2 and for AE1 cyto /AE2 memb .
This result suggested that the AE2 carboxyl-terminal transmembrane domain contained a pH sensor sufficient for expression of most of the WT AE2 pH-regulatory phenotype. However, substitution of the AE1 cytoplasmic domain for that of AE2 in the presence of the AE2 transmembrane domain led to a decreased sensitivity to inhibition of transport activity by protons. Thus, the AE2 amino-terminal cytoplasmic domain exerted a modifier function that was necessary for complete replication of the WT AE2 pH-regulatory phenotype.
Regulation by pH of 36 Cl Ϫ Uptake Mediated by N-terminally Truncated AE2 Polypeptides-The hypothesis of modifier function predicts in simplest form that sufficient truncation of the AE2 cytoplasmic domain should produce a pH versus activity phenotype that resembles that of the AE1 cyto /AE2 memb chimera. In order to localize this proposed modifier function within the N-terminal cytoplasmic domain, two cytoplasmic domain truncation mutants of AE2 were examined. The pH dependence of AE2⌬ N 99 and AE2⌬ N 510 were assayed under the same experimental conditions as had been used to study the chimeric polypeptides. As shown in Fig. 6, the pH versus activity curves for these cytoplasmic domain truncation mutants were of similar shape to that of WT AE2, with complete inhibition of transport activity at pH o 5.0. Whereas the pH dependence curve for AE2⌬99 was indistinguishable from that of WT AE2, the curve for AE2⌬ N 510 was shifted to a more acidic pH value (⌬pH o(50) between wt AE2 and AE2⌬ N 510 ϭ 0.66 Ϯ 0.13 (n ϭ 5); p Ͻ 0.005). This result was unchanged when calculated as the difference between mean pH o(50) values for WT AE2 and AE2⌬ N 510. The pH dependence curve of AE2⌬ N 510 did not differ detectably from that of AE1 cyto / AE2 memb . Thus the region between AE2 amino acids 100 and  and for AE1 cyto /AE2 memb (open circles, n ϭ 13), data from each independent experiment were fitted to a logistic sigmoid equation, normalized to fitted saturation activities, and then averaged. For WT AE1 (filled triangles, n ϭ 6) and AE2 cyto /AE1 memb (filled circles, n ϭ 4), data from each independent experiment were normalized to the highest Cl Ϫ influx activities and then averaged. 510 contained within it a modifier site required for enhancing the proton sensitivity of the pH sensor.
Role of pH i in the Regulation of AE2 Activity by pH-Since the experiments presented in Figs. 5 and 6 were performed under conditions under which both pH o and pH i were varied, the relative contributions of pH o and pH i to the regulation of AE2 by pH were not addressed. However, the localization of a modifier site to the cytoplasmic N-terminal hydrophilic domain of AE2 suggested an important role for pH i . Earlier experiments that demonstrated regulation by pH i of AE2-mediated chloride/HCO 3 exchange in mammalian cells were performed in the presence of CO 2 /bicarbonate (21,25). In these experiments activation of AE2 by alkaline pH i never reached plateau values, probably due to parallel allosteric activation and increasing concentrations of substrate bicarbonate. Therefore, we sought to examine AE2-mediated chloride/base exchange activity in oocytes under conditions that would alter oocyte pH i at constant pH o .
Substitution of increasing amounts of extracellular anion with equimolar acetate (Fig. 7, upper horizontal axis) produced graded increases in acidification of oocytes, as measured with fluorescence ratio imaging of BCECF (Fig. 7, lower horizontal axis). Increasing acetate concentrations were also accompanied by dose-dependent inhibition of AE2-mediated chloride uptake. At pH o 7.4, half-maximal inhibition of AE2 by acetate substitution occurred at 13.5 mM acetate (n ϭ 3), corresponding to an approximate reduction in pH i of 0.16 units (n ϭ 4). In contrast, half-maximal inhibition of AE2 by extracellular acidification (normalized to activity at pH o 7.4 in order to compare the data of Fig. 5 with those of Fig. 7) occurred at pH o 6.83 (n ϭ 13), corresponding to a pH i reduction of only 0.06 units (n ϭ 4).
Injection of sodium acetate, pH 7.4, to an estimated final intracellular concentration of 13 mM neither acidified the oocytes nor inhibited AE2-mediated uptake of 36 Cl Ϫ (not shown). Thus, the inhibitory effect of extracellular acetate substitution was not likely due to simple inhibition of AE2 by competition at an anion substrate site but rather reflected an effect of intracellular acidification. Considered together, these data support the conclusion that pH i contributed significantly, but not entirely, to the regulation of AE2 by pH under conditions in which both pH i and pH o were changed (as presented in Figs. 5 and 6).

DISCUSSION
The present work has used analysis of recombinant chimeric and truncated polypeptides to initiate determination of the structural loci of pH sensitivity of AE2-mediated monovalent anion exchange. Functional studies of recombinant AE polypeptides have been reported in several heterologous expression systems. Although WT AE2 has been functionally expressed in mammalian cells (21,25), heterologous WT AE1 and a version of the AE1 cyto /AE2 memb chimera did not reach the cell surface. In contrast, functional plasmalemmal expression not only of AE2 (25,26) but also of heterologous WT and truncated AE1 (30,31,41) has been reported in Xenopus oocytes. The present study establishes that Xenopus oocytes can also express functional chimeric AE polypeptides at the cell surface and so has set the stage for structure-function studies designed to delineate the amino acid sequences responsible for AE isoform-specific transport properties. We have begun this process with the study of the structural domains responsible for the different pH sensitivities of AE1 and AE2.
In contrast to the experiments that first demonstrated pH sensitivity of heterologous AE2 in Xenopus oocytes (25), the current flux assay was carried out at near-steady-state pH i . In the experiments of Figs. 5 and 6, oocytes were preincubated for 15 min at the desired pH o prior to initiation of the 36 Cl Ϫ influx assay at the same pH o . Oocytes displayed minimal recovery from alkaline pH i during the time course of the experiment in these nominally CO 2 -free conditions. The presence of amiloride during 60-min influx assays to block the endogenous oocyte Na ϩ /H ϩ exchanger (26,36,38), or the performance of 15-min influx assays, achieved near-steady-state pH i during influx periods. As pH o was changed between 9.0 and 5.0, the corresponding steady-state oocyte pH i values ranged from 7.60 to 7.13, respectively. Some oocytes had not achieved maximal ⌬pH i by the end of the 15-min preincubation period (Fig. 4); for these oocytes the influx period included time at pre-steadystate pH i values. Thus, the observed pH-dependent changes in AE-mediated transport activity represented minimal estimates compared with ideal pH i clamp conditions. The experiments demonstrated the requirement for the transmembrane domain of AE2 for display of the characteristic steep pH versus activity curve in WT AE2, AE1 cyto /AE2 memb , AE2⌬ N 99, and AE2⌬ N 510. In contrast, the presence in any polypeptide of the AE1 transmembrane domain produced a broad pH dependence of chloride uptake. Moreover, this pH sensitivity associated with the presence of the AE1 transmembrane domain was only minimally changed by substitution of the AE2 N-terminal cytoplasmic domain for its AE1 counterpart. This apparent lack of cytoplasmic domain specificity differed from the pH sensitivity of the AE2 transmembrane domain, which displayed modulation of affinity for proton equivalents by the contiguous N-terminal cytoplasmic domain of AE2, but not by that of AE1 (Fig. 5). Thus, attachment of the AE1 cytoplasmic domain onto the AE2 transmembrane domain led to an acid shift of 0.7 units in the pH o(50) value of the pH versus activity curve. This shift was reproduced by removal of the first 510 amino acids from the AE2 cytoplasmic domain. However, restricting the truncation to only 99 amino acids led to reproduction of the WT AE2 pH dependence phenotype (Fig. 6).
Definitive assignment of the AE2 transmembrane domain pH sensor to the endofacial or exofacial aspect of the plasmalemma is not yet possible. The combined variation of pH i and pH o did not allow unambiguous assignment of regulatory function to intra-or extracellular protons. Inhibition of AE2-mediated chloride uptake by acetate-induced intracellular acidification at constant pH o (Fig. 7) indeed suggested a role for pH i in this regulation. However, AE2 inhibition by acetate was accompanied by a reduction in pH i of 0.16 units, whereas AE2 inhibition by extracellular acidification was accompanied by a 0.06unit reduction of pH i . In addition, recent experiments have indicated that variation of pH o under conditions of minimal pH i change also can regulate AE2 activity. 4 Thus, pH i very likely could not account completely for the AE2 inhibition produced by lowering pH o in Figs. 5 and 6. These data together allow the proposal that AE2 can be regulated both by pH i and by pH o . Coincident variation of pH i and pH o is a common pathophysiological derangement in the setting of ischemia and hypoxia and thus might cooperatively regulate AE2 activity in vivo.
When AE2 transport activity was plotted as a function of pH o , pH dependence fit simple logistic sigmoid kinetics. 80% of the range of measured AE2 activity was traversed across the 100-fold range of extracellular proton concentration between pH o 6.0 and 8.0 (Fig. 5, upper x axis). pH i could not be easily measured in the same oocytes in which influx measurements were performed, but pH i was measured in separate oocytes otherwise treated identically. When AE2 transport activity was plotted against the pH i values measured in identically treated oocytes (Fig. 5, lower x axis), 80% of the range of measured AE2 activity was traversed across a Յ2-fold range of intracellular proton concentration, between 28 and 51 nM (Table II and Fig.  5). This degree of cooperativity with respect to pH i allows the suggestion of multiple protonatable sensor sites in AE2.
These responses of the AE constructs to combined variation of pH i and pH o suggested a model for pH regulation of AE2 reminiscent of that proposed for NHE1 (32,42). The model proposes that a pH sensor resides within the transmembrane domain. This sensor must be of sufficiently high proton affinity to account for a large part of the difference between the pH sensitivities of WT AE2 and WT AE1. In addition, the AE2 N-terminal cytoplasmic segment harbors a modifier domain whose removal reduces the proton affinity of the transmem-brane domain pH sensor. The reported deletion experiments locate this modifier domain between mouse AE2 cytoplasmic amino acids 100 and 510 (Fig. 6). Whether the modulator directly recognizes the sensor or interacts with it indirectly or via additional polypeptide(s) remains to be determined.
The postulated pH-sensing function could have one or more of the following three topographical dispositions. The pH sensor might detect pH i from the endofacial surface of AE2. Location of the sensor at the AE2 cytoplasmic face is suggested (but not required) by the cytoplasmic location of the modifier domain. Such a sensor might reside within cytoplasmic loops connecting transmembrane spans, within a water-accessible vestibule structure in the plane of the lipid bilayer, or in the C-terminal cytoplasmic tail. The same or a different pH sensor could detect pH o at exofacial residues, including a postulated vestibule structure such as that which might accommodate the sulfonate moieties of the stilbene inhibitors of transport (43). Alternatively, the sensor(s) might be located in a position that allows proton sensing of both intracellular and extracellular solutions, thus providing for dual sensing of pH i and pH o . Glu 681 (44) and Lys 851 (45) of human AE1 may be exposed to both sides of the red cell lipid bilayer. Such residues very likely contribute to the anion binding and translocation pathway through AE1.
In summary, AE2 and AE1 function in Xenopus oocytes differ in their regulation by pH. The steeper pH dependence of AE2 function is mediated by its C-terminal transmembrane domain. Modulation of that pH dependence requires a region of the AE2 N-terminal cytoplasmic domain (the modifier site) between residues 100 and 510. Future experiments will address the individual and cooperative roles of pH i and pH o in regulation of AE2 activity, define more precisely those residues that comprise the pH sensor(s) and modifier site of AE2, and examine the mechanisms by which these two functional domains interact with one another and with other regulatory sites within the AE2 polypeptide.