Evidence for up-regulation of the endogenous Na-K-2Cl co-transporter by molecular interactions with the anion exchanger tAE1 expressed in Xenopus oocyte.

Expression of trout anion exchanger 1 (tAE1) in Xenopus oocyte led to the stimulation of a Na(+)- and Cl(-)-dependent Rb influx. Functional features and pharmacological data strongly suggest that this Rb influx is mediated by the endogenous Na-K-2Cl (NKCC) co-transporter. The functional relationship between expression of tAE1 and activation of the NKCC co-transporter was investigated. Indeed, it was shown previously that tAE1 expressed in Xenopus oocyte induces a strong anion conductance which is correlated with an increased taurine permeability. Measurements of intracellular ion contents ruled out the involvement of any modification of known electrochemical parameters in NKCC co-transporter activation by tAE1. Furthermore, using chimera of tAE1 made with AE1 from other species unable to exhibit anion conductance led to the conclusion that there was no correlation between tAE1 anion conductance and NKCC co-transporter stimulation. Therefore, a possible molecular interaction between tAE1 and the NKCC co-transporter was investigated. Our results clearly show that NKCC activation is dependent upon the C-terminal part of tAE1. Chimeric constructions where tAE1 C-terminal part was substituted by the corresponding part of mouse AE1 abolished co-transporter activation. Moreover, steric encumbrance on the C-terminal end of tAE1 with a specific antibody or with a protein fusion also prevented the co-transporter activation. These data suggest a new role for some anion exchangers in controlling other transporter activity by molecular interactions.

Membrane permeability is an essential property that governs cell physiology. Concentration of intracellular osmolytes and, therefore, of intracellular enzyme activities, specific cell functions, and metabolism depends on cell membrane permeability. To ensure control of intracellular medium, the functioning of membrane transporters must be regulated and coordinated.
Regulation of membrane permeability could be achieved by direct modifications on transporters such as phosphorylation by means of kinase or phosphatase action that could change transport features (1). Another regulation possibility is obtained by interactions between transporters. These interactions could be functional, which means that the functioning of one transporter may create electrochemical conditions favorable to stimulate the functioning of other transporters. Many examples of such functional interactions between membrane transporters are known (for instance, electric coupling between channels, thermodynamic coupling between Na-K-2Cl (NKCC) 1 co-transporter and KCl transporter (2) or autocrine mechanism; Refs. 3,4). But transporter coupling could also be by means of direct protein-protein interaction. Interest in such interactions has been focused on the example of CFTR (cystic fibrosis transmembrane conductance regulator). Different studies report possible molecular interaction of the CFTR with other membrane transporters, such as the Na ϩ channel ENaC (epithelial Na ϩ Channel) (5)(6)(7)(8) and possibly with the Cl Ϫ / HCO 3 Ϫ exchanger (9) (in this last example, stimulation of members of the SLC26 family through molecular interactions with the CFTR is only documented by co-immunoprecipitation experiments). To our knowledge, CFTR is the only example of transporter susceptible to regulate other transporters by protein-protein interactions.
In the present paper, we provide another example of membrane transporter regulation by direct protein interactions that are not related to transport functions. It concerns interaction between the trout anion exchanger, tAE1, and the NKCC cotransporter. This protein catalyzes the simultaneous transport of one Na ϩ with one K ϩ and two Cl Ϫ across cell plasma membrane. It can be detected by measuring a Na ϩ -and Cl Ϫ -dependent K ϩ influx. Moreover, this co-transport is inhibited by micromolar concentrations of bumetanide. Two different isoforms of this co-transporter are known (NKCC1 and NKCC2), and it has been extensively studied in a great number of cell types (for review, see Ref. 10). tAE1 is the anion exchanger of trout erythrocytes. Anion exchangers (AEs) are transmembrane proteins catalyzing the electroneutral exchange of 1 Cl Ϫ for 1 bicarbonate. Up to now, three different genes, named slc4a1, slc4a2, and slc4a3, have been known, each of them coding for different polypeptide products, depending on the splicing and transcription initiation site (for reviews, see Refs. 11,12). All of the AE polypeptides can be divided into two main domains of about the same size: an N-terminal cytoplasmic domain interacting with cytoskeleton and a membrane spanning domain, which is responsible for ion translocation, with a short C-terminal end in the cytoplasm. By expression studies of tAE1 in Xenopus oocytes, we have demonstrated previously that this normally electroneutral anion exchanger is able to form an anion conductive path-way permeable to different charged or neutral osmolytes such as urea, choline, sorbitol, Na ϩ , and K ϩ (13)(14)(15). In physiological conditions, these transport properties of tAE1 are activated by a decrease in intracellular ionic strength in swollen erythrocytes, and they are involved in the cell regulatory volume decrease response (16,17). Here, we focused on another unexpected feature of this anion exchanger. Independently of its own transport ability, this exchanger is able to stimulate the endogenous NKCC co-transporter in Xenopus oocyte. We show that this stimulation is not linked to tAE1 activity but rather to the interaction between the C-terminal end of tAE1 and the NKCC co-transporter. This interaction could involve a direct contact between the two transporters, or it could involve intermediary regulatory proteins. Even though this interaction between tAE1 and the NKCC co-transporter was observed in Xenopus oocyte, it is likely that such interactions could take place in physiological conditions to coordinate transporter activity and regulate cell membrane permeability.
Taking of Oocytes-Xenopus laevis were cooled on ice with MS222 until completely anesthetized and were kept covered with ice during the surgery, according to the procedure recommended by our ethics committee. The surgery consisted of removing about five ovarian lobes containing oocytes. After surgery, the animals were placed in cold water between 0 and 4°C to recover from anesthesia, monitored for 3 hours, and then placed back in their aquaria.
Oocyte Injection-Collected oocytes were washed in modified Barth's saline (MBS) (85 mM NaCl, 1 mM KCl, 2.4 mM NaHCO 3 , 0.82 mM MgSO 4 , 0.33 mM Ca(NO 3 ) 2 , 0.41 mM CaCl 2 , 10 mM HEPES, 4 mM NaOH, pH 7.4, supplemented with 10 units/ml penicillin and 10 g/ml streptomycin). After washing with MBS, defolliculation was obtained by overnight incubation at 19°C in MBS containing between 0.8 and 1.3 mg/ml collagenase (Serva) corresponding to 1 unit/ml, followed by a 30-min incubation in Ca 2ϩ -free MBS. Stage V-VI oocytes were then injected with 50 nl of 100 ng/l cRNA and maintained at 19°C in MBS. This cRNA concentration was found to induce maximal expression of the corresponding proteins in oocyte.
Influx Measurements-Cl influx measurements were done as described previously (13,15). Rb influx measurements were done in media containing 10 Ϫ4 M ouabain to avoid looking at K ϩ movements through the Na-K-ATPase. Before incubation with radioactive medium, oocytes were quickly washed three times with 1.5 ml of radioactive-free medium containing ouabain 10 Ϫ4 M to adapt the oocyte to the eventually different influx conditions. Eight oocytes were incubated at 19°C in 80 l of influx medium containing 86 Rb used as a radioactive substitute for K (Amersham Biosciences) with a specific activity of 36,000 dpm/nmol potassium. After 60 min of incubation in radioactive medium (during the linear phase of the uptake kinetic), the oocytes were washed twice in 8 ml of ice-cold MBS and transferred individually into counting vials. Radioactive Rb uptake in each oocyte was determined after scintillation counting with external standard. The incubation medium was counted in duplicate on 5-l aliquots, using the same protocol to determine the specific activity in each experiment. Rb influx was calculated as the mean of the eight values and expressed as pmol/hour per oocyte. When we compared Rb influx between isosmotic medium (210 mosM) and hyperosmotic medium (270 mosM), both media had the same ion concentrations. Hyperosmotic medium was obtained by adding 60 mosmol of saccharose per liter of MBS. Osmolarity of the media was controlled by measuring osmotic pressure with a Wescor vapor pressure osmometer. The NO 3 Ϫ -containing MBS composition was 88 mM NaNO 3 , 2.4 mM NaHCO 3 , 1 mM KNO 3  "Control oocytes" refers to non-injected oocytes. In previous experiments, non-injected oocytes were compared with water-injected oocytes, and no difference in Cl or Rb influx between the two groups had been observed. The influx experiments were done 1 or 2 days after oocyte injection (see legends to Figs. 1 and 4). No great difference in Cl as well as Rb influxes was observed between 1 and 2 days of protein expression. Thus, data presented that were obtained at day 1 or 2 after injection are comparable.
Determination of Ion and Water Concentrations of Oocytes-To determine the intracellular Cl concentrations, oocytes were incubated in 36 Cl containing MBS up to the equilibrium of specific radioactivity (363 dpm/nmol of Cl). By counting intracellular 36 Cl dpm accumulated, it was possible to deduce the intracellular Cl concentration.
Na ϩ and K ϩ contents were measured as described previously (15). Intracellular ions were extracted from dried oocytes in 4 ml of milliQ water overnight at 4°C. Measurements of sodium and potassium were done with a flame photometer (Eppendorf). Results are expressed in mM (see Table I). The oocyte water content was measured by weighing a group of five oocytes after a quick wash in milliQ water (to remove extracellular ion contaminants) containing 3 H-inuline (to be able to determine the extracellular water volume). The oocytes were weighed FIG. 1. Hyperosmotically stimulated NKCC co-transporter in Xenopus oocyte. A, the Cl Ϫ -dependent Rb influx was measured in control oocytes (non-injected) 1, 2, or 5 days after oocyte collection in isosmotic MBS (white bars) or hyperosmotic MBS (striped bars), both media having the same ion concentration and containing 10 Ϫ4 M of ouabain. These data are the means of eight oocytes individually counted and then averaged. This experiment is one representative experiment of three. B, the Rb influx was measured in non-injected oocyte either in control isosmotic media (Ctrl) or in hyperosmotic media (Hyper). The oocyte incubation media were Cl-containing MBS (white bars), NO 3 Ϫcontaining MBS (dotted bars), or N-methyl-D-glucamine-containing MBS (striped bars). The effect of 10 Ϫ6 M bumetanide was assessed in hyperosmotic MBS (containing Cl). These data are the means of eight oocytes individually counted and then averaged. This experiment is one representative experiment of three.
wet and then dry (they were dried overnight at 80°C). Subtraction of the dry weight from the wet weight after correction for extracellular water trapped between the five oocytes, gives the water content of five oocytes. Assuming that the five oocytes are equivalent, we deduced the intracellular water content of an oocyte to be around 500 nl.
Electrophysiology-Electrophysiological parameters were measured as described previously (13) using the two-electrode voltage clamp technique with a TEV 200 amplifier (Dagan, Minneapolis, MN) monitored by computer through Digidata 1200 A/D converter PC clamp software (Axon Instruments Inc., Foster City, CA). Measurements were done in MBS at room temperature.
Chimera Construction-Except for TB3-gyraseB, all other chimera constructions are described in previously published papers (13). 2 The chimera tAE1-gyraseB was made by PCR with plasmids pSP64tAE1 and pcDNA3Raf-gyrB. This last plasmid was a kind gift from Dr. Michael Farrar (19). tAE1 was amplified with primers 5Ј-tAE1 (AAGCTT-GGGCTGCAGGTCGAG) and 3Ј-tAE1 (GCGGCCGCCTGGCAACGGG-GACTCGTA). The gyrase B was amplified with primers 5Ј-GyrB (GG-CGGCCGCAGCAATTCTTATGAC) and 3Ј-GyrB (GAGCTGTCAACTA-GTGTCGACTTT). The primers 3Ј-tAE1 and 5Ј-gyraseB have an overlapping part of nine nucleotides. 100 l of PCR contained 100 ng of tAE1 or gyraseB plasmids, 200 M 2Ј-deoxynucleoside 5Ј-triphosphate (dNTP), 0.4 M of each primer, 1 l of VentDNA polymerase (NEBiolabs); denaturation was for 2 min at 94°C, followed by 30 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 3 min. The PCR products tAE1 (2.7 kb) and gyraseB (0.7 kb) were then diluted 10 times and mixed in a secondary PCR to amplify the fusion tAE1-gyraseB with primers 5Ј-tAE1 and 3Ј-gyrB in a final volume of 100 l containing 1 l of the two 10ϫ-diluted first PCR products, 200 M dNTP, 0.4 M of each primer, 1 l of Expand high fidelity DNA polymerase (Roche Applied Science). The amplification protocol was the same as for the first PCR. This yielded a 3.4-kb product corresponding to the fusion tAE1-gyraseB that was cloned in pGEMT-easy vector (Promega). Then tAE1-gyrB was subcloned in pSP64polyA (Promega) using Fast-Link DNA ligase (Epicenter) and restriction enzymes from NEBiolabs. Top10FЈ competent cells (Invitrogen) were transfected, and positive recombinant pSP64polyA-tAE1GyrB containing plasmids were selected and analyzed.
Western Blot of Oocyte Membrane Proteins-Oocyte membranes were prepared by homogenization of 20 oocytes (control or injected) in cooled 20 mM Tris-HCl buffer, pH 7.4, with 0.5 mM dithiothreitol and 0.5 mM protease inhibitor (Pefabloc, Roche Applied Science). The mixture was centrifuged at 4°C at 2000 rpm, 4000 rpm, and finally at 6000 rpm in an Eppendorf tube; after each centrifugation, the pellet was discarded. Collected supernatant after the third centrifugation was ultracentrifugated at 65,000 rpm for 30 min at 4°C, and the membrane pellet was solubilized in the homogenization buffer. Protein concentration was measured with a Bio-Rad kit, and 50 g of proteins were then loaded per lane of SDS-PAGE electrophoresis gel. Western blot transfer was done with a semi-dry transfer system from Biometra on nylon membrane (Hybond C extra, Amersham Biosciences). The presence of tAE1 was detected by the antibodies Ab145 directed against a synthetic peptide corresponding to the last 15 amino acids of the C-terminal part of tAE1 or Ab146 directed against a synthetic peptide corresponding to the first 15 amino acids of the N-terminal part of tAE1 (Neosystem, Strasbourg). Immunoglobulins Ab145 and Ab146 were purified from rabbit serum by precipitation with caprilic acid as described by Reik et al. (20). The secondary antibody was a goat IgG-peroxidase (Sigma) that was detected by chemiluminescence (Super Signal West Pico, Pierce).
Chemicals and reagents were from Sigma, unless otherwise stated. Plasmid DNA preparations were done with Qiaquick Miniprep kit from Qiagen, and DNA purifications were done with Machery Nagel nucleospin extract kit.

RESULTS
The presence of the endogenous NKCC co-transport in Xenopus oocyte plasma membrane was characterized by measuring a K ϩ influx that is Cl Ϫ -and Na ϩ -dependent, and sensitive to micromolar concentrations of bumetanide (21,22). Fig. 1A illustrates the ouabain-insensitive K ϩ permeability in Xenopus oocytes between 1 and 5 days after oocyte collection in isos-FIG. 2. Stimulation of NKCC cotransporter by tAE1 expression. A, Rb influx was measured in control oocytes (non-injected, dotted bars) or tAE1-expressing oocytes (white bars) 1 day after injection in isosmotic MBS with 10 Ϫ4 M ouabain. Effect of 10 Ϫ6 M bumetanide and Cl Ϫ or Na ϩ removal was also measured on Rb influx in tAE1-expressing oocyte. The Rb influx in Cl Ϫ -free medium was done in isosmotic NO 3 Ϫ -containing MBS, as published previously (14). The Na ϩ -free medium was MBS in which N-methyl-D-glucamine was substituted for Na ϩ (see "Experimental Procedures" for media composition). B, Rb influx measurements in control oocytes (non-injected, dotted bar) or tAE1 expressing oocytes (white bars) 1 day after injection in hyperosmotic MBS with 10 Ϫ4 M ouabain. Effect of 10 Ϫ6 M bumetanide was measured on the Rb influx in tAE1-expressing oocytes. Means Ϯ S.E., n ϭ 3.
FIG. 3. tAE1 cRNA dose-dependent activation of NKCC cotransporter. Oocytes were injected with different amounts of tAE1-cRNA: 10, 100, and 1000 pg. In each case, Rb influx was measured 1 day after injection in MBS containing 10 Ϫ4 M ouabain and compared with control oocyte (non-injected). Means Ϯ S.E., n ϭ 8. motic or hyperosmotic media. As shown previously by others, this K ϩ permeability is largely Na ϩ -and Cl Ϫ -dependent because substitution of Na ϩ by N-methyl-D-glucamine or Cl Ϫ by NO 3 Ϫ abolished the Rb influx (Fig. 1B). Hyperosmotic stimulation of the NKCC co-transporter increased about 9ϫ the Rb influx of control oocytes. There is no significant difference in the activation of NKCC co-transporter by hyperosmolarity as a function of time. The co-transporter is present and can be activated by hyperosmolarity in oocyte plasma membrane for 5 days after oocyte collection. Moreover, the Rb influx stimulated by hyperosmolarity is highly sensitive to bumetanide as expected (Fig. 1B).
In the next experiments, it was observed that expression of the anion exchanger of trout erythrocyte, tAE1, stimulated a Cl Ϫ -dependent Rb uptake in Xenopus oocyte that was inhibited by bumetanide, by a NO 3 Ϫ substitution of Cl Ϫ , and by substitution of Na ϩ with N-methyl-D-glucamine ( Fig. 2A). This Rb influx is not additive with the Cl Ϫ -dependent Rb uptake stimulated by hyperosmolarity (Fig. 2B); it has the features of the Rb influx mediated by the NKCC co-transporter. In previous work, it was shown that tAE1 is functionally expressed in oocyte plasma membrane where it increases by 6-fold Cl influx of control oocytes as early as 24 h after injection (12). Moreover, it was observed that this Cl influx increase is proportional to the amount of tAE1-cRNA injected in oocytes, and therefore, it was considered as a relevant marker of tAE1 expression level in oocyte plasma membrane. To correlate activation of NKCC co-transporter to the presence of tAE1 in oocyte plasma membrane, Cl Ϫ -dependent, bumetanide-sensitive Rb uptake was measured as a function of the amount of tAE1-cRNA injected in Xenopus oocyte (Fig. 3). This experiment shows a good correlation between NKCC activity and tAE1 expression level.
It should be noticed that the activation of NKCC co-transporter was observed as early as 1 day after tAE1 injection in Xenopus oocyte; then, it persisted over time (Rb influx in tAE1expressing oocyte was 252 Ϯ 33 pmol per oocyte per h; n ϭ 7, 4 days after injection).
The anion exchanger of trout erythrocyte, tAE1, was shown to form an anion conductance permeable to taurine as well as to different organic and inorganic compounds when expressed in Xenopus oocyte (9,10). This large anion conductance imposes a resting potential that is close to the resting potential of Cl Ϫ (resting potential (Em) measured in tAE1-expressing oocyte 1 day after injection is Ϫ36.2 Ϯ 1.9 mV, and the equilibrium potential for Cl (E Cl ) calculated from data Table I is Ϫ35.6 mV). This property is not shared by other members of the AE1 family such as human AE1 (hAE1), mouse AE1 (mAE1), or skate AE1 (skAE1). It should be noticed that skAE1 expressed in Xenopus oocyte has the same transport characteristics as mammalian AE1: it is unable to transport taurine and unable to form an anion conductance. 2 Expression of these other AE1s did not induce any increase in the Cl Ϫ -dependent and bumetanidesensitive Rb influx (Fig. 4). However, these other AE1s induce a Cl exchange in oocyte that is similar to tAE1, suggesting that they are similarly expressed in plasma membrane (13). 2 It could be assumed that activation of NKCC co-transporter is due to the anion conductive activity of tAE1. Classical inhib-itors of tAE1 anion conductance such as NPPB or niflumic acid could not be used to distinguish tAE1 conductance and activation of the co-transporter because these two compounds also inhibit with high efficiency the NKCC co-transport stimulated by hyperosmolarity (5 ϫ 10 Ϫ4 M of NPPB inhibited 100 Ϯ 0.3% of the Cl-dependent Rb influx activated by hyperosmolarity (n ϭ 8) and niflumic acid at 5 ϫ 10 Ϫ4 M inhibited this Rb influx by 90 Ϯ 3%, n ϭ 8). So we used glybenclamide, inhibitor of the CFTR as well as other anion conductances (23,24). As illustrated in Fig. 5A, glybenclamide strongly inhibits the anionconductive activity of tAE1 without any significant effect on the tAE1-mediated activation of NKCC co-transport (Fig. 5B).
Change in intracellular Cl Ϫ concentration is known to activate the NKCC co-transporter (25). Oocyte ion concentrations were measured to check whether tAE1 expression induced any modification in Cl Ϫ concentration 1 day after injection, when activation of the co-transporter was measured. Table I shows that intracellular Na ϩ , K ϩ , and Cl Ϫ concentrations are equivalent between control oocytes and tAE1-expressing oocytes 1 day after injection. Thus, there is no variation of intracellular ion concentrations that could explain the activation of the co-transporter in tAE1-expressing oocytes when we did our measurements.
In the absence of an obvious functional relationship between tAE1 expression and activation of the endogenous NKCC cotransporter in Xenopus oocytes, stimulation by protein-protein interaction was considered.
To assess whether tAE1 is able to stimulate the NKCC co-transporter by protein-protein interactions, different chimeras were done between trout, mouse or skate AE1. Fig. 6A represents a diagram of the different constructions. All of these constructions are functionally expressed in Xenopus oocyte. Except for tAE1⌬N, they have the same Cl Ϫ transport features as does tAE1 (13); this finding suggests a similar amount of protein in oocyte plasma membrane. This possibility is con-

FIG. 4. Effect of different AE1 isoforms on oocyte Cl-dependent Rb influx.
Oocytes were injected with cRNA (5 ng/oocyte) of AE1 from different species: trout (tAE1), mouse (mAE1), or skate (skAE1), and the Cl-dependent Rb influx was measured 2 days later in the presence of ouabain. Control refers to non-injected oocytes. Means Ϯ S.E, n ϭ 3. firmed by Western blot analyses of oocyte membrane proteins (Fig. 6B). In tAE1⌬N, 300 amino-acids in the putative cytoplasmic N-terminal domain (counting 400 amino-acids) were removed; the deletion of almost all of the cytoplasmic domain of the exchanger has an effect upon this protein expression at oocyte plasma membrane, as indicated by a 69% decrease of the Cl Ϫ exchange activity. However, despite a lower Cl Ϫ exchange activity, this deletion does not prevent stimulation of the NKCC co-transport, as illustrated in Fig. 7. Thus, the N-terminal cytoplasmic domain of tAE1 is not important for the stimulation of the co-transporter. In contrast, substitution of the last eight spans of tAE1 by their mouse AE1 counterparts, chimera TZM, abolished the Cl Ϫ -dependent Rb uptake. The reverse chimera made with mAE1 and the last eight spans of tAE1 are not expressed in oocyte. Therefore, we used skAE1, the skate red cell anion exchanger (26), to construct SZT. When the eight C-terminal helices of skAE1 are substituted by tAE1 corresponding domain, the SZT chimera thus obtained was able to transport Cl Ϫ as tAE1 or skAE1, 2 and this chimera was also able to stimulate the NKCC activity (Fig. 7). These data strongly suggest that the C-terminal domain of tAE1 is responsible for the activation of the NKCC co-transporter in Xenopus oocyte. To evaluate precisely the size of the C-terminal domain of tAE1 involved in NKCC co-transporter activation, two other chimeras between trout and mouse AE1 were used: T-5M and T-4M. In these different constructions, the 5 or 4 last putative ␣-helices, respectively, of tAE1 were substituted by their mouse counterparts. As shown previously, these constructs are functional in Xenopus oocyte and induce a Cl Ϫ influx similar to that of tAE1. 2 However, these two chimeras do not stimulate the NKCC co-transporter as tAE1 (Fig. 7). The anion conductance of the different chimeras was indicated under each construction in Fig. 7. The conductance was measured 2 days after oocyte injection. These data show a large discrepancy between chimeras regarding their anion conductance. Furthermore, these differences are not correlated with differences in Rb influx activity. All together, these data suggest the possible interaction between the last four putative helices in the C-terminal part of tAE1 and the NKCC co-transporter. To further investigate this possibility, we attempted to modify the C-terminal end of tAE1 by two means. First, we increased the size of tAE1 C-terminal domain by the fusion of a cytosolic protein, the B subunit of gyrase composed of about 200 amino acids. GyrB domain comes from the bacterial DNA gyrase. As illustrated in Fig. 8A, the fusion protein called tAE1-GyrB is expressed in Xenopus oocyte. Moreover, the addition of GyrB does not have any effect upon the anion conductance and Cl Ϫ transport characteristics (Fig. 8B). However, Fig. 8C shows that the presence of GyrB in the C-terminal part of tAE1 abolished 67% of NKCC co-transporter activation.
Second, we used an antibody raised against the 15 last amino-acids in the C-terminal end of tAE1 to prevent putative interactions between the C-terminal end of tAE1 and the NKCC co-transporter, thus, inhibiting the Cl Ϫ -dependent Rb influx stimulated by tAE1 expression. This antibody was shown to react with tAE1 in a Western blot (Figs. 6A and 8A). Injection of the antibody was done 1, 2, 3 or 5 hours before assessing the Cl Ϫ -dependent Rb influx in either tAE1-expressing oocytes or control oocytes. As control, oocytes expressing tAE1 were injected with water. It was observed that injection of the antibody at any time did not impair anion conductance or FIG. 5. Glybenclamide effect on tAE1 anion conductance and tAE1stimulated NKCC co-transporter. A, conductance (G S) of tAE1-expressing oocyte was measured in the presence or absence of 2 ϫ 10 Ϫ4 M glybenclamide. Measurements were done at day 6 postinjection when tAE1 conductance was high (13). Glybenclamide was prepared at 10 Ϫ1 M in Me 2 SO just before use. Results are expressed in % of the conductance of tAE1-expressing oocyte in the absence of inhibitor (control 100%). Means Ϯ S.E., n ϭ 3. Inset, dose-response curve of tAE1 anion conductance (G S) as a function of glybenclamide concentration (Log M). Oocytes expressing tAE1 for 3 days were preincubated for 15 min in glybenclamide-containing MBS, and conductance was measured in the presence of the corresponding dose of glybenclamide. Data are the means obtained on six oocytes Ϯ S.E. B, Rb influx in tAE1-expressing oocytes was measured 2 days after injection in MBS containing 10 Ϫ4 M ouabain and glybenclamide 2 ϫ 10 Ϫ4 M. Results are expressed as % of tAE1-induced Rb influx in the absence of inhibitors (control ϭ 100%). Rb influx in non-injected oocyte with or without inhibitors was subtracted to corresponding tAE1-induced Rb influx. Means Ϯ S.E., n ϭ 3.
Cl influx mediated by tAE1. In control condition, conductance and Cl influx of tAE1-expressing oocytes were 18.3 Ϯ 2.6 S and 192 Ϯ 14 pmol/min per oocyte (n ϭ 3), and they were 19.8 Ϯ 4.1 S and 202 Ϯ 15 pmol/min per oocyte, n ϭ 3, in these oocytes previously injected with 50 nl of Ab145). Moreover, this antibody was tested regarding its possible inhibitory effect on the endogenous NKCC co-transporter. The hyperosmotically stimulated Rb influx was measured in oocytes previously injected with 50 nl of water or antibody 145. The NKCC cotransporter activity was equivalent in both conditions (387 Ϯ 85 pmol/h per ovocyte in water-injected oocytes versus 355 Ϯ 65 pmol/h per ovocyte in Ab145-injected oocytes, n ϭ 5). In contrast, Fig. 9 shows that injection of antibody 145 inhibits the Cl-dependent Rb influx induced by tAE1. The maximal inhibition was obtained 2 h after antibody injection. The presence of the antibody abolished 41 Ϯ 9% of the tAE1-induced stimulation of NKCC (p Ͻ 0.05). It should be noticed that injection of the same amount of an irrelevant antibody (IgG against rabbit actin) did not impair the Cl-dependent Rb influx in tAE1expressing oocytes (Fig. 9). DISCUSSION In the present report, we show that the Rb influx stimulated by tAE1 expression in Xenopus oocyte is sensitive to micromolar concentrations of bumetanide and to Na ϩ or Cl Ϫ substitution. Moreover, it is not additive with the Rb influx activated by hyperosmotic medium, which strongly suggests that the same transporter is the target of the hyperosmotic shock and tAE1. This transporter has the features of the NKCC co-transporter that is present in oocyte plasma membrane, where it has been studied by different groups (21,22). The stimulation of this co-transporter is observed as soon as tAE1 starts to be expressed in oocyte plasma membrane (1 day after injection; Ref. 13), and the activation is immediately maximal. The simplest interpretation of these results is that tAE1 expression in oocyte membrane activates the endogenous NKCC co-transporter.
Different kinds of interactions between membrane transporters could be envisioned; for example, they could be functional coupling or direct protein-protein interactions. Because tAE1 modifies oocyte membrane permeability, we first examined how these changes may be involved in NKCC co-transporter activation.
As previously shown, tAE1 is able to form an anion channel in oocyte plasma membrane. Up-regulation of NKCC co-transporter by Cl Ϫ channels has been reported in different papers. For instance, in epithelial cells, opening of Cl Ϫ channels in the apical membrane stimulates NKCC uptake in the basolateral membrane (27,28). This stimulation seems to be related to intracellular Cl Ϫ depletion by a molecular mechanism still not well understood. Co-expression of NKCC co-transporter and FIG. 6. Schema of different anion exchangers, chimerical constructions and immunodetection of these proteins. A, tAE1, mAE1, skAE1, and chimerical constructions between cytoplasmic and transmembrane domains of the exchangers. The extracellular loop connecting spans 5 and 6, which is long in tAE1 and much shorter in all other AE1, is called the Z-loop in tAE1. The topology of the exchangers is the one described by Fujinaga et al. (18), with 13 ␣ helices crossing the lipid bilayer, the N terminus and C terminus part of the protein being in the cytoplasm. In TZM, the eight helices after the Z-loop were substituted by the corresponding helices in mAE1. In T-5M and T-4M, the five (T-5M) or four (T-4M) last helices of tAE1 were substituted by the five or four last helices of mAE1. In SZT, the N-terminal domain and the five first helices of tAE1 were substituted by the corresponding part of skAE1. tAE1⌬N is a tAE1 deleted of the first 311 amino-acids (corresponding to 3/4 of the cytoplasmic domain of the protein). the KCl co-transporter in human embryonic kidney cells showed that KCl depletion induced by KCl co-transporter activates NKCC co-transporter (2). Moreover, in immature neuronal cells, activation of GABA A receptors decreases intracellular Cl Ϫ concentration, and this consequently stimulates NKCC co-transporter. tAE1-expressing oocytes exhibit a high Cl Ϫ conductance that allows the cell to reach equilibrium potential for Cl Ϫ (13). Our data clearly show that the depolarization in tAE1-expressing oocytes is not associated with a decrease in intracellular Cl Ϫ concentration at day 1 post-injection (Table I). Indeed, in control oocytes, equilibrium potential for Cl Ϫ , calculated from data Table I, is Ϫ35.6 mV. This value properly fits to membrane potential measured in tAE1-expressing oocytes at day 1 postinjection (Em ϭ Ϫ36.2 Ϯ 1.9 mV). Thus, tAE1 expressing oocytes reach the equilibrium potential for Cl Ϫ without any change in Cl Ϫ concentration. To conclude, the absence of intracellular Cl Ϫ depletion ruled out this possibility in the observed stimulation of the NKCC co-transporter.
Furthermore, tAE1 anion conductance also does not seem to be the triggering factor of NKCC co-transporter. Indeed, the conductive properties of tAE1 could be inhibited by glybenclamide (Fig. 5A) without modifying activation of the co-transporter (Fig. 5B). This result suggests the absence of functional links between tAE1 and the co-transporter.
In a previous work, different chimeras between tAE1 and other AE1s unable to form a Cl Ϫ channel were used to determine the domain of the protein involved in the channel formation. 2 These chimeras strengthened the absence of correlation between conductance of tAE1 and co-transporter activation. As illustrated in Fig. 7, all of the chimeras ending with mAE1 sequences were unable to stimulate the NKCC co-transporter despite the fact that they have an anion conductance (T-4M and T-5M) or no anion conductance (TZM). In contrast, chimera SZT, which has a similar conductance to T-4M or T-5M, is able to maximally stimulate the co-transporter. Activation of NKCC co-transporter is maximal for different values of conductance: G ϭ 15.2 S (tAE1), 6.1 S (SZT), or 2.8 S (tAE1⌬N); for equivalent conductances, activation is either maximal (SZT with G ϭ 6.1 Ϯ 1.0 S) or null (T-5M with G ϭ 4.53 Ϯ 1.07 S). This definitely rules out a role for the tAE1 chloride channel in NKCC co-transporter activation.
It seems obvious that activation of the co-transporter in tAE1-expressing oocytes is not related to known thermodynamic events because of tAE1 functioning but rather because of the presence of this protein in the plasma membrane. This stimulation is dependent upon at least the last four putative spanning domains in the C-terminal part of tAE1, because their substitution by corresponding helices of mAE1 abolished stimulation of the co-transporter. It is possible to interfere in the activation by increasing the size of the cytoplasmic tail of FIG. 8. Fusion protein tAE1-gyrase B. A, Western blot of tAE1 or tAE1GyrB fusion protein expressed in Xenopus oocyte membrane. Membranes were prepared 2 days after injection, as described under "Experimental Procedures." The presence of tAE1 was detected by Ab145 (against C-terminal end, lane b) or Ab146 (against N-terminal end, lane d). tAE1GyrB was detected by Ab146. The lower bands around 100 kDa correspond to unglycosylated or partially glycosylated tAE1, and the upper bands correspond to glycosylated tAE1 (controlled by deglycosylation experiments, not shown). The fusion tAE1-GyrB, unglycosylated and glycosylated, migrated with an apparent molecular mass of 120 kDa for the unglycosylated form and 140 kDa for the glycosylated form (lane e) instead of 100 kDa and 120 kDa for the wild tAE1. The B gyrase has a molecular mass of 24 kDa. Mass markers in kDa are drawn on the left of the picture. Lane a corresponds to control oocytes (non-injected) reacting with Ab145, and lane c corresponds to control oocytes reacting with Ab146. This Western blot is representative of at least five different experiments. B, Cl Ϫ influx was measured in oocyte-expressing tAE1 or the fusion protein tAE1-gyraseB, 2 days after injection of cRNA. Uptake was done in 15 min (during the linear phase of the uptake) in MBS with a specific radioactivity of 363 dpm/nmol of Cl Ϫ . Control is non-injected oocytes. Data presented are from one representative experiment of three. Means Ϯ S.E., n ϭ 8 oocytes. C, Rb influx was measured in MBS containing 10 Ϫ4 M ouabain with tAE1 or tAE1-gyraseB-expressing oocytes 2 days after injection. Data presented are from one representative experiment of three different experiments. Means Ϯ S.E., n ϭ 8 oocytes.
FIG. 9. Inhibition of tAE1-stimulated NKCC co-transporter by a specific antibody against tAE1 C-terminal end. Oocytes expressing tAE1 for 1 day and control oocytes were injected with 50 nl of antibody Ab145, diluted 50ϫ in DEPC-treated water 2 h before measuring Cl-dependent Rb influx. Control is non-injected oocytes. To check the effect of injection on this Rb influx, reference tAE1-expressing oocytes were injected with 50 nl of DEPC-treated water or 50 nl of an irrelevant antibody (against rabbit actin, Sigma, diluted to inject the same amount of IgG as for Ab145) 2 h before measuring the Rb influx. Means Ϯ S.E., n ϭ 7 experiments (except with Ab against actin, n ϭ 3; p Ͻ 0.05, comparison of tAE1 ϩ water and tAE1 ϩ Ab145). tAE1 (fusion tAE1gyraseB). The same result is observed when a specific antibody to the C-terminal extremity is injected in oocytes. These data strongly suggest that tAE1 is able to interact with the co-transporter. Further experiments should determine whether this is a direct interaction between the two transporters or if it involves an intermediary regulatory protein. It is well known that anion exchangers are able to interact by their large N-terminal cytoplasmic domain with different proteins in erythrocytes (29,30). The only described interaction site localized in the short cytoplasmic C-terminal domain of the protein is a binding site for carbonic anhydrase (31). However, our results provide further evidence for the role of the Cterminal domain of tAE1 in interactions with other proteins.
The ability of tAE1 to stimulate a co-transporter in Xenopus oocyte could be extrapolated to the erythrocyte where tAE1 is the most abundant plasma membrane protein. It was shown that this anion exchanger has a key role in trout red cell regulatory volume decrease response, acting as a channel permeable to taurine, Na ϩ , and K ϩ (32). Hyposmotic swelling of trout erythrocyte induces a K ϩ loss mediated by two pathways: a K-Cl co-transporter and a Cl-independent pathway. The stoichiometry of solute lost during regulatory volume decrease is seven taurine with three cations (K ϩ and Na ϩ ) and three Cl Ϫ (32). Such finely tuned osmolyte transport requires precise coordination between the two transporters KCl and tAE1. It is tempting to assume that, by interacting with the K-Cl cotransporter in erythrocyte, tAE1 is able to regulate its activity. Moreover, it is well known that AE1 is able to interact with hemoglobin by means of its binding to erythrocyte cytoskeleton. This binding is sensitive to hemoglobin oxygenation (33). The data associated with our results, which suggest a role of tAE1 in regulating other transport functioning, might provide an explanation of oxygen control of the KCl cotransporter as well as the Na ϩ /H ϩ exchanger in trout erythrocyte (34,35). Indeed, the oxygen sensitivity of these two transporters could be mediated by tAE1 interacting with hemoglobin on one side and with these transporters on the other.
The CFTR, a Cl Ϫ channel, was the first transporter studied for the possibility it might regulate other transporters (6,9,36,37,38). Our data provide an example other than the CFTR that supports the existence of a mechanism of membrane transport regulation by molecular interactions (direct or indirect) between transporters themselves. We have no explanation about the molecular mechanism underlying activation of NKCC cotransporter by tAE1, but regulation through interactions with a common regulatory protein could be hypothesized. Indeed, there is increasing evidence that membrane transporters may coordinate their functions through their common interactions with PDZ domains of regulatory proteins such as, for instance, EBP50 (39) or NHE3 kinase A regulatory protein (40).