Functional Comparison of the K+-Cl−Cotransporters KCC1 and KCC4*

The K+-Cl−cotransporters (KCCs) are members of the cation-chloride cotransporter gene family and fall into two phylogenetic subgroups: KCC2 paired with KCC4 and KCC1 paired with KCC3. We report a functional comparison inXenopus oocytes of KCC1 and KCC4, widely expressed representatives of these two subgroups. KCC1 and KCC4 exhibit differential sensitivity to transport inhibitors, such that KCC4 is much less sensitive to bumetanide and furosemide. The efficacy of these anion inhibitors is critically dependent on the concentration of extracellular K+, with much higher inhibition in 50 mm K+ versus 2 mmK+. KCC4 is also uniquely sensitive to 10 mmbarium and to 2 mm trichlormethiazide. Kinetic characterization reveals divergent affinities for K+(K m values of ∼25.5 and 17.5 mm for KCC1 and KCC4, respectively), probably due to variation within the second transmembrane segment. Although the two isoforms have equivalent affinities for Cl−, they differ in the anion selectivity of K+ transport (Cl− > SCN− = Br− > PO4 −3 > I−for KCC1 and Cl− > Br− > PO4 −3 = I− > SCN−for KCC4). Both KCCs express minimal K+-Cl−cotransport under isotonic conditions, with significant activation by cell swelling under hypotonic conditions. The cysteine-alkylating agentN-ethylmaleimide activates K+-Cl− cotransport in isotonic conditions but abrogates hypotonic activation, an unexpected dissociation of N-ethylmaleimide sensitivity and volume sensitivity. Although KCC4 is consistently more volume-sensitive, the hypotonic activation of both isoforms is critically dependent on protein phosphatase 1. Overall, the functional comparison of these cloned K+-Cl− cotransporters reveals important functional, pharmacological, and kinetic differences with both physiological and mechanistic implications.

In a great majority of cells, the plasma membrane is permeable to water. Movement of water across the cell membrane is largely dependent on the osmotic pressure gradient between the intracellular and extracellular space, such that water transport accompanies changes in the concentration of osmotically active molecules. Thus, when intracellular osmolarity exceeds that of the extracellular milieu, cell volume increases due to the movement of water into the cell. To cope with the resultant cell swelling, cells have developed a series of complex mechanisms to achieve a regulatory volume decrease, primarily through the activation of efflux mechanisms for intracellular ions. In particular, net electroneutral release of K ϩ -Cl Ϫ is achieved by K ϩ -Cl Ϫ cotransport, the simultaneous operation of K ϩ /H ϩ and Cl Ϫ /HCO 3 Ϫ exchangers, or through parallel, swelling-activated K ϩ and Cl Ϫ channels (1).
K ϩ -Cl Ϫ cotransport was first described in red blood cells as a swelling-and NEM 1 -activated K ϩ efflux mechanism (2,3), and red cells remain the primary model tissue for this class of ion transport. However, functional and physiological evidence has also been reported for the existence of K ϩ -Cl Ϫ cotransport in neurons (4), vascular smooth muscle (5), endothelium (6), epithelia (7,8), heart (9), and skeletal muscle (10). Consequently, K ϩ -Cl Ϫ cotransport has been implicated not only in regulatory volume decrease, but also in transepithelial salt absorption (8), renal K ϩ secretion (11), myocardial K ϩ loss during ischemia (9), and regulation of neuronal Cl Ϫ concentration (4). The physiological mechanisms invoked in cell volume regulation may also have broader roles in phenomena such as cell growth and apoptosis (1).
The extent of molecular heterogeneity in K ϩ -Cl Ϫ cotransport was unexpected, even after the identification of KCC1 and KCC2. In consequence, next to nothing is known about the functional and pharmacological properties of the four major KCC isoforms, or indeed of the physiological role of each isoform. One exception is the recent recognition that KCC2 encodes a developmentally regulated Cl Ϫ extrusion mechanism in neurons, with crucial secondary effects on the response to ␥-aminobutyric acid and other neurotransmitters that activate neuronal chloride conductance (4). To begin to understand the physiology and function of the individual KCCs, it is essential to characterize the functional properties of each isoform. KCC1 cDNAs from human, mouse, rabbit, pig, and Caenorhabditis elegans have been functionally expressed in human embryonic kidney cells (HEK 293) and in Xenopus laevis oocytes (13,17,18). Rat KCC2 and human KCC3 cDNAs have also been expressed in HEK 293 cells (15,19,20). From these studies, it is already quite clear that there are differences in the functional and pharmacological properties of the K ϩ -Cl Ϫ cotransporter isoforms. For example, kinetic analysis indicates that KCC2 exhibits significantly a higher affinity for potassium (14) than that of KCC1 (13) or KCC3 (20). In contrast, KCC2 is uniquely volume-insensitive, exhibiting minimal if any activation by cell swelling and considerable isoosmotic transport activity (21). In the present study, we extended the functional and pharmacological characterization of the recently cloned mouse KCC4 (16), using the Xenopus laevis oocyte expression system. The functional comparison of shark and human BSC2/NKCC1 and rabbit BSC1/NKCC2 has yielded important structure-function information for the Na ϩ -K ϩ -2Cl Ϫ cotransporters (22)(23)(24). With this paradigm in mind, we simultaneously studied the functional properties of KCC1 in our expression system and report significant functional and pharmacological differences between these representatives of the two molecular subgroups of the KCCs.

MATERIALS AND METHODS
Xenopus laevis Oocyte Preparation-Adult female Xenopus laevis frogs were purchased from Carolina Biological Supply Company (Burlington, NC) and maintained at the Institution animal facility under constant control of room temperature and humidity, 16°C and 65%, respectively. Frogs were fed with frog brittle dry food from Carolina Biological Supply Company, and water was changed twice a week. Oocytes were surgically collected from anesthetized animals under 0.17% tricaine and incubated for 1 h with vigorous shaking in frog Ringer ND96 (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl, and 5 mM Hepes, pH 7.4) in the presence of 2 mg/ml of collagenase B. Oocytes were then washed four times in ND96, manually defolliculated, and incubated overnight in ND96 at 18°C. On the next day, stage V-VI oocytes (25) were injected with 50 nl of water containing 0.25-0.5 g/l of cRNA in vitro transcribed from mouse KCC4 (KCC4) or rabbit KCC1 (KCC1) cDNA. Oocytes were incubated at 18°C for 4 days in ND96 supplemented with 2.5 mM sodium pyruvate and 5 mg/100 ml gentamicin; this incubation medium was changed every 24 h. On the day of the experiment, oocytes were switched to Cl Ϫ -free ND96 (96 mM Na ϩisethionate, 2 mM K ϩ gluconate, 6.0 mM Ca 2ϩ gluconate, 1.0 mM Mg 2ϩ gluconate, 5 mM Hepes, 2.5 mM sodium pyruvate, 5 mg/100 ml gentamicin, pH 7.4) 2 h prior to the uptake assay.
The full-length KCC4 and KCC1 cDNAs were previously subcloned into the high expression vector pGEMHE (16); rabbit KCC1 was a gift of Dr. Bliss Forbush III. To prepare cRNA, the KCC1 and KCC4 cDNAs were linearized at their 3Ј-ends with NheI and then transcribed in vitro using the T7 RNA polymerase mMESSAGE kit (Ambion). Transcription product integrity was confirmed on agarose gels, and concentration was determined by absorbance reading at 260 nm (DU 640, Beckman, Fullerton, CA). cRNA was stored frozen in aliquots at Ϫ80°C until used.
Assessment of K ϩ -Cl Ϫ Cotransport-K ϩ -Cl Ϫ cotransport was assessed by measuring tracer 86 Rb ϩ uptake (NEN Life Science Products) in experimental groups of at least 15 oocytes. Since both KCC4 and KCC1 express minimal activity under isotonic conditions (see "Results"), 86 Rb ϩ uptake was generally assessed in oocytes swollen by a 30-min incubation period in a hypotonic K ϩ and Cl Ϫ -free medium (50 mM N-methyl-D-glucamine (NMDG) gluconate, 4.6 mM Ca 2ϩ gluconate, 1.0 mM Mg 2ϩ gluconate, 5 mM Hepes, pH 7.4) with 1 mM ouabain, followed by a 60-min uptake period in a hypotonic Na ϩ -free medium with variable K ϩ -Cl Ϫ content. K ϩ and Cl Ϫ concentrations were varied separately using combinations of KCl, NMDG chloride, potassium gluconate, and NMDG gluconate, for a maximal total concentration of 50 mM; an uptake solution with 50 mM K ϩ -Cl Ϫ did not contain NMDG chloride, potassium gluconate, or NMDG gluconate, for example. All uptake solutions also contained 1.8 mM CaCl 2 , 1 mM MgCl 2 , 5 mM Hepes, pH 7.4, and were supplemented with 1 mM ouabain and 5.0 Ci/ml 86 Rb ϩ . Isotonic conditions were generated by supplementing the same solutions with 3.5 g/100 ml sucrose to reach isosmolar conditions for oocytes (ϳ210 mosmol/kg). Ouabain was added to prevent 86 Rb ϩ uptake via the Na ϩ -K ϩ -ATPase. The absence of extracellular Na ϩ and the hypotonicity of the uptake medium prevented 86 Rb ϩ uptake via the endogenous Na ϩ -K ϩ -2Cl Ϫ cotransporter that is present in oocytes (26).
All uptakes were performed at 32°C temperature. At the end of the uptake period, oocytes were washed five times in ice-cold uptake solution without isotope to remove extracellular fluid tracer. Oocytes were dissolved in 10% SDS, and tracer activity was determined for each oocyte by ␤-scintillation counting.
To determine the ion transport kinetics of KCC4 and KCC1, we performed experiments using varying concentrations of K ϩ and Cl Ϫ . The sensitivity for several inhibitors was assessed by exposing groups of oocytes to the inhibitors at concentrations varying from 20 M to 2 mM. For these experiments, the desired concentration of the inhibitor was present during both the incubation and uptake periods, except when noted.
Statistical Analysis-Statistical significance is defined as two-tailed p Ͻ 0.05, and the results are presented as mean Ϯ S.E. The significance of the differences between groups was tested by one-way analysis of variance with multiple comparison using Bonferroni correction or by the Kruskal-Wallis one-way analysis of variance on ranks with the Dunn's method for multiple comparison procedures, as needed.

Heterologous Expression of KCC4 and KCC1 in Xenopus
Oocytes-In isotonic conditions, no differences were observed among KCC4, KCC1, and water-injected oocytes (data not shown). When uptakes were performed under hypotonic conditions, microinjection of KCC4 and KCC1 cRNAs resulted in significant K ϩ -Cl Ϫ cotransport activity, as compared with control oocytes that were injected with water. Fig. 1 summarizes five experiments in which oocytes from different frogs were injected with water or KCC4 or KCC1 cRNA, followed by 86 Rb ϩ uptake assay using a hypotonic uptake solution containing 10 and 50 mM of extracellular K ϩ and Cl Ϫ , respectively. In control oocytes, 86 Rb ϩ uptake was 588 Ϯ 91 pmol⅐oocyte Ϫ1 ⅐h Ϫ1 in the presence of Cl Ϫ and 147 Ϯ 23 pmol⅐oocyte Ϫ1 ⅐h Ϫ1 in the absence of Cl Ϫ , indicating the presence of an endogenous K ϩ -Cl Ϫ cotransporter. Microinjection of KCC4 cRNA resulted in an increased 86 Rb ϩ uptake to 24,457 Ϯ 3,476 pmol⅐oocyte Ϫ1 ⅐h Ϫ1 . This 86 Rb ϩ uptake was Cl Ϫ -dependent, in that uptake in KCC4 oocytes in the absence of extracellular Cl Ϫ was 1723 Ϯ 402 pmol⅐oocyte Ϫ1 ⅐h Ϫ1 . In oocytes microinjected with KCC1, 86 Rb ϩ uptake increased to 12,632 Ϯ 2205 pmol⅐oocyte Ϫ1 ⅐h Ϫ1 , and the influx was Cl Ϫ -dependent. The difference in the amount of uptake between KCC4 and KCC1 was statistically significant (p Ͻ 0.05). Although equal amounts of KCC4 and KCC1 cRNA were injected for all experiments that we performed during the study, using multiple cRNA preparations, the relative expression level under hypotonic conditions has always been KCC4 Ͼ KCC1. In addition, for each KCC, the absolute uptake varied from frog to frog. Although KCC1 and KCC4 were also compared head-to-head using oocytes from the same frog for individual experiments, to ease the comparison of the two isoforms we present much of the results as the percentage of Cl Ϫ -dependent 86 Rb ϩ uptake. Thus, 100% generally represents the uptake observed in the KCC4 or KCC1 control group minus uptake observed in the water-injected oocytes.
Inhibitor Profile of KCC4 and KCC1-The effect of the loop diuretics furosemide and bumetanide was initially assessed using two different concentrations of extracellular K ϩ : 2 and 50 mM. In uptake medium with a K ϩ concentration of 2 mM, relative KCC4 activity was 61 Ϯ 3 and 90 Ϯ 4% in the presence of 2 mM furosemide or bumetanide, respectively. Interestingly, the inhibition of KCC4 by loop diuretics was augmented when the uptake medium contained 50 mM K ϩ ; under these conditions, the KCC4 activity was 9 Ϯ 4 or 17 Ϯ 4% in the presence of furosemide or bumetanide, respectively. In contrast, for KCC1 this effect of extracellular K ϩ was not observed for furosemide and was marginal for bumetanide. KCC1 function in the presence of furosemide was 9 Ϯ 2% in 2 mM K ϩ and 18 Ϯ 8% in 50 mM K ϩ (p not significant), and in the presence of bumetanide it was 51 Ϯ 12 versus 19 Ϯ 7% in 2 and 50 mM K ϩ , respectively (p ϭ 0.05; t ϭ 1.99). To further define the differences in the K ϩ effect on the sensitivity to loop diuretics between KCC4 and KCC1, we assessed the inhibitory effect of furosemide and bumetanide at several concentrations of extracellular K ϩ . The results of these series of experiments are shown in Fig. 2. The percentage inhibition of KCC4 by both furosemide and bumetanide was significantly affected by extracellular K ϩ (Fig. 2, upper panels). The minimal and maximal inhibition by both loop diuretics was observed at 2 and 6 mM, respectively; no further effect was observed at higher K ϩ concentrations. In contrast, the percentage of KCC1 inhibition by either furosemide or bumetanide did not vary as a function of extracellular K ϩ concentration (Fig. 2, lower panels). Thus, to define differences between the two KCCs in sensitivity to loop diuretics, we used a 10 mM concentration of extracellular K ϩ to assess the concentration curves for furosemide and bumetanide inhibition upon the Cl Ϫ -dependent 86 Rb ϩ uptake induced by KCC4 or KCC1. As Fig. 3 illustrates, KCC4 exhibits apparent half-maximal inhibition (K 0.5 ) values of ϳ900 M for both furosemide and bumetanide. These are lower than the respective values for KCC1 (ϳ180 M for furosemide and bumetanide). Therefore, KCC4 clearly exhibits a lower affinity for loop diuretics than does KCC1. The inhibition of KCC1 by furosemide in Fig. 3 suggests the possibility of a second affinity site for the loop diuretic. However, this inhibition fitted well to a Michaelis-Menten kinetics pattern with one inhibitor-binding site. The data did not fit to an equation with two binding sites (data not shown).
The sensitivity of the KCCs to other inhibitors of red cell K ϩ -Cl Ϫ cotransport was also assessed in oocytes injected with KCC4 or KCC1. Fig. 4 illustrates the effect of 100 M DIDS and 100 M DIOA on the 86 Rb ϩ uptake induced by the microinjection of each KCC cRNA. The effect of extracellular K ϩ concentration on the inhibition of cotransport was very dramatic for DIDS. When the concentration of extracellular K ϩ was 2 mM, the addition of DIDS to the extracellular medium resulted in reduction of KCC4 function to 65 Ϯ 10% (p Ͻ 0.003) and of KCC1 to 85 Ϯ 6% (p ϭ 0.113, not significant). In contrast, when 50 mM of extracellular K ϩ was used, DIDS resulted in significant decrease of KCC4 and KCC1 to 13 Ϯ 4 and 13 Ϯ 2%, respectively. The addition of 100 M of DIOA to the extracellular medium also resulted in inhibition of the KCCs. However, inhibition of KCC4 was higher when extracellular K ϩ was lower, although this was not the case for KCC1. DIOA is reportedly specific for K ϩ -Cl Ϫ cotransport over Na ϩ -K ϩ -2Cl Ϫ cotransport (27), and the same concentration of DIOA had no effect on the function of the Na ϩ -K ϩ -2Cl Ϫ cotransport activity of Xenopus oocytes (26) (data not shown).
We also tested the effect of a 2 mM concentration of the thiazide diuretic trichlormethiazide on the percentage of chlo- In all experiments, the Cl Ϫ concentration of the extracellular medium was 50 mM, whereas the [K ϩ ] e increased from 2 to 20 mM. The mean uptake in the absence of loop diuretic for each KCC was taken as the 100% of uptake, and data from diuretic-treated groups was normalized to the uptake in this control group. Experimental groups were exposed to a 2 mM concentration of furosemide or bumetanide during the incubation and uptake periods. Each point represents the mean Ϯ S.E. of at least 15 oocytes.
ride-dependent 86 Rb ϩ uptake. Surprisingly, given the supposed specificity of thiazides for Na ϩ -Cl Ϫ cotransport (28), KCC4 was moderately sensitive to trichlormethiazide. As we observed with furosemide and DIDS, the higher the extracellular K ϩ , the higher the inhibition by thiazides, since in 2 mM of extracellular K ϩ 86 Rb ϩ uptake was reduced to 79 Ϯ 3%, and at 50 mM it was reduced to 57 Ϯ 9%. This difference was significant (p Ͻ 0.01). In KCC1-injected oocytes, trichlormethiazide reduced 86 Rb ϩ uptake by a statistically significant amount to 64 Ϯ 4% in 2 mM K ϩ ; this inhibitory effect was not statistically significant at 50 mM K ϩ (74 Ϯ 8% reduction in activity). Consistent sensitivity to trichlormethiazide is thus unique to KCC4.
Independent studies have suggested that barium can inhibit renal K ϩ -Cl Ϫ cotransporters (7,8,29). We thus assessed the effect of 10 mM extracellular barium on the function of KCC4 and KCC1. Fig. 5 shows that when 10 mM BaCl 2 was added to the uptake medium, KCC4-induced influx was reduced to 58 Ϯ 4.3% of the uptake observed in KCC4-injected control oocytes. KCC1 function was only reduced to 79 Ϯ 4.2%, hence the inhibitory effect of barium was significantly greater for KCC4 than for KCC1 (p Ͻ 0.01).
Kinetic Properties of KCC4 and KCC1-To determine and compare the kinetic properties of KCC4 and KCC1 in the same expression system, we measured 86 Rb ϩ uptake in KCC4-and KCC1-injected oocytes as a function of the concentration of each transported ion. The results of these series of experiments are depicted in Fig. 6. Uptakes were performed with K ϩ or Cl Ϫ fixed at 50 mM, with changing concentrations of the counterion from 0 to 50 mM. Uptakes were also measured in water injected oocytes (data not shown), and the mean values for water groups were subtracted from corresponding KCC groups in order to assess only the 86 Rb ϩ uptake mediated by each heterologously expressed isoform. As shown in Fig. 1, 86 Rb ϩ uptake in waterinjected oocytes was low, such that this correction was generally minor. In the case of KCC4, 86 Rb ϩ influx increased as the concentration of each transported ion was raised, until a plateau phase was reached at ion concentrations greater than 20 -40 mM, compatible with Michaelis-Menten behavior. The calculated apparent K m and V max for extracellular K ϩ concentration were 17  Rb ϩ uptake was assessed in control groups of 2 and 50 mM K ϩ concentration, in the absence of inhibitors, and experimental groups were exposed to a 100 M of DIDS or DIOA during incubation and uptake periods. Each bar represents the mean Ϯ S.E. of at least 15 oocytes. Anion Dependence of KCC4 and KCC1-It has been shown that some extracellular anions other than Cl Ϫ can support ion translocation through the K ϩ -Cl Ϫ cotransporter of both sheep and human erythrocytes (30). It was thus of interest to measure 86 Rb ϩ transport by KCC4 and KCC1 in the presence of different anions. The 86 Rb ϩ influx of KCC4-and KCC1-injected oocytes using an uptake solution containing 40 mM potassium gluconate and 10 mM KCl served as the reference activity for these experiments, as compared with uptake activity in oocytes exposed to medium containing 40 mM potassium gluconate and 10 mM of KBr, KH 2 PO 4 , KI, potassium gluconate, or KSCN. Fig. 7 shows the percentage of KCC4 (upper panel) and KCC1 (lower panel) function when uptakes were performed using these different anion substitutions. KCC4 shows the higher 86 Rb ϩ influx in the presence of 10 mM KCl. 86 Rb ϩ influx was still observed in the presence of other anions: 58 Ϯ 9% with 10 mM KBr, 22 Ϯ 5.9% with 10 mM KH 2 PO 4 , and 17 Ϯ 3.8% with KI, whereas potassium gluconate and KSCN did not support transport. These results are in contrast to those observed in KCC1-injected oocytes, for which the order of anion-supported transport was Cl Ϫ Ͼ SCN Ϫ ϭ Br Ϫ Ͼ PO 4 Ϫ3 Ͼ I Ͼ gluconate. Regulation of KCC4 and KCC1-One of the most distinctive characteristics of K ϩ -Cl Ϫ cotransport in several cells and species is activation by the alkylating agent NEM (2). We therefore analyzed the effect of NEM on 86 Rb ϩ influx in groups of oocytes under isotonic or hypotonic conditions. Again, in all of the experiments in which we assessed 86 Rb ϩ influx in oocytes that were incubated in isotonic medium (ϳ 210 mosmol/kg), the uptake observed in KCC4-or KCC1-injected oocytes was not different from the uptake in water-injected oocytes. However, the addition of 1 mM NEM in isotonic conditions resulted in a 5-fold activation of KCC4 (214 Ϯ 12 pmol⅐oocyte Ϫ1 ⅐h Ϫ1 in the KCC4 control group versus 1062 Ϯ 70 pmol⅐oocyte Ϫ1 ⅐h Ϫ1 in the NEM-treated group, p Ͻ 0.001) and a 2.6-fold activation of KCC1 (120 Ϯ 27 versus 319 Ϯ 76 pmol⅐oocyte Ϫ1 ⅐h Ϫ1 , p Ͻ 0.05) (Fig. 8, A and B). Of note, when uptakes were performed in hypotonic medium, the addition of NEM resulted in a dramatic inhibition of both isoforms (Fig. 8, C and D), such that 86 Rb ϩ  KCC4 (A and B) or KCC1 (C and D). A and C, K ϩ dependence of 86 Rb ϩ uptake for each KCC. B and D, Cl Ϫ dependence of 86 Rb ϩ uptake for each KCC. Uptakes were performed with K ϩ or Cl Ϫ fixed at 50 mM, varying the concentration of the appropriate counterion from 0 to 50 mM, as indicated. Uptakes were also measured in water-injected oocytes (data not shown), and the mean values for the corresponding water groups were subtracted to analyze only the 86 Rb ϩ uptake due to each injected KCC. Lines were fit using the Michaelis-Menten equation. Data are expressed as uptakes in pmol⅐oocyte Ϫ1 ⅐h Ϫ1 , each point represents the mean of at least 25 oocytes. uptakes induced by KCC4 and KCC1 were reduced by 68 and 55%, respectively. In the same experiments, 86 Rb ϩ uptake due to the endogenous oocyte K ϩ -Cl Ϫ cotransporter (H 2 O-injected oocytes) was significantly increased when uptakes were done under both isotonic and hypotonic conditions (data not shown).
It has been known for some time that the inhibition of protein phosphatases prevents the activation of red cell K ϩ -Cl Ϫ cotransport by either cell swelling or NEM. Since the role of phosphatases in the control of the cloned KCCs is unclear, we studied the effect of three inhibitors of protein phosphatases. We used 100 nM calyculin A, which inhibits the function of protein phosphatases 1 and 2A. The relative role of specific phosphatases was assessed using okadaic acid at 1 nM, a concentration that only affects protein phosphatase 2A, and cypermethrin at 100 pM, a concentration in which this compound inhibits the function of protein phosphatase 2B. As Fig. 9 shows, the addition of calyculin A completely prevents the activation of KCC4 and KCC1 by cell swelling. In contrast, neither okadaic acid, nor cypermethrin prevented this activation. These results indicate that protein phosphatase 1 is required for the activation of both KCC4 and KCC1 by cell swelling. DISCUSSION We have recently shown (16) that heterologous expression of the mouse KCC4 cDNA induced the expression of a 86 Rb ϩ influx pathway that is activated by cell swelling, dependent on the presence of extracellular Cl Ϫ , and inhibited by 2 mM of the loop diuretic furosemide. These data established that KCC4 functions as a K ϩ -Cl Ϫ cotransporter. The present study extends these initial observations and defines the functional properties of KCC4 in greater detail. In addition, we have studied the functional properties of KCC1, the other widely distributed KCC isoform, using the same experimental protocols. Since sequence comparisons of the four KCC proteins pair KCC4 with KCC2 and KCC1 with KCC3, this study also constitutes the first direct functional comparison of the two subgroups of the KCCs.
Our data indicate that KCC1 and KCC4 express minimal K ϩ -Cl Ϫ cotransport in unstimulated cells under isotonic conditions but are strongly activated by cell swelling induced by hypotonic conditions. In our expression system, neither KCC4nor KCC1-injected oocytes exhibited a significant increase of 86 Rb ϩ uptake over water-injected oocytes when incubated in isotonic media during the influx period. When oocytes were incubated in hypotonic conditions, however, both cotransport-ers were markedly activated, albeit with a different magnitude (KCC4 Ͼ KCC1). These findings differ from previous observations (13,15,17,19,20) that indicate minimal hypotonic activation of KCC1, KCC2, and KCC3 when these cotransporters were expressed in HEK 293 cells, but agree with the cell swelling-induced activation of KCC1 found by Su et al. (18) using Xenopus oocytes as an expression system. Thus, when expressed in Xenopus oocytes, KCC4 and KCC1 cotransporters can be activated by cell swelling, suggesting that HEK 293 cells may not possess the appropriate signaling pathways for the activation of the cotransporters by swelling. In comparison, volume-regulated transport pathways recapitulate their in vivo physiology when expressed in Xenopus oocytes. Thus, shrinkage-activated transport pathways such as the Na ϩ -K ϩ -2Cl Ϫ cotransporter (26) or the epithelial sodium channel ENaC (31) and swelling-activated pathways such as the calcium-activated intermediate K ϩ channel mIK1 (32) are regulated appropriately in this expression system. Of the four KCCs, KCC4 seems to be the isoform that exhibits the highest activation by hypotonicity. Influx mediated by KCC4 in this study was consistently higher than KCC1, although the amount of injected cRNA was equivalent. There are, however, several possible explanations for the greater hypotonic activation of KCC4. For example, KCC4 cRNA may be intrinsically more stable or better translated than KCC1 in oocytes. Of note, however, rat KCC2 reportedly encodes a K ϩ -Cl Ϫ cotransporter with significant activity under isotonic conditions when expressed in Xenopus oocytes and only minimal activation by cell swelling (21). Therefore, it is likely that much of the observed differences in volume sensitivity is due to variation in the structure of the four KCC proteins.
The two major loop diuretics inhibit KCC4, with an inhibitor sensitivity that is lower than that observed for KCC1. The reported effect of external potassium ([K ϩ ] e ) on the inhibition of K ϩ -Cl Ϫ cotransport by loop diuretics (33) was observed for KCC4 but not KCC1. We observed a significantly different effect of furosemide and bumetanide on KCC4 at variable [K ϩ ] e , with the minimal and maximal effect at [K ϩ ] e of 2 and 6 mM, respectively. In contrast, no effect of [K ϩ ] e was observed for the inhibition of KCC1 by loop diuretics. This finding suggests that, as the isoform with the lower inhibitor affinity, the inhibition of KCC4 is more dependent on the positive effect of [K ϩ ] e on the interaction between the transporter protein and loop diuretics. Similar to our loop diuretic experiments, the anion transport inhibitor DIDS inhibited the function of KCC4 and KCC1, with an apparent K i that was dramatically lowered by an increase in [K ϩ ] e . Almost no effect was observed at 2 mM [K ϩ ] e , while in 50 mM [K ϩ ] e , 86 Rb ϩ influx was completely blocked by a 100 M DIDS concentration. This relationship between [K ϩ ] e and the inhibition of the K ϩ -Cl Ϫ cotransporter by DIDS was previously observed in low potassium sheep red blood cells (34) and was explained by the existence of two sites for K ϩ in the cotransporter: a modifier site and a transport site. Of interest, DIDS can also inhibit the function of the thiazidesensitive Na ϩ -Cl Ϫ cotransporter but has no effect on the bumetanide-sensitive Na ϩ -K ϩ -2Cl Ϫ cotransporter (26). The acid alkaloid DIOA, considered a specific inhibitor of red cell K ϩ -Cl Ϫ cotransport (27), also inhibited KCC4 and KCC1. However, in contrast to DIDS and loop diuretics, all of which primarily inhibit anion transporters and exchangers (35,36), the higher the [K ϩ ] e , the lower the efficacy of DIOA. However, even in a very high [K ϩ ] e (50 mM), the inhibition of 86 Rb ϩ influx by 100 M DIOA was still greater than 50%.
The increased 86 Rb ϩ uptake induced by KCC4 was also inhibited by about 20% in low and 40% in high [K ϩ ] e by 2 mM concentration of the thiazide-diuretic trichlormethiazide. The members of the electroneutral cation chloride coupled cotransporters have been defined in part due to their sensitivity to diuretics. The Na ϩ -K ϩ -2Cl Ϫ cotransporters are sensitive to loop diuretics, derivates of sulfamoylbenzoic acid, and resistant to the benzothiadiazine derivates, whereas the Na ϩ -Cl Ϫ cotransporter is inhibited by thiazides but not affected by loop diuretics (12). Our results suggest that KCC4, which exhibits a low degree of identity with the sodium-dependent cation-chloride cotransporters (ϳ22%), can be inhibited not only by loop diuretics but also by thiazide-type diuretics. A similar observation has been reported by Harling et al. (37), who showed that the plant cation-chloride cotransporter AXI 4, which exhibits the highest sequence identity with the KCCs (36 -38%), can also be inhibited by bumetanide, furosemide, and the thiazide-like diuretic metolazone.
We have found that KCC4 and KCC1 can be blocked by the addition of 10 mM BaCl 2 to the uptake medium, with a relative sensitivity of KCC4 Ͼ KCC1. Although red cell K ϩ -Cl Ϫ cotransport is at least partially sensitive to quinidine derivatives (38), this is the first indication that the cloned K ϩ -Cl Ϫ cotransporters are directly sensitive to BaCl 2 . This observation is also consistent with the controversial proposal by Greger and Schlatter (7) that the basolateral membrane of renal thick ascending limb cells contains a barium-sensitive K ϩ -Cl Ϫ cotransporter. The observation by Amlal et al. (8) that a thick limb chloride-dependent, barium-sensitive NH 4 ϩ transport mechanism is only modestly sensitive to furosemide suggests that KCC4 is the isoform present in the basolateral membrane of the mammalian thick ascending limb. In this regard, Liapis et al. (39) failed to detect KCC1 mRNA in human thick ascending limb.
Kinetic analyses reveal that KCC4 and KCC1 exhibit very similar affinities for extracellular Cl Ϫ (K m values in 16.1 Ϯ 4.2 and 17.2 Ϯ 8.3 mM, respectively), whereas the differences in the affinity for extracellular K ϩ approached statistical significance (K m values 17.5 Ϯ 2.7 and 25.5 Ϯ 3.2 mM, respectively; p ϭ 0.08). Our results for KCC1 agree with those reported by Gillen et al. (13). It is known that the central core of 12 transmembrane (TM) segments determines the kinetic properties of the cation-chloride cotransporters (40). In this regard, the otherwise identical TM segments of the four KCC proteins differ primarily at amino acid residues within TM2, TM4, and TM7. An elegant series of studies have implicated TM2 in the determination of cation (Na ϩ and K ϩ ) affinity in the Na ϩ -K ϩ -2Cl Ϫ cotransporters, whereas residues within TM4 and TM7 appear to affect anion affinity (23,24,40,41). However, sequence comparisons of the entire gene family indicate that only TM7 is particularly conserved between the Na ϩ -dependent and Na ϩindependent cation-chloride cotransporters; thus, these observations may not translate to the K ϩ -Cl Ϫ cotransporters. Of particular significance, however, the reported affinity of KCC2 for K ϩ (ϳ5 mM) is closer to that of KCC4 than KCC1, and KCC2 shares significant identity with KCC4 within TM2. Therefore, as in the Na ϩ -K ϩ -2Cl Ϫ cotransporters, TM2 may play a major role in the determination of cation affinity. The K ϩ affinity of individual K ϩ -Cl Ϫ cotransporters may be of major physiological significance, since under conditions wherein extracellular K ϩ increases, such as cardiac ischemia (9) and neuronal activity (4), the higher affinity KCCs may function as K ϩ -Cl Ϫ influx pathways.
KCC4 and KCC1 exhibit surprisingly similar affinities for chloride. However, they do differ in another parameter of anion transport, the anion selectivity or "anion-series" of K ϩ -Cl Ϫ cotransport. KCC4 and KCC1 thus differ in the profile of 86 Rb ϩ transport that can be sustained by different anions. In KCC4, about 50% of the function can be observed in the presence of Br Ϫ , and some transport is still present with PO 4 Ϫ3 , whereas in KCC1 70% of 86 Rb ϩ uptake can be obtained in the absence of Cl Ϫ , when either Br Ϫ or SCN Ϫ are present in the extracellular medium. Whether these differences are encoded by subtle variation in TM4 and TM7 will require further study; however, differences in anion selectivity were crucial for the identification of the anion channel pore in the CLC chloride channels (42,43).
The functional properties of KCC4 and KCC1 observed in the present study suggest that it is unlikely that either of these isoforms is the predominant K ϩ -Cl Ϫ cotransporter expressed in red blood cells. On the one hand, Delpire and Lauf (44) observed that the K ϩ -Cl Ϫ cotransporter in hyposmotically swollen low K sheep erythrocytes exhibited a K m for extracellular K ϩ of ϳ55 mM, which is very different from the K m obtained in the present study for KCC4 and KCC1. It has also been shown in sheep red blood cells (45) that K ϩ influx is higher in the presence of Br Ϫ than in the presence of Cl Ϫ ; neither KCC4 nor KCC1 exhibited this behavior. Moreover, although it has been shown that KCC1 mRNA is expressed in mouse erythroleukemic cells, it is not present in circulating reticulocytes (46), and the KCC2 isoform is expressed exclusively in the central nervous system (14). Thus, taking all the information together, our kinetic and anion substitution experiments suggest that the major K ϩ -Cl Ϫ cotransporter in erythrocytes is either KCC3 or an as-yet-unidentified isoform.
K ϩ -Cl Ϫ cotransport was initially defined as a red cell transport pathway that is activated by the alkylating agent NEM. Multiple laboratories have since found that pretreatment of erythrocytes from several species with 1 mM NEM results in significant activation of this Cl Ϫ -dependent K ϩ transport pathway (2). It is still unclear if the activating effect of NEM is related to NEM-induced dephosphorylation, via activation of an upstream kinase, or to direct modification of the thiol groups on the cotransporter. There are reports supporting both possibilities (47) (for reviews, see Refs. 2 and 3). However, like the activation by cell swelling, NEM-activated transport is prevented by phosphatase inhibitors, suggesting a positive effect of NEM on upstream signaling pathways (48 -52). In this regard, our data show a very interesting behavior of the KCCs when exposed to NEM. Under isotonic conditions, NEM stimulated the function of KCC4 and KCC1, as well as the endogenous K ϩ -Cl Ϫ cotransporter of the oocytes. Su et al. (18) also observed KCC1 activation by NEM in oocytes. In contrast, in hypotonic conditions, when we exposed oocytes to NEM, both KCC4 and KCC1 were inhibited. The fact that KCC4 and KCC1 were activated by NEM in isotonic conditions suggests that oocytes possess the intracellular pathways that NEM requires for activation of the cotransporters. The mechanisms by which swelling-activated KCC4 and KCC1 are inhibited by NEM are still unclear. It has been shown in sheep red cells that K ϩ -Cl Ϫ cotransporters can be activated or inhibited by NEM through high and low affinity stimulatory thiols, respectively (47,53,54). Further experiments will be required to clarify this issue; however, a reconciliation of these and previous observations is that there are direct inhibitory sites on the transporter proteins themselves and stimulatory sites on upstream kinases. In this regard, there are several transmembrane or juxtamembrane cysteines in the predicted KCC proteins, and transmembrane cysteines were recently implicated in the differential sensitivity of Na ϩ -K ϩ -2Cl Ϫ cotransporters to cysteine-reactive compounds (55).
Over the last decade or so, several laboratories have suggested that dephosphorylation of the K ϩ -Cl Ϫ cotransporter is required for its activation, since inhibition of protein phosphatases prevents the swelling-and NEM-induced activation (48 -50, 56, 57). Our data support this hypothesis, since the protein phosphatase inhibitor calyculin A completely abrogates hypotonic activation of KCC4 and KCC1. Calyculin A is known to inhibit both protein phosphatase 1 and 2A (58). To discriminate between phosphatases, we also tested the effect of okadaic acid in a concentration of 1 nM, which inhibits only protein phosphatase 2A, and cypermethrin, which inhibits only protein phosphatase 2B (59,60). Since these two compounds did not affect activation of KCCs, we suggest that, at least in Xenopus oocytes, protein phosphatase 1 is the phosphatase that is involved in activation of the KCCs during cell swelling. In addition to a role in cell volume regulation, activation of K ϩ -Cl Ϫ cotransport by protein phosphatase 1 may play a role in transepithelial transport of salt in the kidney and other epithelial organs. There is evidence for a swelling-activated basolateral K ϩ -Cl Ϫ cotransporter in the proximal tubule (61,62), where protein phosphatase 1 also is responsible for activating the basolateral Na ϩ -K ϩ -ATPase (63); protein phosphatase 1 may thus function to couple the pump to basolateral Cl Ϫ and K ϩ exit through the KCCs. Of more specific relevance to K ϩ -Cl Ϫ cotransport, the combined data for NEM, phosphatase inhibition, and cell swelling dissociates for the first time these various control points for this transport pathway.
In conclusion, we have found significant regulatory, kinetic, and pharmacological differences between KCC4 and KCC1.
Despite differences in their relative activation by cell swelling, KCC4 and KCC1 share a requirement for dephosphorylation by a protein phosphatase for swelling-induced activity. As previously shown for K ϩ -Cl Ϫ cotransport in red cells (48,50,56,64), the relevant protein phosphatase is probably protein phosphatase 1. The two KCC isoforms in this study differ slightly in affinity for K ϩ , presumably due to variation within transmembrane 2, a region of the cation-chloride cotransporter proteins previously implicated in cation affinity (22)(23)(24). Ion affinity may also be of physiological relevance, in that lower affinity K ϩ -Cl Ϫ cotransporters, such as KCC1 and the red cell K ϩ -Cl Ϫ cotransporter (potentially KCC3), may function exclusively as efflux mechanisms. As proposed initially by Payne (19), the higher affinity isoforms (KCC2 and now KCC4) may function as both efflux and influx pathways. Such a duality has been verified experimentally in neurons (65), where synaptic activity may increase extracellular K ϩ to the point that KCC2 mediates K ϩ -Cl Ϫ influx. KCC4 transcript is in turn particularly abundant in heart, where K ϩ efflux during ischemia appears to involve K ϩ -Cl Ϫ cotransport (9); again, a higher affinity isoform may play a role in reclaiming this intracellular K ϩ in the postischemic myocardium. Finally, the pharmacological characterization of KCC4 fulfills the prediction, based on observations of the physiology of renal thick ascending limb cells (7,66), that K ϩ -Cl Ϫ cotransporters may be sensitive to barium, widely considered a specific inhibitor of K ϩ channels.