Acute pH-dependent regulation of AE2-mediated anion exchange involves discrete local surfaces of the NH2-terminal cytoplasmic domain.

We have previously defined in the NH2-terminal cytoplasmic domain of the mouse AE2/SLC4A2 anion exchanger a critical role for the highly conserved amino acids (aa) 336-347 in determining wild-type pH sensitivity of anion transport. We have now engineered hexa-Ala ((A)6) and individual amino acid substitutions to investigate the importance to pH-dependent regulation of AE2 activity of the larger surrounding region of aa 312-578. 4,4'-Diisothiocyanostilbene-2,2'-disulfonic acid (DIDS)-sensitive 36Cl- efflux from AE2-expressing Xenopus oocytes was monitored during changes in pHi or pHo in HEPES-buffered and in 5% CO2/HCO3- -buffered conditions. Wild-type AE2-mediated 36Cl- efflux was profoundly inhibited at low pHo, with a pHo(50) value = 6.75 +/- 0.05 and was stimulated up to 10-fold by intracellular alkalinization. Individual mutation of several amino acid residues at non-contiguous sites preceding or following the conserved sequence aa 336-347 attenuated pHi and/or pHo sensitivity of 36Cl- efflux. The largest attenuation of pH sensitivity occurred with the AE2 mutant (A)6357-362. This effect was phenocopied by AE2 H360E, suggesting a crucial role for His360. Homology modeling of the three-dimensional structure of the AE2 NH2-terminal cytoplasmic domain (based on the structure of the corresponding region of human AE1) predicts that those residues shown by mutagenesis to be functionally important define at least one localized surface region necessary for regulation of AE2 activity by pH.

The anion exchangers AE1, AE2, and AE3 are polypeptide products of the SLC4 bicarbonate transporter gene superfamily that mediate Na ϩ -independent Cl Ϫ /HCO 3 Ϫ exchange. Anion exchangers contribute to the regulation of intracellular pH (pH i ), cell volume, tonicity, and intracellular Cl (Cl Ϫ ) in metazoan cells (1)(2)(3)(4)(5). The AE polypeptides share a highly conserved hydrophobic, polytopic, transmembrane domain of Ͼ500 amino acids (aa), 1 with a short COOH-terminal cytoplasmic tail capable of binding carbonic anhydrase II (6,7). This transmembrane domain is preceded by a less extensively conserved hydrophilic NH 2 -terminal cytoplasmic domain of 400 -700 aa (3). The polytopic transmembrane domain can mediate anion exchange in the absence of nearly the entire NH 2 -terminal cytoplasmic domain (8,9). Whereas the NH 2 -terminal cytoplasmic domain of erythroid AE1 is important for its binding to multiple cytoskeletal proteins, glycolytic enzymes, and hemoglobin (10), functions of the cytoplasmic NH 2 -terminal domains of AE2 and AE3 remain less extensively investigated.
The polypeptide products of the various SLC4 AE anion exchanger genes differ in their acute regulation by pH. Native AE1-mediated Cl Ϫ /HCO 3 Ϫ exchange in erythrocytes (11) and heterologous AE1-mediated Cl Ϫ /Cl Ϫ exchange in Xenopus oocytes (12,13) both display a broad pH versus activity profile, consistent with the primary role of erythroid AE1 in facilitating CO 2 /HCO 3 Ϫ exchange between the respiring tissues and lungs (14). In contrast, non-erythroid, Na ϩ -independent Cl Ϫ /HCO 3 Ϫ exchange in many cell types is sensitively regulated by changes in pH i (e.g. Refs. 15 and 16), consistent with its proposed role in pH i recovery from alkaline loads. Recombinant AE2 when expressed in tissue culture cells (17,18) or in Xenopus oocytes (12,13,19) is highly sensitive to changes in both pH o and pH i . Similarly, recombinant cardiac AE3 expressed in Xenopus oocytes is responsive to changes in pH i (19).
Structure-function experiments have indicated that an operationally defined AE2 "pH sensor site" that enhances pH o sensitivity of anion transport is localized to the transmembrane domain (13). This study also demonstrated that the extracellular proton sensitivity (pH o(50) value) of AE2-mediated 36 Cl Ϫ influx (ϳ7.0 under conditions of unclamped pH i ) was acidshifted by up to 0.7 units following various truncations of the cytoplasmic NH 2 -terminal domain. This result suggested the presence, between aa 99 and 510 of the 705-aa AE2 NH 2terminal cytoplasmic domain, of a "pH modifier site" that modulates anion transport activity of the transmembrane domain. More recently, 36 Cl Ϫ efflux assays in AE2-expressing oocytes revealed that pH sensitivity of the transporter requires the integrity of two non-contiguous regions of the NH 2 -terminal cytoplasmic domain: the highly conserved aa 336 -347 (20) and the less well conserved region of aa 391-510 (19). The con-served residues aa 336 -347 play a similar role in pH-dependent regulation of the AE3 polypeptide and are likely to be important in other SLC4 transporters, as well (20).
In the present work we extend our mutagenic analysis to the AE2 NH 2 -terminal cytoplasmic domain residues that precede and follow the highly conserved aa 336 -347 sequence, as guided by our earlier sequential deletion studies (18,20). We demonstrate that several non-contiguous regions of the AE2 NH 2 -terminal cytoplasmic domain contain conserved amino acids whose mutation alters the regulation of anion exchange activity by pH i and/or by pH o . Homology modeling suggests that a substantial subset of these residues contributes to at least one pH-sensitive surface of this domain. A preliminary report of this work has been published (21). 36 Cl was purchased from ICN (Irvine, CA). Other chemical reagents were of analytical grade and obtained from Sigma, Calbiochem, or Fluka. Restriction enzymes and T4 DNA ligase were from New England BioLabs (Beverly, MA). Taq DNA polymerase and dNTPs were from Promega (Madison, WI) or Invitrogen.

Reagents-Na
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, pH 7.40. Flux media lacked sodium pyruvate. pH values of 7.0, 8.0, and 8.5 in room air flux media were achieved with 5 mM HEPES. 5 mM MES was used for room air flux media of pH values 5.0 and 6.0. In Cl Ϫ -free solutions, NaCl was replaced isosmotically with 96 mM sodium isethionate, and equimolar potassium, calcium, and magnesium gluconate substituted for the corresponding Cl Ϫ salts. CO 2 /HCO 3 Ϫ -buffered solutions of pH 7.4 were saturated with 5% CO 2 , 95% air at room temperature for ϳ1 h, and differed from Cl Ϫ -free ND-96 in the replacement of 24 mM sodium isethionate with 24 mM NaHCO 3 . The pH of CO 2 /HCO 3 Ϫ -buffered solutions was verified prior to each experiment. Addition to flux media of the weak acid salt, sodium butyrate, was as an equimolar substitution for NaCl.
Mutant AE2 cDNAs-Murine AE2 encoded in plasmid p⌬X (13) was used as template for polymerase chain reaction. The AE2 hexa-Ala bloc substitution mutants (A) 6 357-362, (A) 6 391-396, (A) 6 397-402, and (A) 6 403-408 were constructed by a four-primer PCR method as described (13,19,22). Single residue missense mutations were constructed by the same method. Integrity of PCR products and ligation junctions was confirmed by DNA sequencing of both strands. AE2 mutant D578K was constructed using the QuikChange Mutagenesis kit (Stratagene) and sequenced only at the site of the mutation. Oligonucleotide primers were obtained from Biosynthesis (Woodlands, TX) or Sigma (Haverhill, UK); primer sequences are available upon request.
cRNA Expression in Xenopus Oocytes-Mature female Xenopus (Xenopus One, Madison, WI, or Blades, UK) were maintained and subjected to partial ovariectomy as described (12). Stage V-VI oocytes were manually defolliculated following incubation of ovarian fragments with 2 mg/ml collagenase A (Roche Molecular Biochemicals) for 60 min in ND-96 solution containing 50 ng/ml gentamycin and 2.5 mM sodium pyruvate. Oocytes were injected on the same day with cRNA or with H 2 O in a volume of 50 nl. Capped cRNA was transcribed from linearized cDNA templates with the T7 MEGAscript kit (Ambion, Austin, TX), and resuspended in diethyl pyrocarbonate-treated water. RNA integrity was confirmed by agarose gel electrophoresis in formaldehyde, and RNA concentration was estimated by A 260 . Injected oocytes were then maintained for 2-6 days at 19°C. 36 Cl Ϫ Efflux Measurements-Individual oocytes in Cl Ϫ -free ND-96 were injected with 50 nl of 130 mM Na 36 Cl (10,000 -12,000 cpm). Following a 5-10-min recovery period, the efflux assay was initiated by transfer of individual oocytes to 6-ml borosilicate glass tubes, each containing 1 ml of efflux solution. At intervals of 3 min, 0.95 ml of this efflux solution was removed for scintillation counting and replaced with an equal volume of fresh efflux solution. Following completion of the assay with a final efflux period in the presence of the anion transport inhibitor DIDS (200 M), each oocyte was lysed in 100 l of 2% SDS. Samples were counted for 3-5 min such that the magnitude of 2 S.D. was Ͻ5% of the sample mean.
Experimental data were plotted as ln (% cpm remaining in the oocyte) versus time. 36 Cl Ϫ efflux rate constants were measured from linear fits to data from the last three time points sampled for each experimental condition. All single time point values for 36 Cl Ϫ efflux from AE2 cRNA-injected oocytes into chloride medium exceeded 150 cpm. Efflux cpm values for water-injected oocytes (40 -90 cpm) were indistinguishable from those for AE2 cRNA-injected oocytes in the presence of DIDS. Both of these values exceeded machine background values of 20 cpm (peak 30 cpm). Within each experiment, water-injected and AE2 cRNA-injected oocytes from the same frog were subjected to parallel measurements. On each experimental day, 36 Cl Ϫ efflux activity of the tested mutant AE2 polypeptides was compared at pH o 7.4 to wild-type AE2 activity ("basal activity"). Each AE2 mutant was tested in oocytes from at least two frogs, and most in oocytes from three frogs. These data are summarized in Supplemental Materials Table I.
Measurement of pH o Dependence of 36 Cl Ϫ Efflux-Individual oocytes maintained in Cl Ϫ -free solution at pH o 7.4 were exposed sequentially to (Cl Ϫ containing) ND-96 at pH 5.0, 6.0, 7.0, 8.0, and 8.5, then to a solution at pH 8.5 in the presence of DIDS (13,19). Rate constants measured at each pH o value for wild-type AE2 and for the tested AE2 mutants in each individual experiment were fit (Sigma Plot) to the following first-order logistic sigmoid equation,  36 Cl Ϫ efflux did not differ when the experiment was performed with the order of pH o change reversed or randomized (20). pH o dependence was also unchanged when pH i was clamped during pH o variation by simultaneously changing the butyrate concentration in the bath (20). All pH o(50) data for wild-type and mutant AE2 polypeptides are summarized in Supplemental Materials Table I.
Measurement of pH i Dependence of 36 Cl Ϫ Efflux-To vary pH i at constant pH o , 36 Cl Ϫ -injected oocytes were preincubated for 30 min prior to the start of the experiment in pH 7.4 Cl Ϫ -free solution containing 40 mM sodium butyrate. 36 Cl Ϫ efflux was then initiated by transfer of oocytes into Cl Ϫ -containing solution in the continued presence of butyrate. The oocytes were then transferred into efflux medium containing Cl Ϫ but lacking weak acid. In these conditions, pH i increases 0.50 Ϯ 0.03 units after butyrate removal, with a time constant of ϳ6 min, and butyrate is neither an inhibitor of nor a substrate for AE2 (19).
In our previous work, we defined the pH i sensitivity of AE2 activity as the -fold increase in the rate constant for 36 Cl Ϫ efflux following a rise of pH i (induced by butyrate removal) (Supplemental Materials Fig. S1, left panel). This method of data presentation places no upper limit on the value for stimulation. It is, however, influenced more by small differences in the low 36 Cl Ϫ efflux rate constant measured in the presence of butyrate (i.e. at low pH i ) than by changes in the higher, and better resolved, efflux rate constant measured subsequently in its absence (at high pH i ). In the present work, we have therefore adopted a more conservative presentation of pH i sensitivity, namely, the fractional reduction in the 36 Cl Ϫ efflux rate constant associated with a reduction of pH i . Note that this value is the reciprocal of the value previously reported for the same assay. The ratio is referred to as the "normalized 36 Cl Ϫ efflux rate constant" in response to 40 mM butyrate, and reflects the response to acidic pH i at unchanged pH o . Mean values of this normalized 36 Cl Ϫ efflux rate constant for each AE2 mutant were compared with that of wild-type AE2 tested simultaneously by Student's unpaired t test. The level of significance was taken as p Ͻ 0.05. All normalized 36 Cl Ϫ efflux rate constants in units of % are summarized for wild-type and mutant AE2 polypeptides in Supplemental Materials Table I.
Modeling kAE2 Cytoplasmic NH 2 -terminal Structure-The crystal structure of a portion of the human erythroid AE1 cytoplasmic NH 2terminal domain (10) was used as a template to model the corresponding portion of mouse AE2 cytoplasmic NH 2 -terminal domain. Geno3D (geno3d-pbil.ibcp.fr) was provided with the AE1 template (GenBank TM pdb1hynQ-0) to generate a predicted structure for AE2 cytoplasmic NH 2 -terminal domain aa 317-610, and rendered with DeepView/ SwissPDB viewer 3.7 (us.expasy.org/spdbv/). DeepView's Nearest Neighbor function was applied to detect amino acids predicted to reside within defined Å distances of any residue of interest. Ion pair interaction distances were interpreted according to ion-pair distances measured in a series of polypeptides of known structure (23).

Roles of Candidate SH3 Binding Domain and Histidine
Residues within Amino Acids 312-317 for AE2 Regulation by pH o and pH i -We reported previously that the region encompassing amino acids 310 -347 within the NH 2 -terminal cytoplasmic domain of the murine AE2 anion exchanger is required for normal regulation of AE2-mediated Cl Ϫ transport by pH o and pH i (19,20). As shown in Fig. 1A, aa 312-317 includes two prolines that comprise a potential SH3 domain binding site, as well as two potentially titratable histidine residues. We mutated each residue individually to Ala and assessed the effects on Cl Ϫ /Cl Ϫ exchange of varying pH o or pH i . Fig. 1B indicates that most mutants exhibited 36 Cl Ϫ efflux activity at pH o 7.4 comparable with that of wild-type AE2. Fig. 1C profiles 36 Cl Ϫ efflux activity (normalized as described under "Materials and Methods") as a function of pH o for wild-type AE2, the two AE2 mutants P313A and H317A, and the double mutant H314A/ H317A. The pH o value at which the rate constant for wild-type, AE2-mediated 36 Cl Ϫ efflux was half-maximal (pH o(50) ) at 6.75 Ϯ 0.05 (n ϭ 24; Fig 1, C and E). The single mutant P313A and the double mutant H313A/H317A each exhibited a pH o dependence indistinguishable from that of wild-type AE2, whereas the pH o(50) value of the AE2 mutant H317A was shifted to a more acidic pH value of 6.20 Ϯ 013 (n ϭ 11, p Ͻ 0.05). Fig. 1E summarizes, for all mutants tested, the pH o(50) values measured from data like that shown in Fig. 1C. Although mutagenesis of either His 314 or His 317 led to an acidshift in pH o(50) value, the double mutant H314A/H317A exhibited wild-type regulation by pH o .
We next investigated the effect of changing pH i at a constant pH o upon activity of the same set of AE2 mutants. Bath addition and removal of the weak acid, butyrate, respectively, decreases and increases oocyte pH i (19). Fig. 1D shows a representative efflux trace in which 36 Cl Ϫ efflux via wild-type AE2, and via the mutants P313A, H317A, and H314A/H317A, was reduced at low pH i (butyrate addition) and subsequently stimulated when pH i was elevated (by butyrate removal). Fig. 1F summarizes similar experiments for all mutants tested, results are expressed as the 36 Cl Ϫ efflux rate constant at low pH i (in the presence of 40 mM butyrate) normalized to that measured at high pH i (following extracellular butyrate removal, see "Materials and Methods" and Supplemental Materials Fig.  S1). The data do not suggest involvement of the proline and histidine residues in pH i -dependent regulation of AE2 activity, because their mutation to alanine did not prevent reduction of 36 Cl Ϫ efflux at low pH i . Note that while the AE2 mutants, H314A and H317A, were altered in their sensitivity to pH o their sensitivity to pH i was unchanged (Fig. 1F). The data also indicate that pH-dependent regulation of AE2 does not require binding of a SH3 domain-containing protein at this site, because the proline mutants exhibited normal sensitivity to both pH i and pH o .
Individual Mutations within AE2 aa 318 -323 Identify Residues Important for Regulation of Cl Ϫ Transport by pH o and pH i -Our previous study using hexa-Ala bloc substitutions implicated aa 318 -323 as important for the regulation of AE2 activity by pH (20). However, the AE2 mutant (A) 6 318 -323 expressed too low an activity at pH o 7.4 to allow study of inhibition by extracellular acidification. Therefore, we re-evaluated this region by alanine scan of the individual residues ( Fig. 2A). Fig. 2B shows that all mutants studied retained 36 Cl Ϫ efflux activity at pH o 7.4, sufficient for an analysis of pH sensitivity. Fig. 2C shows that the AE2 mutant, E318A (open circles), exhibited a pH o dependence similar to that of wild-type AE2 (filled circles), except for its incomplete inhibition at pH o 5.0. In contrast, the AE2 mutant F320A (filled squares) and the double mutant E318A/E322G (open squares) displayed acidshifted pH o(50) values of 5.83 Ϯ 0.12 (n ϭ 10, p Ͻ 0.05) and 6.20 Ϯ 0.13 (n ϭ 7, p Ͻ 0.05), respectively. Because both of these mutants also exhibited incomplete inhibition at pH o 5.0, these pH o(50) values represent maximal estimates. Fig. 2E summarizes the pH o sensitivity of all the point mutants, plotted as pH o(50) values derived from results like those illustrated in Fig. 2C. Two individual mutations (F320A and L323A) and the (inadvertently derived) double mutation E318A/E322G led to significant shifts in pH o(50) values relative to that of wild-type AE2. Alanine substitution of other residues did not significantly shift pH o(50) values. Mutation of the hydrophobic phenylalanine (Phe 320 ) to the similarly sized but more hydrophilic tyrosine (as in the corresponding position of mAE1) also acid-shifted the pH o(50) value, but to a lesser degree. Note that, although individual mutations of glutamate 318 or 322 to alanine had no effect on the pH o(50) value, there was an acid-shift for the double mutant.
We next tested the pH i sensitivity of the same AE2 mutants. Fig 2D shows a representative experiment in which 36 Cl Ϫ efflux via wild-type AE2 and the mutant E318A was inhibited at low pH i (in the presence of butyrate) and subsequently stimulated by an elevation of pH i (induced by extracellular butyrate removal). In contrast, mutants F320A and E318A/E322G both showed a significantly reduced inhibition at low pH i (and therefore a reduced stimulation upon elevation of pH i ). Fig. 2F summarizes the pH i sensitivity of all the point mutants tested here in comparison with wild-type AE2. The mutants F320A, F320Y, and L323A, and the double mutants E318A/E322G and E318V/E322A each displayed a decrease in 36 Cl Ϫ efflux activity at low pH i (in the presence of 40 mM butyrate) that was significantly lower than wild-type. This loss of inhibition by acid pH i was also manifest in AE2 F320A-mediated Cl Ϫ /HCO 3 Ϫ exchange (not shown). These data therefore identify multiple amino acid residues whose mutation alters wild-type regulation of AE2 activity by both pH i and pH o. The next segment of AE2, aa 324 -335, was not further examined because the AE2 mutants A (6) 324 -329 and A (6) 330 -335 exhibited wild-type regulation by both pH o and pH i (20).
The Effect of Individual Mutations within AE2 aa 348 -355 on Regulation by pH i -We determined previously that the adjacent, strongly conserved AE2 region encompassing aa 336 -347 was critical for wild-type sensitivity to pH o and pH i (20). However, the AE2 region aa 348 -355 immediately adjacent to the conserved critical region of aa 336 -347 is only partially conserved across the SLC4 gene family. This adjacent region includes several glutamic acid residues. Prompted by our earlier identification of Glu 346 and Glu 347 as important in regulation of AE2 by pH, we subjected AE2 residues 348 -355 individually to alanine scan mutagenesis. Each of these AE2 mutants displayed near wild-type 36 Cl Ϫ efflux activity at pH o 7.4, and retained wild-type regulation by pH i (Supplemental Materials Table I).
Mutagenesis of AE2 Residues 356 -362 Identifies a Conserved Histidine Residue Important for pH i and pH o Sensitivity-The high degree of conservation of AE2 Trp 356 among SLC4 polypeptides extends through the adjacent region encompassing aa 357-362. Therefore, we proceeded to test the role of this region by hexa-Ala bloc substitution, and by individual Ala substitution of Trp 356 and the candidate pH-sensitive His 360 residue (Fig. 3A). All mutants were highly active at pH 7.4 (Fig.  3B). The pH o versus 36 Cl Ϫ efflux activity profiles of W356A, (A) 6 357-362, and H360E were each acid-shifted when compared with that of wild-type AE2 (Fig. 3C). Whereas the pH o(50) value of W356A was acid-shifted to a moderate degree, the pH o(50) value of (A) 6 357-362 was acid-shifted by ϳ1 pH unit. As pH-dependent Regulation of Anion Exchanger AE2 summarized in Fig. 3C, this shift was partially phenocopied by the individual mutant H360E, suggesting a role for His 360 in the regulation of AE2 by pH o .
We next compared pH i sensitivity of AE2 mutants W356A, (A) 6 357-362, and H360E with wild-type AE2 (Fig. 3D). In contrast to wild-type AE2, which exhibits a low rate constant at

pH-dependent Regulation of Anion Exchanger AE2
acid pH i and is subsequently stimulated by raising pH i , each of these mutants had robust 36 Cl Ϫ efflux activity at low pH i and were minimally responsive to subsequent elevation of pH i .
These data are the first to define a histidine residue within the NH 2 -terminal cytoplasmic domain whose mutation alters AE2 regulation by both pH o and pH i .   (19). Subsequent deletion analysis narrowed this area to aa 391-410 (data not shown). As an initial step toward identification of individual amino acid residues required for this regulation, we tested the functional properties of three hexa-Ala bloc substitutions spanning this region and we then tested individual substitution mutants (Fig. 4A). As shown in Fig. 4B, only (A) 6 397-402 mutant lacked activity at pH o 7.4 sufficient to evaluate its regulation by pH o (Fig. 4C). Fig. 4C summarizes pH o sensitivity of the active mutants. The (A) 6 391-396 mutant exhibited a pH o(50) value of 6.95 Ϯ 0.09 (n ϭ 6, p Ͼ 0.05), similar to that of wild-type AE2, 6.71 Ϯ 0.09 (n ϭ 16), whereas the (A) 6 403-408 mutant displayed a significantly acid-shifted pH o(50) value of 6.05 Ϯ 0.19 (n ϭ 10, p Ͻ 0.05). Thus aa 403-408 appear important for pH o -dependent regulation of AE2. In contrast, both AE2 mutants (A) 6 403-408 and (A) 6 391-396 exhibited wild-type pH i sensitivity (p Ͼ 0.05) (Fig. 4D). The AE2 (A) 6 397-402 mutant appeared to be insensitive to regulation by pH i (Fig. 4D), but its low basal activity (Fig. 4B) mandates caution in interpreting this potentially interesting result.
Amino acids with fixed-charge side chains play important roles in pH-dependent AE2 regulation by aa 336 -347 (20,24). We therefore tested possible regulatory contributions from AE2 residues Glu 399 , Asp 405 , and Lys 408 , by studying function of the individual Ala substitution mutants. All mutants displayed robust activity at pH o 7.4 (Fig. 4B), and wild-type pH o(50) values (Fig. 4C), and wild-type or near wild-type sensitivity to pH i (Fig. 4D). Overall, the experiments highlight aa 403-408 as important for pH o -dependent but not pH i -dependent regulation of AE2. However, the individual Ala substitutions studied did not suffice to alter AE2 regulation by pH.
Role of a Predicted Side Chain Interaction in the Regulation of AE2-mediated Cl Ϫ Transport by pH o and pH i -AE2 mutants E346A and H360E each altered pH-dependent regulation of AE2 (Fig. 3) (20). Structural modeling of the AE2 NH 2 -terminal cytoplasmic domain based on the corresponding AE1 structure (10) revealed Ͻ5-Å separation between the side chains of residues Glu 346 and His 360 (Fig. 5A). Therefore, we investigated the functional importance of this putative intra-monomeric interaction by creation of the paired mutations E346H, H360E, and the double mutant. Fig. 5B shows that all mutants displayed sufficient activity to allow analysis of 36 Cl Ϫ efflux during changes of pH o and pH i . The pH o(50) values of mutants E346H and H360E were each acid-shifted compared with wildtype AE2 (Fig. 5D). However, this altered phenotype was not rescued in the AE2 double mutant E346H/H360E, as shown in Fig. 5, C and D.
The AE2 single and double mutants were also assessed for altered regulation by pH i (Fig. 5E). Whereas wild-type AE2 exhibited ϳ20% 36 Cl Ϫ efflux activity at low pH i (in the presence of butyrate), AE2 mutants E346H and H360E each retained ϳ80% efflux activity. Again, however, the double mutant E346H/H360E failed to rescue the altered phenotype. We have shown previously that single amino acid substitutions within the highly conserved region aa 336 -347 that alter pH i sensitivity of Cl Ϫ /Cl Ϫ exchange similarly alter the pH sensitivity of Cl Ϫ /HCO 3 Ϫ exchange (20). We therefore examined pH i sensitivity of Cl Ϫ /HCO 3 Ϫ exchange mediated by AE2 H360E and the double mutant E346H/H360E (Fig. 5F). The 36 Cl Ϫ efflux activity of both mutants in the presence of butyrate was significantly higher than for wild-type AE2. Thus, the presence of HCO 3 Ϫ did not allow the double mutant to rescue the altered pH i sensitivity of the two individual mutants. The combined data fail to support a functional role for Glu 346 /His 360 interaction in determining the pH sensitivity of AE2. The data do, however, confirm that AE2 structure-function relationships derived from measurements of pH i sensitivity of Cl Ϫ /Cl Ϫ exchange in Xenopus oocytes apply equally to the pH i sensitivity of AE2-mediated Cl Ϫ /HCO 3 Ϫ exchange.

Role in AE2 Regulation of a Residue Predicted in AE1 to
Control pH-dependent Conformational Change-Mutation of the conserved hAE1 NH 2 -terminal cytoplasmic domain amino acid residues Glu 291 to Gln, Arg, or Leu abolishes the pH-dependent change in Stokes radius 2 normally observed with both natural and recombinant AE1 cytoplasmic domain dimers. This phenotype most likely arises by alteration of the hinge angle between the large globular domain and the dimerization arm (25), as revealed by the crystal structure of the hAE1 NH 2 -terminal cytoplasmic domain oligomer (10). We hypothesized that a similar pH-dependent movement around a homologous hinge region might contribute to pH-dependent regulation of AE2-mediated anion exchange. We therefore tested the effect of the corresponding AE2 mutation D578K on pH-sensitive anion exchange. AE2 D578K had wild-type activity at pH o 7.4 (Fig. 6B), and exhibited a modestly acid-shifted pH o(50) value of 6.46 Ϯ 0.07 (n ϭ 16, Fig. 6D). In addition, AE2 D578K exhibited only minimal sensitivity to pH i , retaining nearly 60% activity in the presence of butyrate (Fig. 6E). Thus, a mutation in AE2 which, in the corresponding region of AE1 changes the pH-dependent dimer structure, is also involved in the pH sensitivity of the 36 Cl Ϫ transport.
Because the carboxylate of AE2 D578 is modeled to reside 7 Å from the ⑀-amino group of the pH-sensitive Lys 344 (Fig. 6A), we tested the hypothesis that a mid-range electrostatic interaction between Asp 578 and Lys 344 might contribute to basal AE2 function or to its regulation by pH. The AE2 mutation K344D exhibited a substantial acid shift in pH o(50) with a value of 5.86 Ϯ 0.14 (n ϭ 13) (p Ͻ 0.05 compared with the wild-type value of 6.75 Ϯ 0.10, n ϭ 14) (Fig. 6D). However, the double mutant D578K/K344D, with its acid-shifted pH o(50) value of 5.91 Ϯ 0.06 (n ϭ 11), failed to rescue wild-type pH o sensitivity (Fig. 6, C and D).
A role for the putative Asp 578 /Lys 344 interaction was also tested in the regulation of AE2 by pH i (Fig. 6E). Whereas AE2 D578K displayed a moderately reduced sensitivity to pH i (ϳ60% activity in presence of 40 mM butyrate, compared with wild-type AE2 (ϳ20% activity, n ϭ 17), the K344D mutant was completely insensitive to pH i . The double mutant K344D/ D578K was similarly insensitive to pH i not only in the assay of Cl Ϫ /Cl Ϫ exchange (Fig. 6E), but also in the assay of Cl Ϫ /HCO 3 Ϫ exchange (Fig. 6F). Thus, the absence of a phenotypic second site reversion in the pH regulation profiles of AE2 D578K/ K344E fails to support a role for electrostatic interaction between these two nearby residues in the regulation of AE2 by pH.
Modeling the Non-contiguous Amino Acids Required for pH o and pH i Regulation of AE2 within the Putative Structure of AE2 aa 317-623-We have modeled the mouse AE2 NH 2terminal cytoplasmic domain (Fig. 7) based upon the crystal structure of the hAE1 NH 2 -terminal cytoplasmic domain (10). Although the structured hAE1 aa 55-356 correspond to mAE2 aa 317-650, the DeepView program yielded a model that omitted the poorly conserved AE2 residues 624 -650. The ␤ sheet, ␤1 (in the Zhang et al. (10) nomenclature), is predicted to begin soon after AE2 aa 317. Amino acid residues shown experimentally to be important for regulation of AE2-mediated Cl Ϫ transport by pH o or pH i (or both) are mapped onto the putative structure in ribbon format (Fig. 7A), and in space-filling format (Fig. 7B). Fig. 7B shows that several of the pH-sensitive residues identified in the current study, including His 317 , Glu 318 , Phe 320 , Trp 356 , and His 360 form, together with portions of the 2 J. Zhou and P. S. Low, personal communication.

pH-dependent Regulation of Anion Exchanger AE2
previously studied aa 336 -347, a contiguous surface. The pHsensitive Asp 578 is immediately adjacent to this surface. In contrast, the less intensively studied pH-sensitive regions of aa 403-408 (in pink) and 397-402 (out of view at lower right) are located on the other side of this model of part of the AE2 NH 2 -terminal cytoplasmic domain (Fig. 7B). The sequence alignment of Fig. 7C reveals strong conservation of pH-sensitive AE2 residues in the closely related Na ϩ -independent Cl Ϫ / HCO 3 Ϫ exchanger, AE3, and throughout the SLC4 gene family, including Na ϩ -dependent bicarbonate transporters not known to transport chloride. Although Na ϩ -HCO 3 Ϫ cotransport in cardiomyocytes is regulated by pH independently of HCO 3 Ϫ concentration (26), acute regulation by pH of individual recombinant Na ϩ -dependent HCO 3 Ϫ transporters has yet to be reported. The conservation among NBC cotransporters of amino acid residues implicated in regulation of AE2 and AE3 activities by pH strongly suggests similarity in mechanism of regulation. Ϫ -buffered medium, from (n) oocytes expressing wild-type or mutant AE2. Asterisks (*) indicate p Ͻ 0.05 compared with wild-type AE2.

DISCUSSION
The current work expands considerably the range of mutations in the AE2 NH 2 -terminal cytoplasmic domain known to alter pH i and pH o sensitivity of anion transport. The importance of this domain in conferring pH sensitivity has already been documented, with particular emphasis on aa 336 -347, the most highly conserved portion of the domain among bicarbonate transporters in the SLC4 superfamily (19,20). Less well conserved NH 2 -terminal cytoplasmic regions have also been implicated, including aa 318 -323 (20). Additional residues influencing pH sensitivity reside in the AE2 transmembrane domain (13,19).
Our present work highlights NH 2 -terminal cytoplasmic domain residues that influence pH sensitivity, but which are non-contiguous with the highly conserved aa 336 -347 region. Mutation of these residues decreases pH sensitivity of AE2-FIG. 6. Role of residues with predicted importance in pH-sensitive conformational sensing in the regulation of AE2-mediated Cl ؊ transport by pH. A, structure of AE2 aa 317-610 as modeled on the template of the crystal structure of the AE1 NH 2 -terminal cytoplasmic domain (see "Materials and Methods"). The predicted proximity of functionally important Lys 344 and to Asp 578 (predicted to be of conformational importance) suggested a possible contribution of their electrostatic interaction to regulation of AE2 by pH. B, 36 Cl Ϫ efflux rate constants for (n) oocytes measured at pH o 7.4 in oocytes expressing wild-type or mutant AE2 polypeptides (mean Ϯ S.E.). C, regulation by pH o of normalized 36 Cl Ϫ efflux from oocytes expressing wild-type AE2 (filled circles) and AE2 K344D/D578K (filled squares). Values are mean Ϯ S.E. D, pH o(50) values of wild-type and mutant AE2 polypeptides. E, normalized 36 Cl Ϫ efflux rate constant (ϮS.E.) in the presence of 40 mM butyrate in HEPES-buffered medium from (n) oocytes expressing wild-type AE2 or the indicated mutants. F, normalized 36 Cl Ϫ efflux rate constant (ϮS.E.) in the presence of 40 mM butyrate in Cl Ϫ -free HCO 3 Ϫ -buffered medium from (n) oocytes expressing wild-type or mutant AE2 polypeptides. Asterisks (*) indicate p Ͻ 0.05 compared with wild-type AE2.

pH-dependent Regulation of Anion Exchanger AE2
mediated Cl Ϫ transport. It is notable, therefore, that homology modeling based on the structure of the corresponding region of human AE1 places many of these newly studied residues at the domain surface immediately adjacent to residues aa 336 -347. Together, these clustered residues delineate a localized surface of the NH 2 -terminal cytoplasmic domain associated with AE2 regulation by pH i and pH o . Furthermore, in select examples, the structural requirements for pH-dependent regulation of AE2-mediated Cl Ϫ /Cl Ϫ exchange have been shown to apply equally to regulation of Cl Ϫ /HCO 3 Ϫ exchange.

Role of Charged and Titratable Amino Acid Residues in AE2 Regulation by pH
Histidine-Exofacial His protonation controls the pHdependent activity of many ion channel polypeptides (27)(28)(29).
In the case of acid/base transporters, mutation of two juxtamembrane His residues of the COOH-terminal intracellular cytoplasmic domain of the NHE3 Na ϩ /H ϩ exchanger acid shifts its pH i sensitivity (30). This result suggests that His residue(s) may also be important for pH sensing by the AE2 protein. In the present work, individual Ala substitutions of AE2 His 314 or His 317 each acid-shifted the pH o sensitivity but were without effect on the pH i sensitivity. Protonation of these His residues, however, is unlikely to account for pH o sensitivity, as the double mutant H314A/H317A had no effect (Fig. 1). This pattern contrasts with the consequences of His mutagenesis in the NH 2 -terminal and COOH-terminal cytoplasmic tails of the K ϩ channel Kir1.1. In that protein, although two His residues of the NH 2 -terminal tail and one of the 6 His residues in the COOH-terminal tail are dispensable for maintenance of pH ). The structural model (above) and the linear schematic (below) each indicate residues that when mutated alter AE2 regulation by pH i (yellow), pH o (red), or by both pH i and pH o (orange). B, space filling structure of AE2 aa 317-610. Surface amino acid residues are marked with the same yellow-red-orange color scheme. Pro 610 (blue) is the most COOH-terminal surface residue in this view. AE2 aa 403-408 are at the bottom in pink. Mutation en bloc altered sensitivity only to pH o , but individual mutations with the same effect have yet to be identified. AE2 aa 397-402 are located out of view on the far side at the bottom right, adjacent to aa 403-408. Leu 323 is modeled to be not at the domain surface. C, the sequences of mouse AE2 aa 312-323, 336 -347 (see Stewart et al.,20), 356 -362, and 397-408 are aligned with corresponding regions of other SLC4 anion transporters. Boxes mark conserved residues whose mutation alters regulation of AE2-mediated Cl Ϫ transport by pH, with the color code as in A and B, the EMBL/GenBank/DDBJ accession number for these sequences are: mAE2 (murine anion exchanger 2; J04036), mAE3 (murine anion exchanger 3; AAA40692), mNCBE (murine sodium-dependent chloride/bicarbonate exchange; BAB17922), hNDCBE1 (human sodium-dependent chloride/bicarbonate exchange 1; AF069512), DmNDAE1 (Drosophlia melanogaster sodium-dependent anion exchange 1; AF047468), hNBCe2 (human sodium bicarbonate cotransport 2; AF293337), hNBCe1 (human sodium bicarbonate cotransport 1; AF007216), rNBCn1 (rat sodium bicarbonate cotransport 1; AF070475), hAE4 (human anion exchanger 4; AF332961), hAE1 (human anion exchanger 1; CAA31128), mAE1 (murine anion exchanger 1; J02756), trAE1 (trout anion exchanger 1; Z50848), hBTR1 (human bicarbonate transporter 1; AF336127). hBTR1 lacks evident similarity with these AE2 segments (alignment as in Parker et al., 2001) (37).

pH-dependent Regulation of Anion Exchanger AE2
sensitivity, individual mutation of four of the COOH-terminal tail His residues attenuates pH sensitivity, and mutation of all four residues nearly completely abolishes it (27).
In contrast to the complex effects of His 314 and His 317 mutations in AE2, the single H360E mutation exerted a dramatic effect on both pH i and pH o sensitivity. His 360 resides within the more highly conserved stretch of aa 357-362. Its mutation replicated the effects of the A (6) 357-362 hexa-Ala bloc substitution in acid shifting the pH o versus activity curve of AE2, and in nearly completely abolishing the regulation of transport by pH i (Fig. 3). This is the largest attenuation of pH sensitivity yet observed for an individual mutation in the AE2 NH 2 -terminal cytoplasmic domain. His 360 may thus form part of the "pH modulator" site of AE2. Homology modeling predicts surface exposure of His 360 on this portion of AE2. The functional consequences of the H360E mutation may reflect a predicted side chain rotation and displacement that changes the shape as well as the electrostatics of the local protein surface.
Glutamate-In addition to histidine, mutation of other NH 2terminal charged residues can influence the pH sensitivity of AE2. Most notable among these are the two glutamate residues Glu 346 and Glu 347 (within the highly conserved region, aa 336 -347) that have previously been shown to be important for conferring pH sensitivity (20). We now report that although other individual glutamate AE2 mutants, namely E318A and E322A, exhibited essentially wild-type pH sensitivity, the double mutant exhibited an acid-shifted pH o dependence and a loss of pH i sensitivity (Fig. 2). Individually mutating other glutamate residues to alanine (Glu 350 , Glu 351 , Glu 352 , and Glu 354 ) was without effect (Supplemental Materials Table I), despite the fact that these residues are close to the important Trp 356 highlighted in Fig. 7 (see also Ref. 20). Similarly, the individual mutant E399A did not replicate the altered pH regulation of the AE2 bloc mutant A (6) 397-402 (Fig. 4). In summary, of 10 glutamate residues mutated so far in the NH 2 -terminal region, no more than four seem to play a role in setting the pH sensitivity of the transporter.
Aspartate and Lysine-We had hypothesized that charged residues Asp 405 and Lys 408 may be responsible for the acidshifted pH o(50) value of AE2 hexa-Ala bloc substitution mutant A (6) 403-408 as their side chains are predicted to face outward (see Fig. 7). However, individual mutation of each residue of alanine was again without effect on regulation of AE2 by pH o or pH i , suggesting that neither residue acting individually mediates pH o sensitivity of AE2. It remains to be investigated whether the double charge mutation or individual mutation of one of the uncharged residues in this region might contribute to pH o regulation of AE2.
Mutation of human AE1 residue Glu 291 abolishes pH-dependent changes in Stokes radius of the soluble recombinant AE1 NH 2 -terminal cytoplasmic domain dimer. 2 In view of this, we tested the possibility that mutation of the corresponding AE2 residue, Asp 578 , might modulate or abolish pH sensitivity of AE2-mediated Cl Ϫ transport. Indeed, AE2 D578K exhibited a significant acid shift in pH o(50) value, along with significant reduction in pH i sensitivity (Fig. 6). Thus, our results may reflect a similar pH-dependent shape change in AE2 that is associated with a regulatory change in anion transport.
Tests of Charge-Pair Interaction within AE2-Because AE2 D578K exhibited altered pH sensitivity, we examined the modeled AE2 structure for residues that might interact with Asp 578 . Nearest neighbor search analysis (DeepView; see "Materials and Methods") estimated a distance of ϳ7 Å between Asp 578 and Lys 344 , whose mutation we earlier showed to alter pH sensitivity of AE2 (Fig. 6A). This distance is within the 8 -9-Å range of stabilizing electrostatic interactions detected in a survey of many protein structures (23). We therefore hypothesized that a charge reversal experiment with the AE2 double mutant D578K/K344D might rescue the wild-type pH sensitivity lost in the individual mutants, but this was not the case (Fig. 6, D-F).
A nearest neighbor search was also performed for His 360 , whose mutation produced such a large alteration in the pH dependence of AE2 (see above). This revealed within a distance of 4 Å the residue Glu 346 that, when changed to Ala, selectively altered regulation by pH o (20). The individual mutations H360E and E346H indeed each altered regulation of AE2 by both pH o and pH i (Fig. 5, C-F), in CO 2 /HCO 3 Ϫ -buffered as well as in HEPES-buffered conditions. Again, however, the AE2 double mutant E346H/H360E failed to restore wild-type regulation by pH. Thus, initial analysis of two possible intramolecular electrostatic side chain interactions fails to provide functional evidence for their importance, perhaps reflecting a structural rather than a regulatory role for the interactions. Alternatively, it might reflect inaccuracies in the AE2 structure predicted by homology modeling, such that other residues near His 360 and Asp 578 might be more appropriate targets for mutagenesis.

Role of Non-charged Residues in AE2 Regulation by pH i
Proline-Although charged, proton-titratable residues are attractive candidates for a pH sensor of the transporter it is clear that, for AE2, non-charged residues must also play a role (20). The proline-rich AE2 sequence PRARPRAPHKPHEVF (aa 306 -320) meets "moderate stringency" criteria as a potential binding site for SH3 domains of cortactin or phosphatidylinositol 3-kinase p85 subunit (scansite.mit.edu), and so proline residues may conceivably play a role in the biochemical regulation of AE2 activity. Although the modestly conserved AE2 aa 312-317 do not correspond to a structured region of the crystallized AE1 cytoplasmic domain (10), proline mutagenesis was used to test for possible involvement of an SH3-binding protein in the pH-dependent regulation of anion transport. However, neither basal activity nor pH-regulated activity of AE2 was altered by individual or tandem Ala substitution of Pro 313 and Pro 316 . In contrast, the two proline-rich regions of the NHE2 Na ϩ /H ϩ exchanger are known to bind in vitro to a range of SH3 domains, although NHE1, NHE3, and NHE4 lack this property (31). Mutation of both proline-rich regions in NHE2 is reported to alter the polarity of sorting of the transport protein in epithelial cells, although it does not alter the kinetic regulation of ion transport (31). In contrast, AE2 interactions with SH3 domain-containing proteins have yet to be reported, and the phosphatidylinositol 3-kinase inhibitor wortmannin is without effect on AE2-mediated rates of 36 Cl Ϫ efflux at pH o 7.4 (24). Overall there is, to date, no compelling evidence for a role of the NH 2 -terminal cytoplasmic domain prolines or SH3 domainbinding proteins in the regulation of AE2 by pH.
Other Hydrophobic and Aromatic Residues-In the present work, individual mutation of AE2 aa 318 -323 has identified the conserved residue Phe 320 as important in AE2 regulation by pH. Mutation of Phe 320 to Ala, or to the corresponding mouse AE1 residue Tyr, acid-shifted the pH o versus activity profile and substantially decreased pH i sensitivity. The mutation L323A produced similar alterations in AE2 regulation by both pH o and pH i (Fig. 2). Phe 320 and Leu 323 are predicted to project their side chains to opposite sides of the "␤1 sheet" as suggested by homology modeling with AE1, with Leu 323 buried and Phe 320 exposed at the surface (see Fig. 7). Both substitutions of Phe 320 are likely to alter the size of the hydrophobic surface patch to which the phenyl side chain of Phe 320 is normally predicted to contribute, a feature that may lead to the observed reduction in pH sensitivity. The predicted surface location of Phe 320 suggests that its phenol group may interact with either the COOH-terminal transmembrane domain or directly with the lipid bilayer.
Within aa 348 -356 immediately adjacent to the highly conserved AE2 aa 336 -347, only the mutation W356A substantially decreased AE2 pH i sensitivity (Fig. 3). This Trp residue is also highly conserved within the SLC4 gene family (Fig. 7). AE2 homology modeling predicts partial surface exposure of the Trp 356 indole, where it is surrounded within a putative 5-Å radius by multiple glutamates and arginines (see "Materials and Methods"). The W356A substitution predicts a large reduction in the hydrophobic patch area, accompanied by rearrangement of surrounding charged residues. It is therefore of interest that, in the membrane-spanning gramicidin channel, tryptophan side chains near the lipid-water boundary are oriented away from the aqueous ion pore and serve to tune the proton conductance of the channel (32).

Larger Scale Structural Considerations in the Regulation of AE2 by pH
The involvement of NH 2 -terminal cytoplasmic domain amino acids in the regulation of AE2 by pH i is more readily explained than their involvement in regulation by pH o . In the latter case, direct protonation by extracellular H ϩ is not a possibility. Mutations of AE2 NH 2 -terminal cytoplasmic domain residues might alter the pH o sensitivity of AE2 by transmission of a local conformational change across the bilayer to residues accessible to extracellular solvent. This would represent a type of static "inside-out signaling," akin to that exhibited by extracellular matrix receptors. Some of the pH regulation-associated residues may be important in maintaining the structural integrity of the NH 2 -terminal cytoplasmic domain. This could be achieved through intra-monomeric or inter-monomeric interactions. For example, a conformational change accompanying a mutation might indirectly alter the conformation of the AE2 substrate binding pocket, or the conformation of a portion of the transmembrane domain required for anion translocation, thereby altering the pH sensitivity of the transporter. A mutation may also alter the pK of other critical residues whose protonation/deprotonation controls a local or larger scale conformational change within the NH 2 -terminal cytoplasmic domain. In addition, the integrated control of anion transport by pH is likely to depend on interactions between the surface of the NH 2 -terminal cytoplasmic domain and the cytoplasmic vestibule of the anion translocation pathway. Alternatively there may be interaction between the NH 2 -terminal cytoplasmic surface and transmembrane domain structures that remotely stabilize the transport pathway.

Non-contiguous NH 2 -terminal Cytoplasmic Domain Amino Acids Cluster to Form a Surface That Contributes to AE2 Regulation by pH
The finding that multiple, non-contiguous regions of the NH 2 -terminal cytoplasmic domain of AE2 are involved in defining the pH i and pH o sensitivity of AE2 has similarities with the regulation by pH of potassium channels (33,34) and the Na ϩ /H ϩ exchanger, NHE1 (35,36). In those proteins, noncontiguous amino acids were proposed to be involved in pH gating, or to constitute part of the "pH sensor." In the present work, many of the residues that influence the pH sensitivity of AE2 are predicted to form a single, localized surface on the cytoplasmic domain. This surface cluster of 12 residues includes 7 residues with charged side chains. The relatively high charge density of this area may be affected by local changes of pH. Thus, this surface may modulate the sensing of pH by the transmembrane anion-translocating portion of the AE2 protein.

Conservation of Amino Acid Residues Important in pH Regulation of AE2
As shown in Fig. 7 many of the residues newly identified as important for normal regulation of AE2-mediated Cl Ϫ transport by pH o and pH i are conserved among other SLC4 bicarbonate transporters. The degree of conservation of a region or of individual residues may reflect the functional importance to pH-dependent regulation of related transporters. For example, the AE3 glutamate residues that correspond to the functionally important Glu 346 and Glu 347 of AE2 are also important for the pH i -dependent modulation of AE3 (20). Similarly, the cytoplasmic subdomains found to be important in the pH i dependence of NHE1 activity are comparably conserved in other sodium/hydrogen exchangers (35).
In summary, the results of the present study suggest the presence of a localized pH-sensitive surface on the AE2 NH 2terminal cytoplasmic domain that contributes to the pH o and pH i dependence of AE2-mediated anion transport. This highlights the complexity of the pH sensing apparatus. In addition to structures within the NH 2 -terminal cytoplasmic domain, identification of other pH sensing structures within the COOHterminal transmembrane domain, as well as the anion transport sites themselves, will be required to elucidate more completely the molecular mechanism of AE2 regulation by pH. Experiments toward this end are underway.