Solute Carrier Family 26 Member a2 (Slc26a2) Protein Functions as an Electroneutral SO42−/OH−/Cl− Exchanger Regulated by Extracellular Cl−*

Background: Slc26a2 is an SO42− transporter, mutations in which cause diastrophic dysplasia. How Slc26a2 transports SO42− is unknown. Results: We found that Slc26a2 exchanges SO42− for 2OH− or 2Cl− and is regulated by a promiscuous extracellular anion site. Conclusion: Slc26a2 functions as SO42−/2OH− or SO42−/2Cl− exchanger, regulated by extracellular Cl−o. Significance: The findings should help in understanding aberrant SLC26A2 function in diastrophic dysplasia. Slc26a2 is a ubiquitously expressed SO42− transporter with high expression levels in cartilage and several epithelia. Mutations in SLC26A2 are associated with diastrophic dysplasia. The mechanism by which Slc26a2 transports SO42− and the ion gradients that mediate SO42− uptake are poorly understood. We report here that Slc26a2 functions as an SO42−/2OH−, SO42−/2Cl−, and SO42−/OH−/Cl− exchanger, depending on the Cl− and OH− gradients. At inward Cl− and outward pH gradients (high Cl−o and low pHo) Slc26a2 functions primarily as an SO42−o/2OH−i exchanger. At low Cl−o and high pHo Slc26a2 functions increasingly as an SO42−o/2Cl−i exchanger. The reverse is observed for SO42−i/2OH−o and SO42−i/2Cl−o exchange. Slc26a2 also exchanges Cl− for I−, Br−, and NO3− and Cl−o competes with SO42− on the transport site. Interestingly, Slc26a2 is regulated by an extracellular anion site, required to activate SO42−i/2OH−o exchange. Slc26a2 can transport oxalate in exchange for OH− and/or Cl− with properties similar to SO42− transport. Modeling of the Slc26a2 transmembrane domain (TMD) structure identified a conserved extracellular sequence 367GFXXP371 between TMD7 and TMD8 close to the conserved Glu417 in the permeation pathway. Mutation of Glu417 eliminated transport by Slc26a2, whereas mutation of Phe368 increased the affinity for SO42−o 8-fold while reducing the affinity for Cl−o 2 fold, but without affecting regulation by Cl−o. These findings clarify the mechanism of net SO42− transport and describe a novel regulation of Slc26a2 by an extracellular anion binding site and should help in further understanding aberrant SLC26A2 function in diastrophic dysplasia.


Slc26a2 is a ubiquitously expressed SO 4
2؊ transporter with high expression levels in cartilage and several epithelia. Mutations in SLC26A2 are associated with diastrophic dysplasia. The mechanism by which Slc26a2 transports SO 4 2؊ and the ion gradients that mediate SO 4 2؊ uptake are poorly understood. We report here that Slc26a2 functions as an SO   Protein sulfation, and thus SO 4 2Ϫ , is essential for cellular and tissue survival. Many proteins undergo post-translational modification by sulfation. Tyrosine sulfation of signaling molecules, like the G protein-coupled receptor chemokine receptors (1), modifies signaling pathways. Protein sulfation contributes to detoxification of endogenous compounds (2). A critical role of protein sulfation is sulfation of proteoglycans (3). Proteoglycans are constituents of the extracellular matrix that mediate the cell response to growth factors (4). Several disorders are caused by mutations in genes that affect proteoglycan synthesis or sulfation. The sulfate groups in proteoglycans are critical in formation of active domains, and the high polyanionic charge density of the proteoglycans is neutralized by SO 4 2Ϫ (5). Sulfation of secretory proteins, like digestive enzymes and mucins, is essential for their synthesis, processing through the biosynthetic pathway and packaging in secretory granules (6). Hence, understanding SO 4 2Ϫ homeostasis is essential for understanding cell development and function.
Although Slc26a1 has limited tissue distribution, Slc26a2 is ubiquitously expressed with particularly high levels in developing and mature cartilage as well as in epithelial tissues like pancreas, salivary glands, colon, bronchial glands, tracheal epithelium, and eccrine sweat glands (20,21). The central role of Slc26a2 in supplying the bulk of cellular SO 4 2Ϫ is evident from the lethality of deletion of the SLC26A2 gene in humans and mice (20,22), mainly due to under-sulfation of proteoglycans leading to aberrant development (23). Indeed, measurement of SO 4 2Ϫ uptake in fibroblast from patients with a severe form of the disease showed reduced or lack of SO 4 2Ϫ uptake (20,24). Most mutations causing diastrophic dysplasia are missense mutations that affect either trafficking to the plasma membrane or showed reduced SO 4 2Ϫ transport (25,26). The phenotype of chondrodysplasias is highly variable, ranging from mild (27) to lethal before or shortly after birth (11). To better understand the disease and cellular SO 4 2Ϫ homeostasis, it is necessary to understand transport and regulation of Slc26a2. To date, characterization of transport by Slc26a2 was based on measurement of isotopic fluxes (24,25,28) that are the sum of both net and exchange fluxes, with the exchange dominating the fluxes. These studies revealed that Slc26a2 can transport SO 4 2Ϫ , Cl Ϫ , and oxalate (24,25,28), and a recent detailed characterization of the fluxes suggested that Slc26a2 functions as an electroneutral transporter when mediating isotopic fluxes. SO 4 2Ϫ fluxes appeared to be sensitive to intracellular and extracellular pH (24). An unusual finding was that inhibition of SO 4 2Ϫ and oxalate isotopic uptake by external Cl Ϫ exhibited simple saturation, whereas Slc26a2-mediated exchange of intracellular SO 4 2Ϫ , oxalate, or Cl Ϫ for external Cl Ϫ was non-saturable (24), suggesting that the measured fluxes, at least isotopic efflux, is mostly exchange rather than net fluxes.
The available information is not sufficient to determine the mode of SO 4 2Ϫ and other ions transport by Slc26a2 and the cellular ionic gradients that drive net transport. We used Xenopus oocytes expressing Slc26a2 to report that Slc26a2 functions as SO  Ϫ . Slc26a2 activity is regulated by an extracellular anion binding site, which is not involved in ion transport. Modeling of the Slc26a2 transmembrane sector identified an extracellular loop, which contains the conserved sequence 367 GFXXP 371 in the vicinity of the gating Glu 417 as a potential part of the permeation pathway. These findings should help in further understanding ion transport by the SLC26 transporters and aberrant SLC26A2 function in diastrophic dysplasia.

EXPERIMENTAL PROCEDURES
Solutions and Reagents-For experiments in oocytes, the standard HEPES-buffered ND96 solution contained (in mM): 96 NaCl, 2 KCl, 1.8 CaCl 2 , 1 MgCl 2 , and 5 HEPES, pH 7.5. Cl Ϫ -free solutions were prepared by replacing chloride with gluconate in the presence of calcium cyclamate substituted for CaCl 2 . A 100 mM solution of diisothiocyanostilbene-2,2Ј-disulfonic acid (DIDS) (Invitrogen) dissolved in DMSO was prepared freshly and diluted to a final concentrations of 10 or 50 M in the relevant solutions. All other chemicals and reagents were purchased from Sigma.
cRNA Preparation-The pCMV-Sport6-Slc26a2 (Gen-Bank TM /EMBL/DDBJ, accession no. BC028345) was purchased from Open Biosystems and was used as template for cRNA preparation. The plasmid was linearized with NotI and used to transcribe cRNA with an mMESSAGE mMACHINE Sp6 kit (Life Technologies, Applied Biosystems), respectively. Mutation in Slc26a2 were generated by a site-directed mutagenesis kit (Agilent Technologies) and verified by sequencing.
Biotinylation and Western Blot Analysis-To monitor surface expression of Slc26a2 WT, E417A, and E417K, HEK cells transfected with vector alone or Myc-tagged Slc26a2 constructs were incubated with EZ link Sulfo-NHS-LC-Biotin (0.5 mg/ml, Thermo Fisher Scientific) for 30 min at room temperature. Subsequent steps were as previously described (29) with the following modifications: 50 l of 1:1 slurry of immobilized avidin beads (Thermo Fisher Scientific) was added to 300 g of protein in 300 l of cell extract, and the mixture was incubated overnight. To monitor protein expression the PVDF membranes were incubated overnight with anti-Myc antibodies diluted 1:1,000 (Cell Signaling) and for 1 h with HRP-conjugated goat anti-mouse (Invitrogen) diluted 1:2,000. For ␤-actin detection membranes were incubated for 1 h with monoclonal anti-␤-actin peroxidase (Sigma-Aldrich) diluted 1:20,000.
Xenopus laevis Oocyte Preparation-All experiments in this study were conducted under the National Institutes of Health guidelines for research on animals, and experimental protocols were approved by the Institutional Animal Care and Use Committee. Oocytes were isolated by partial ovariectomy of anesthetized female X. laevis (Xenopus Express, Brooksville, FL) and treated by collagenase B (Roche Applied Science), as described previously (30). Stage V-VI oocytes were injected with 10 ng of cRNA using glass micropipettes and a microinjection device (Nanoliter 2000; World Precision Instruments) in a final volume of 27.6 nl. Control oocytes were injected with equal volumes of H 2 O. Oocytes were incubated at 18°C in ND96 supplemented with 2.5 mM pyruvate and antibiotics and were studied 72-144 h after injection.
Voltage, pH, and Cl Ϫ Measurement in Oocytes-Voltage recordings were performed at room temperature with two-electrode voltage clamp, exactly as described previously (29,30). Voltage, pH i , and Cl Ϫ i concentrations were measured as detailed previously (31,32). In the present study, the Cl Ϫ -sensitive electrode was also used to record intracellular Br Ϫ , I Ϫ , and NO 3 Ϫ with the resin and the procedure used to measure Cl Ϫ (see "Results").
Measurement of Buffer Capacity-To determine OH Ϫ (H ϩ ) fluxes by Slc26a2 we determined the buffer capacity of oocytes bathed in HEPES-buffered medium. Because we can measure both Cl Ϫ i and pH i , we determined the buffer capacity directly rather than relying on pH i changes induced by weak acids.
Supplemental Fig. 1A shows that two consecutive injections of the oocytes with 13.8 nl of 100 mM HCl reduced pH i and increased Cl Ϫ i . Similar determination in five experiments and using the pH i and Cl Ϫ i changes of the first injection resulted in a buffer capacity of 17.1 Ϯ 2.2/pH unit, which is similar to that reported by others (33).
Modeling and Prediction of the Slc26a2 Transmembrane Domains Structure-The transmembrane sector of the mouse Slc26a2 model was generated using the Deepview Swiss-PDB viewer by Raw sequence fit of the Slc26a2 sequence (NCBI accession no. NP_031911) onto the putative Slc26a6 model previously generated by us based on structural similarity to the bacterial ClC-ec protein (29). The predicted binding site of DIDS on the Slc26a2 model was performed with the AutoD-ockVina software (34), according to software tutorial instructions. Briefly, the box grid determining the Slc26a2 region of binding was set using the AutoDockTools software with the following coordinates (center: x ϭ 0.472, y ϭ 1.222, z ϭ 0.472) (size: x ϭ 30, y ϭ 26, z ϭ 24). Exhaustiveness level was set to default. AutoDockTools was further used to select all rotatable bonds of the DIDS molecule. The AutoDockVina software generated nine different models, and herein we present the best model as ranked by the software with a predicted affinity of Ϫ8.9 kcal/mol for the binding of Slc26a2 and DIDS. The final model (cartoon and surface representations) was generated using PyMOL (Schrödinger, LLC). /OH Ϫ /Cl Ϫ Exchanger-Slc26a2-mediated net fluxes were assayed in Xenopus oocytes by measuring intracellular pH (pH i ) and Cl Ϫ (Cl Ϫ i ) and the membrane potential in the same oocytes. Fig. 1A shows that removal of extracellular Cl Ϫ (Cl Ϫ o ) had no effect on pH i and the membrane potential with a slow rate of reduction in Cl Ϫ i . Exposing Slc26a2-expressing oocytes bathed in Cl Ϫ -free solution to 0.2 mM SO 4 2Ϫ resulted in a precipitous reduction in pH i and Cl Ϫ i . Removal of SO 4

RESULTS AND DISCUSSION
2Ϫ o with the concomitant addition of Cl Ϫ o resulted in increased pH i and Cl Ϫ i . Fig. 1C shows almost no change in Cl Ϫ i and pH i in waterinjected oocytes under the same conditions. Reduction in pH i can be due to H ϩ influx or OH Ϫ efflux. Because some of the SO 4 2Ϫ transport is coupled to Cl Ϫ , and the Cl Ϫ coupling is affected by pH o (see below), we will refer to the transported ion as OH Ϫ , although we cannot distinguish between the transport of OH Ϫ and H ϩ . The average Slc26a2mediated SO 4 2Ϫ -coupled net Cl Ϫ and OH Ϫ transports are shown in Fig. 1B   obtained from these experiments were used to calculate the apparent K m for SO 4 2Ϫ that were then plotted as a function of Cl Ϫ o (Fig. 2C). The linear relationship in Fig. 2C (24). This may reflect the different dependence of Cl Ϫ o of the half (exchange) and full turnover cycle (net) of transport by Slc26a2.
Many of the SLC26 transporters can transport HCO 3 Ϫ in exchange for Cl Ϫ (9). However, Fig. 2D shows that Slc26a2 does not function as a Cl Ϫ /HCO 3 Ϫ exchanger. The capacity of Slc26a2 to transport other anions, such as I Ϫ , Br Ϫ and NO 3 Ϫ , in addition to SO 4 2Ϫ , OH Ϫ and Cl Ϫ was further tested by measuring their intracellular concentration. Supplemental Fig. 1B shows that the resin used to detect Cl Ϫ can also detect Br Ϫ and NO 3 Ϫ ϳ10 times better than Cl Ϫ and I Ϫ ϳ100 times better than Cl Ϫ (see also (29)). The left panel of Fig. 2D shows that exposing Slc26a2-expressing oocytes to Cl Ϫ -free solution containing 2 mM I Ϫ resulted in a rapid influx of I Ϫ . Removal of I Ϫ in the absence of Cl Ϫ o stopped the influx. To initiate I Ϫ efflux it was necessary to add Cl Ϫ o , with as little as 1 mM Cl Ϫ o resulting is nearly maximal rate of I Ϫ efflux. Similar behavior was observed with Br Ϫ and NO 3 Ϫ (Fig. 2D, right panel) and no I Ϫ (Fig. 2D, FEBRUARY  2Ϫ and Ox 2Ϫ transport (see below), suggesting that the transport rate follows the pH gradient rather than substrate species.

Slc26a2 Transport Properties and Regulation
Coupling of SO 4 2Ϫ transport to both Cl Ϫ and OH Ϫ may function to ensure SO 4 2Ϫ uptake under acidic and alkaline conditions. Slc26a2 is expressed in the luminal membrane of polarized cells (31,35) that can be exposed to acidic and alkaline pH. In the stomach and synovial fluid pH is acidic (36,37) (Fig. 4B). Activation of the exchange was not specific for Cl Ϫ . Fig. 4C shows that 1 mM external Cl Ϫ , Br Ϫ , I Ϫ , NO 3 Ϫ and SCN Ϫ similarly activated SO 4 2Ϫ i /2OH Ϫ o exchange. Only 1 mM F Ϫ did not activate the exchange (Fig. 4C), but actually inhibited the exchange initiated by the other anions (not shown).
The findings in Fig. 4 suggest that SO 4 2Ϫ transport by Slc26a2 is regulated by interaction of an anion with a regulatory site. The regulatory site is not selective for Cl Ϫ , but because Cl Ϫ is the major extracellular anion, Slc26a2 is likely regulated by may be by stabilization of an active Slc26a2 conformation. However, the exact mechanism remains to be elucidated. The physiological significance of regulation of Slc26a2 activity by Cl Ϫ o is not known at present. The Cl Ϫ content in the GI tract is high in the range of 100 -150 mM and is determined largely by acid secretion (40). On the other hand, urine Cl Ϫ can be below 4 mM when prerenal azotemia occurres with metabolic alkalosis (41) and regulation of Slc26a2 by Cl Ϫ o can become significant. In addition, the luminal membrane-localized Slc26a2 is exposed to variable Cl Ϫ concentrations, as low Cl Ϫ in ducts that absorb the Cl Ϫ , such as the pancreatic (38) and salivary (32) ducts, the  A Potential Slc26a2 Permeation Pathway-In a previous study we developed a model of the Slc26a6 transmembrane sector to search for motifs that determine the function of the electrogenic Slc26 transporters as coupled and uncoupled transporters (29). The modeling identified a glutamate (Glu Ϫ ) conserved in all Slc26 transporters that has the same orientation as Glu Ϫ E148 in the Cl Ϫ permeation pathway of the ClC transporters (44 -47). Interestingly, a recent study utilized the predicted Slc26a6 model and the crystal structure of the Slc26a5 STAS domain to assemble a detailed putative structure of Slc26 transporters (48). This structure showed a surprising similarity to the low resolution structure of a bacterial Slc26 homologue obtained using SANS (small angle neutron scattering) method in terms of symmetry and size. Notably, this study suggested that Slc26 functions as a dimer, as was previously suggested based on the predicted similarity of Slc26a6 to the bacterial ClC-ec dimeric crystal structure. Therefore, assuming a similar overall architecture for Slc26 transporters, we thread the Slc26a2 transmembrane sector on the Slc26a6 model to determine the localization of the conserved Glu Ϫ Glu 417 (Fig.   5). Another purpose of the modeling was to identify additional determinants of the Slc26a2 ion permeation pathway and perhaps the extracellular Cl Ϫ regulatory site. The motif GSGIP was identified as a potential anion (Cl Ϫ ) binding site that is conserved in the ClC transporters (44,49). Mutations of residues within this motif altered ionic selectivity and coupling in the yeast and mammalian ClCs (50,51). We searched for a similar motif in the Slc26 transporters. Although identical motif is not present in the Slc26 transporters, supplemental Fig. 2 shows the presence of the well conserved sequence GFXXP. The structural model in Fig. 5 shows the predicted localization of the Slc26a2 conserved Glu Ϫ Glu 417 and phenylalanine Phe 368 and of a potential DIDS binding site.
To test the prediction in Fig. 5 we first determined the sensitivity of Slc26a2 to DIDS. Fig. 6A shows that 50 M DIDS completely inhibited SO 4 2Ϫ -driven OH Ϫ (H ϩ ) efflux and most of the SO 4 2Ϫ -driven Cl Ϫ efflux. The residual DIDS-insensitive Cl Ϫ efflux is likely not mediated by Slc26a2 but by a DIDS-insensitive transporter native to the oocytes. Fig. 6B shows that DIDS inhibited SO 4 2Ϫ efflux when added after SO 4 2Ϫ uptake. Also in this case DIDS completely inhibited OH Ϫ (H ϩ ) influx, but with a residual Cl Ϫ influx. Similar results were obtained with 10 and 50 M DIDS, indicating that the DIDS sensitivity of Slc26a2 is in the same range of that reported for Slc26a6 (52). Fig. 6C summarizes the rates of OH Ϫ and Cl Ϫ fluxes in the absence and presence of SO 4 2Ϫ and DIDS, indicating that at pH o of 7.5 and the absence of Cl Ϫ o ϳ60% of SO 4 2Ϫ uptake is coupled to OH Ϫ efflux and 40% to Cl Ϫ efflux. Fig. 6D test another prediction of the model in Fig. 5 by neutralizing (Slc26a2(E417A)) or reversing (Slc26a2(E417K)) the charge of the conserved Glu 417 . Both mutations eliminated SO 4 2Ϫ (Fig. 6D) and I Ϫ (not shown) transport activity. Inhibition of transport was not due to altered trafficking of the mutants to the plasma membrane (Fig. 6E).
The sequence GFXXP is predicted to be in the extracellular loop between transmembrane domains (TMDs) 7 and 8, with Phe 368 predicted to be in the entrance of the permeation pathway (Fig. 5). The mutations G367A and P371A had no effect on SO 4 2Ϫ transport or its coupling to Cl Ϫ and OH Ϫ (not shown). However the F368A mutation had multiple effects. Fig. 7A shows that Slc26a2(F368A) is ϳ50% less active than wild-type Slc26a2 in exchanging SO 4 2Ϫ for OH Ϫ (left traces) and Cl Ϫ (right traces). Most notably, the F368A mutation increased the apparent affinity of Slc26a2 for SO 4 2Ϫ by ϳ8-fold to reduce the apparent K m for SO 4 2Ϫ o from 79 Ϯ 7 to 9.7 Ϯ 0.7 M. Unexpectedly from competition between SO 4 2Ϫ o and Cl Ϫ o (Fig. 2), the F368A mutation increased the apparent K m for inhibition of SO 4 2Ϫ uptake by Cl Ϫ o from 26 to 50 mM (Fig. 7C). Hence, Phe 368 appears to control the access of SO 4 2Ϫ and Cl Ϫ to the permeation pathway. Interestingly, Fig. 7D shows that the F368A mutation had no effect of the apparent affinity for the Cl The findings in Fig. 7, A-C provide additional evidence for the importance of the GSGIP or the GFXXP motifs in the function of the Cl Ϫ transporters, in addition to the two additional  FEBRUARY 10, 2012 • VOLUME 287 • NUMBER 7 GXXXP motifs that participate in Cl Ϫ transport in the bacterial ClCs (44). The bacterial ClC-ec1 crystal structure shows that the permeation pathway has three Cl Ϫ interacting sites (44 -46, 49). Ser 107 and Gly 108 in the GSGIP motif coordinate the Cl Ϫ ion in the internal substrate site, and the side chain of Ser 107 participates in binding of the middle Cl Ϫ (45,49 Fig. 8B shows that     FEBRUARY 10, 2012 • VOLUME 287 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 5129 prominent as the increased apparent affinity for SO 4 2Ϫ (Fig. 7), it was in the same direction. The results in Fig. 8, A-D indicate that the properties of Ox 2Ϫ transport closely resemble those of SO 4 2Ϫ transport, although at the same conditions the Ox 2Ϫ transport rate was ϳ50% slower than the SO 4 2Ϫ transport rate. In summary, the present study reports the mechanism of SO 4 2Ϫ and Ox 2Ϫ transport by Slc26a2. Both anions are transported in exchange for Cl Ϫ and OH Ϫ or by cotransport with H ϩ . Based on the rate of the coupled OH Ϫ (H ϩ ) fluxes in the absence of Cl Ϫ o and at substrate concentration of 1 mM, net SO 4 2Ϫ transport by Slc26a2 is about twice faster than net Ox 2Ϫ transport. Under normal conditions plasma oxalate is in the micromolar range and even in patients with primary hyperoxaluria plasma oxalate is around 40 M (53). Moreover, although Slc26a2 is expressed at high level in the luminal membrane of colonic crypts (21), SO 4 2Ϫ in the colon can be in the millimolar range both in human (54) and animals (55) that will favor SO 4 2Ϫ uptake by Slc26a2. Indeed, the colon is a major site of SO 4 2Ϫ absorption (54,56) that is likely mediated by Slc26a2. Similarly, although Slc26a2 is expressed in the proximal tubule luminal membrane (31), SO 4 2Ϫ concentration in the proximal tubule is in the millimolar range, and although the role of Slc26a2 in the kidney is not know, if any it is likely to function mainly as an SO 4 2Ϫ transporter (8). The only possible scenario where Slc26a2 can affect Ox 2Ϫ homeostasis is by mediating Ox 2Ϫ secretion in exchange for external SO 4 2Ϫ when external Cl Ϫ is low and pH is high. Even then, this process will be inhibited by the high cytoplasmic Cl Ϫ typical of epithelia and by intracellular SO 4 2Ϫ . Thus Slc26a2 is not likely to play a major role in oxalate metabolism in the colon or the kidney.

Slc26a2 Transport Properties and Regulation
The permeation pathway includes the conserved SLC26 transporter Glu Ϫ and may lay between TMD7 and TMD8, where a phenylalanine conserved in the loop predicted to connect the TMDs may control SO 4 2Ϫ and Cl Ϫ access to the permeation pathway. As yet, mutations of these residues, or even in the vicinity of these residues, have not been found in patients with diastrophic dysplasia (57). This is most likely because Slc26a2 is an essential gene and the mutations that markedly affect Slc26a2 activity may not be compatible with life. Indeed, analysis of several disease causing Slc26a2 mutations showed retention of some SO 4 2Ϫ transport capacity by the mutants and a good correlation between loss of SO 4 2Ϫ transport and disease severity (25,26). The coupling of SO 4 2Ϫ transport to both OH Ϫ and Cl Ϫ likely serves to ensure transport at both acidic pH when most SO 4 2Ϫ uptake is mediated by SO 4 2Ϫ /2OH Ϫ exchange and alkaline pH when most SO 4 2Ϫ uptake is mediated by SO 4 2Ϫ /2Cl Ϫ exchange. Slc26a2 is also regulated by an extracellular anion binding site different from the transport site, the physiological function of which remains to be determined, although it may control SO 4 2Ϫ uptake when Cl Ϫ o is very low.