Comparison of Na-K-Cl Cotransporters

The Na-K-Cl cotransporter (NKCC) mediates the coupled movement of ions into most animal cells, playing important roles in maintenance of cell volume and in epithelial Cl transport. Two forms of NKCC have been described: NKCC1, the “housekeeping” isoform that is also responsible for Cl accumulation in secretory epithelial cells, and NKCC2, which mediates apical Na+K+Cl entry into renal epithelial cells. Here we examine the kinetic properties of NKCC1, NKCC2, and the endogenous HEK-293 cell cotransporter. Stable expression of rabbit NKCC2A was obtained in HEK-293 cells utilizing a chimera (h1r2A0.7) in which the 5′-untranslated region and cDNA encoding 104 amino acids of the N terminus are replaced by the corresponding sequence of NKCC1. h1r2A0.7 exhibits Na and Cl affinities near those of NKCC1, but it has a 4-fold lower Rb affinity, and a 3-fold higher affinity for the inhibitor bumetanide. The activity of h1r2A0.7 is increased on incubation in low [Cl] media as is NKCC1, but the resting level of activity is higher in h1r2A0.7 and activation is more rapid. h1r2A0.7 exhibits an appropriate volume response, unlike NKCC1 for which concomitant changes in [Cl]i appear to be the overriding factor. These results support a model in which apical NKCC2 activity is matched to basolateral Cl exit through changes in [Cl]i. Reverse transcriptase-polymerase chain reaction of HEK-293 cell mRNA is positive with NKCC1 primers and negative with NKCC2 primers. Surprisingly, we found that the behavior of the endogenous HEK cell Na-K-Cl cotransporter is unlike either of the two forms which have been described: compared with NKCC1, HEK cell cotransporter has a 2.5-fold lower Na affinity, an 8-fold lower Rb affinity, and a 4-fold higher bumetanide affinity. These results suggest the presence of a novel isoform of NKCC in HEK-293 cells.

The Na-K-Cl cotransporter (NKCC or BSC) 1 mediates the coupled movement of Na, K, and Cl ions across the plasma membrane of animal cells. The transporter plays an important role in electrolyte movement across polarized epithelia and is also thought to be involved in regulation of intracellular volume and intracellular [Cl] (1,2). NKCC is a member of the Na-coupled group of cation-chloride cotransporters (CCCs) (1,3), a family which also includes K-Cl cotransporters (KCC) (3,4). Three Na-coupled cation-chloride cotransporters have been described to date. 1) The "secretory" (or "housekeeping" or "basolateral") Na-K-Cl cotransporter, NKCC1 (or BSC2), is widely distributed in mammalian tissues (5,6) and is especially prominent in the basolateral membranes of secretory epithelial cells; within the kidney, NKCC1 is found in epithelial cells in the collecting duct and in the glomerulus (7,8). 2) The "renal" or "apical" Na-K-Cl cotransporter, NKCC2 (or BSC1) (9,10), is found only in the apical membrane of epithelial cells in the thick ascending limb of the loop of Henle (TAL) (11)(12)(13)(14)(15). Three splice variants of NKCC2 (A, B, and F), differing in the sequence of the second predicted transmembrane domain, are differentially distributed along the nephron (9,11). 3) The Na-Cl cotransporter, NCC (or TSC) (16), is restricted to the apical membrane of the distal tubule in the mammalian kidney.
The activity of the Na-K-Cl cotransporter is increased in most cells in response to cell shrinkage, leading to a regulatory increase in cell volume (1). Additionally, in secretory epithelia, cotransporter activity is strongly regulated as part of the process controlling fluid secretion: it appears that a decrease in intracellular [Cl] is the message which triggers an increase in cotransport activity and thereby achieves apical-basolateral communication (17)(18)(19)(20). We have shown that for NKCC1, modulation of transport in response to both volume change and [Cl] i change involves direct phosphorylation of the NKCC1 protein (17,20).
We have recently used chimeras of human and shark NKCC1 to identify regions that are responsible for mediating the binding characteristics of the transporters, taking advantage of 5-fold species differences in kinetic constants for ion translocation and bumetanide inhibition (21). When the N and the C termini were interchanged between species, we found no significant change in kinetic parameters, indicating that it is the large central transmembrane domain of the NKCC protein that encodes the differences in ion and bumetanide binding.
The function of NKCC has been studied by expression of transporter cDNAs in mammalian cell lines (22) and in Xenopus oocytes (10). Mammalian expression systems offer considerable advantages in reproducibility and in the ability to perform assays under a large number of conditions. We have determined the characteristics of NKCC1-mediated transport using stable expression in HEK-293 cells (6, 23) but unfortunately have been unable to obtain functional expression of NKCC2 using the same methods (9). Similarly, it has been difficult to express NKCC1 in oocytes (5).
In this project we have been able to characterize ion transport mediated by NKCC2 utilizing a chimera (h 1 r 2A 0.7) in which 104 amino acids of the N terminus are replaced by corresponding residues of NKCC1. Apparently, translation or processing efficiency is higher with the NKCC1 5Ј-UTR and N terminus. Most of this region is very poorly conserved from one species to another and from one isoform to another, both for NKCC and KCC (3,24). Since our previous experiments demonstrate that neither the N nor the C terminus contributes to the differences in ion affinities between sNKCC1 and hNKCC1 (21), we do not anticipate that the N-terminal change in h 1 r 2A 0.7 significantly alters the function of NKCC2.
The HEK-293 cell line used for expression of NKCC in this and previous studies is derived from human embryonic kidney, immortalized by adenovirus transformation (25). HEK cells have a rather low level of endogenous ion fluxes, including Na-K-Cl cotransport. The de-differentiated line does not exhibit epithelial characteristics, and it is therefore not possible to predict which isoform of Na-K-Cl cotransporter might be present.
In this study, we compare the kinetic and regulatory behavior of NKCC1, h 1 r 2A 0.7, and the endogenous HEK cell cotransporter. We find that NKCC1 and NKCC2 (as h 1 r 2A 0.7) are different from one another in ion and bumetanide affinities as well as in the relative sensitivities to cell volume and [Cl]. Surprisingly, we find also that the endogenous Na-K-Cl cotransporter in HEK cells exhibits unique functional features, its behavior being different from that of both NKCC1 and NKCC2. Part of this work has been previously reported in abstract form (26).

EXPERIMENTAL PROCEDURES
The Chimera h 1 r 2A 0.7-h 1 r 2A 0.7 is composed of the entire coding region of the renal Na-K-Cl cotransporter (rNKCC2A), except that two-thirds of the N terminus is replaced by the corresponding region in hNKCC1 (Fig. 1). A common NcoI restriction site in hNKCC1 and rNKCC2A was used to create the chimeric junction 0.7 (Thr 217 /Met 218 in hNKCC1 and Thr 104 /Met 105 in rNKCC2A). This site occurs in a conserved region in the N terminus, 75 amino acids before the first putative transmembrane domain. The cDNA was prepared by simultaneous ligation of four fragments into the pJB20 expression vector (27) (6200-bp length) at EcoRI-KpnI sites. The fragments were: EcoRI-NcoI from hNKCC1 (900-bp length); and NcoI-SphI (250-bp length), SphI-XbaI (1440-bp length), and XbaI-KpnI (1650-bp length) from rNKCC2A. The final construct was analyzed by automated sequencing and restriction analysis.
Cell Lines-Control HEK cells, mock-transfected cells, and lines stably expressing sNKCC1 and hNKCC1 were the same as in Ref. 21. h 1 r 2A 0.7 cDNA was transfected into HEK cells by calcium phosphate precipitation, and stable lines were isolated by G-418 resistance, as described previously (6,21). T 84 cells, obtained from J. Madara, were as in Ref. 6. NIH-3T3 cells and the E12a mutant (28) were from T. G. O'Brien. All lines were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, penicillin, streptomy-cin, and G-418 (for transfected cells), as described previously (21). We have studied two lines of h 1 r 2A 0.7 from separate transfections; the results for the two lines were indistinguishable from one another, and we have combined the data as reported here. Similarly, several experiments with hNKCC1 were performed with a different line reported in Ref. 6, with indistinguishable results. All experiments with sNKCC1 were carried out with the cell line described in Ref. 21. This line appears to be functionally the same as a line described in Ref. 22 except that the cotransport flux is somewhat higher.
Each concentration curve was carried out in a single row of the 96-well plate, and in each experiment there were 2-6 replicate rows. Counts of the 1-min 86 Rb influx were normalized to the value at the highest ion concentration, or to the longest time point in an activation time course, or to the value of uninhibited flux in inhibition studies. In previous experiments, we determined that at confluence, the coefficient of standard variation of protein content in several wells is quite small, approximately 10% of the mean. This coefficient is similar to that calculated for absolute counts in several wells under the same conditions, and therefore, it has not been useful to routinely determine protein on a well-by-well basis. Rows with obvious rogue values were omitted from averages, which in all cases was less than one row in 10. Data are expressed as means Ϯ S.E. among all rows in several experiments (on average 20 -30 rows in 5-8 experiments). Similarly, K m and K i values were obtained on a per row basis by non-linear least squares curve fitting using the Simplex algorithm (program PLOT, B. Forbush). Where error bars are not visible, they are smaller than the symbols. We have not attempted to correct fluxes in transfected cells for a potential background contribution from HEK cell cotransporter because we have evidence that expression of exogenous cotransporter suppresses endogenous cotransporter. 2 RNA Extraction and Reverse-transcriptase-PCR-Poly(A)-selected RNA was isolated from T 84 cells, 3T3 fibroblasts, E12a cells, and HEK cells. Confluent cells from 10-cm dishes were homogenized and digested for 1 h in 200 g/ml proteinase K, 0.5% SDS, 100 mM NaCl, 20 mM Tris-Cl and 1 mM EDTA, at pH 8.0 and at 37°C. After adjusting NaCl concentration to 400 mM, the cell lysates were incubated with oligo(dT)cellulose for 4 h at room temperature. Poly(A) RNA was eluted in 1 mM EDTA, 0.05% SDS, pH 8.0, and concentrated by ethanol precipitation. Poly(A) RNA was primed with a gene-specific antisense oligonucleotide derived from a conserved region in the third transmembrane domain of either NKCC1 (gaatcacgactgtaatggctccaa; base pair 1422) or NKCC2 (ctccagagatgttggcaccagcaag); bp 1487). The primers were extended with 2 R. Behnke, P. Isenring, and B. Forbush, unpublished data. the enzyme AMV in an appropriate reaction buffer with 0.4 M dNTPs. For PCR, a sense oligonucleotide derived from another conserved sequence in the N terminus of NKCC1 (gcgcaccttcggccacaacaccatg; bp 794) or NKCC2 (ccgagttcggtgggtcaataggcttg; bp 1041) was added to the first cDNA strand, along with the DNA polymerases Taq and PWO.
The NKCC1 primers correspond to regions that are identical between human and mouse NKCC1; NKCC2 primers correspond to a region identical in mNKCC2 and rNKCC2 and with 1 and 2 mismatches in hNKCC2. The NKCC1 target sequences are less than 20% identical to corresponding NKCC2 sequences, and NKCC2 target sequences are, respectively, less than 20 and 50% identical to corresponding NKCC1 regions.
The region of interest (630 bp in NKCC1 and 440 bp in NKCC2) was amplified through 40 cycles of PCR with denaturation steps of 40 s at 95°C, annealing 50 s at 50°C, and extension 1 min at 70°C. To test for amplification yields, 0.1 ng of plasmid DNA (either hNKCC1/Bluescript SK Ϫ or rNKCC2A/Bluescript SK Ϫ ) was processed in the same manner as the poly(A) RNA, including an incubation period with reverse transcriptase. Samples of PCR reactions were analyzed on a 1.2% agarose gel stained with ethidium bromide (10 g/ml).

RESULTS
Functional Expression of NKCC2-In previous efforts we have been unable to measure ion fluxes of rNKCC2 transfected in mammalian cell lines (COS cells, HEK, 3T3, Madin-Darby canine kidney (MDCK)), 3 apparently because of low levels of expression or poor cell surface delivery (9). Also, truncation and modification of the 5Ј-UTR of rNKCC2 did not increase functional expression. 3 We have successfully approached the problem using a chimera, h 1 r 2A 0.7, in which the 5Ј-UTR and cDNA encoding the first 104 amino acids of rNKCC2A were replaced with the corresponding region from hNKCC1. The N terminus of the cation-chloride cotransporters is very poorly conserved across isoforms and species (3,24), and we have shown that it does not play a role in sNKCC1/hNKCC1 ion affinity differences (21). When stimulated by preincubation in low [Cl] medium (see below), h 1 r 2A 0.7-transfected cell lines were found to transport 86 Rb about 2.5-fold faster than control HEK cells (data not shown). 4 This is a lower level of transport than obtained on transfection of sNKCC1 (4 -6-fold above control) or hNKCC1 (7-9-fold above control), but as will be seen below, the properties of h 1 r 2A 0.7 and HEK-cell cotransporter are readily distinguished.
Kinetic Behavior of Cotransporters-To compare the transport behavior of NKCC isoforms, we measured the dependence of 86 Rb influx on Na, Rb, Cl, and bumetanide concentration for sNKCC1, hNKCC1, h 1 r 2A 0.7, as well as for HEK cells. The results are illustrated in Figs. 2 and 4, and the K m for Na, Rb, and Cl and the K i for bumetanide are summarized in Fig. 3. As has been consistently noted for the Na-K-Cl cotransporter, Na, Rb, and bumetanide dependences of 86 Rb influx fit a model of ligand binding at a single site (Fig. 2), whereas the relation between [Cl] and 86 Rb influx is sigmoidal, consistent with two binding and translocation sites for Cl (Fig. 4). The data generally agree well with values that have been previously reported by this laboratory in separate studies (6,21,22). 5 The kinetic data demonstrate that NKCC2A (as h 1 r 2A 0.7) presents distinct ligand binding characteristics. The K m of h 1 r 2A 0.7 for Na is 40% lower than that of hNKCC1, and the K m for Rb is 4-fold higher compared with hNKCC1. In comparison to previous measurements of K m values for ion transport in mouse kidney cortical TAL (29) and in a mouse medullary TAL cell line (30), K m(Cl) is in the same range (34 compared with Ӎ50 and 67 mM, respectively), K m(Rb) is significantly greater (8 compared with 1-2 and 1.3 mM), and K m(Na) is similar to that reported for the TAL line (10 versus 7 mM) but greater than recorded in the cortical TAL (2-3 mM). While some of these differences may be attributable to methodology, it is also possible that they reflect differences in ion affinities of the three splice variants of NKCC2 that are differentially distributed along the TAL (9,11).
In examining the kinetic behavior of endogenous HEK cell cotransport, we were surprised to find that it was significantly different both from NKCC1 (noted previously in Ref. 6) and from h 1 r 2A 0.7. In particular, Na and Rb affinities are much 3 J. Payne and B. Forbush, unpublished data. 4 We have been unable to obtain functional expression of B and F splice variants of NKCC2 using similar chimeric constructs (J. Payne and B. Forbush, unpublished data). 5 The values for K i(bumetanide) are rather different from one report to another; the values reported here are 2-fold higher than in Ref. 22 and Ref. 6 and 5-fold higher than in Ref. 21. Our procedures remain the same, and we speculate that the difference is due to differences in the lots of bumetanide (recently obtained from Sigma; in earlier work, also from P. Feit and from Hoechst); we have not yet tested this hypothesis. Despite these absolute differences, the relative differences from one isoform or species to another are similar within each set of determinations. We also note that the value of K m(Cl) for sNKCC1 in Ref. 6    lower in the untransfected HEK cell compared with cells transfected with NKCC1 or h 1 r 2A 0.7; the bumetanide affinity is similar to that of h 1 r 2A 0.7 and 3-fold higher than that of hNKCC1. As illustrated in Fig. 3, there was no difference between untransfected and mock-transfected HEK-cells, demonstrating that the kinetic properties of the transporter are not affected by the transfection process, by the maintenance of foreign DNA in the cell line, or by expression of the G-418resistance gene product. These data suggest that the endogenous Na-K-Cl cotransporter in HEK cells is a distinct form.
Inhibition of Cotransport Activity by Hg-Inorganic mercury has been shown to inhibit electrolyte secretion in the spiny dogfish rectal gland in a manner consistent with binding of Hg to sNKCC1 (31) . Fig. 5, A and B, shows the effect of Hg on cotransport activity in transfected and untransfected HEK cells. The shark rectal gland cotransporter, sNKCC1, is seen to be 50% inhibited at 25 M Hg. This is consistent with the concentration range in which secretion is inhibited (31), but direct comparison is difficult due to substantial differences in experimental conditions. The mammalian cotransporters, particularly the endogenous HEK cotransporter, are much less sensitive to Hg, presumably reflecting differences in the number or accessibility of cysteine residues in the cotransporter protein.
Regulation of NKCC-We have previously reported that HEK cells possess the requisite machinery to regulate NKCC1 via changes in intracellular [Cl], and we have routinely utilized a preincubation in low [Cl] medium to bring about activation of NKCC1 (6,21,22). Cl loss and cotransporter activation is accelerated from t1 ⁄2 Ӎ 40 min (22) to t1 ⁄2 Ӎ 10 min (21) by decreasing the tonicity of the low [Cl] medium, presumably as a result of increased Cl loss through swelling-activated K-Cl cotransport (3). Fig. 6A illustrates the time courses of activation of hNKCC1, h 1 r 2A 0.7, and untransfected HEK cells in low Cl hypotonic medium. The data clearly demonstrate that like NKCC1, h 1 r 2A 0.7 is activated by a decrease in Cl.
The rate of activation in low Cl is seen to be quite different among the cotransporters (Fig. 6A), with a half time ranging from 2 min for h 1 r 2A 0.7 to 6 min for untransfected HEK cells and 12 min for NKCC1. This result suggests that maximal activation occurs with less decrease in [Cl] i for h 1 r 2A 0.7 (and HEK cells) than for NKCC1. We have confirmed that cellular Cl loss occurs in response to low Cl pre-incubation, 6 but we have not examined the time course of loss. It is possible that Cl loss from HEK cells proceeds at different rates in various cell lines and that this underlies the differences in activation time. There is little reason to accept this alternative explanation since it is very unlikely that untransfected HEK cells would lose Cl 6 times faster than hNKCC1-transfected cells.
In Fig. 6A, it may be noted that the level of cotransporter activity in resting cells (that is with no pre-stimulation) varies from one form to another; this resting level is plotted in the bar graph in Fig. 6B. Importantly, the renal cotransporter NKCC2 (as h 1 r 2A 0.7) has substantial activity under resting conditions, approximately 22% of maximal. On the other hand, hNKCC1 activity is almost undetectable in the resting condition, approximately 2% of the maximal level.
The Na-K-Cl cotransporter is known to be activated by cell shrinkage in a wide variety of cell types (1). The effect of pre-incubation in media of different osmolalities is illustrated in Fig. 6C. Surprisingly, neither untransfected nor NKCC1transfected HEK cells demonstrate the expected response to cell volume. Rather, there is significant decrease in transport activity at elevated tonicity, both in normal ionic conditions and following activation by incubation in low [Cl]. We suggest that this effect is due to the concentrative effect of cell shrinkage that increases cell [Cl] and thereby inactivates the cotransporter, overriding a weak or non-existent volume response of the normal type. This explanation has been previously offered for paradoxical shrinkage-activation of the shark rectal gland cotransporter (32).
The NKCC2 chimera does exhibit a modest volume response, 6 C. Gillen and B. Forbush, unpublished data. increasing in activity by about 50% with a 2-fold increase in tonicity (Fig. 6C, right panel). This demonstrates that HEK cells do have volume-response machinery. It is possible that the appropriate response is seen with h 1 r 2A 0.7 but not with NKCC1 because the counteracting effect of increased [Cl] i is a stronger modulator of NKCC1.
mRNA Encoding NKCC-To determine if the HEK cell cotransporter is closely related to either of the two known NKCC isoforms, we used RT-PCR to amplify a 630-bp sequence from NKCC1 and a 440-bp sequence from NKCC2. As illustrated in Fig. 7, the NKCC1 oligonucleotides (top panel) yielded a PCR product of the expected size for all the cell lines tested, whereas the result with NKCC2 oligonucleotides (bottom panel) was negative for each line. The high amplification yields in HEK cells demonstrate either that hNKCC1 is present in HEK cells or that the endogenous cotransporter is very similar to NKCC1 in the regions of the PCR primers. On the other hand, it is apparent that the HEK cell does not express an appreciable level of hNKCC2 mRNA.
In this experiment we also analyzed a line of mouse 3T3 fibroblasts (E12a) that has been shown to be deficient in Na-K-Cl cotransport activity (28). As seen in Fig. 7, NKCC1 is expressed in the E12a line as well as in the control 3T3 cells (Fig. 7). This result indicates that the mutation in E12a results in a defect that arises beyond the point of transcription, i.e. decreased translation of NKCC mRNA or a defective cotransport protein. DISCUSSION The results presented here provide a comparison of kinetic characteristics of hNKCC1, a rNKCC2A chimera, and the native HEK cell cotransporter. These kinetic parameters are summarized in Table I. As considered above, the characteristics of ion and bumetanide binding as well as the characteristics of regulation are different for each of the cotransporter forms.
This work provides the first detailed description of an isolated form of NKCC2. NKCC2A is found in the apical membrane of the thick ascending limb of the loop of Henle in the mammalian kidney (11,13,15), specifically in segments found in the outer stripe of the outer medulla and in the inner cortex (11). Our results demonstrate that NKCC2A is kinetically different from hNKCC1, particularly in having a lower Rb affinity (K m Ӎ 8 mM). This is a surprisingly high value, and it would appear to ensure that the cotransporter is not saturated with luminal K under physiological conditions. Compared with NKCC1, NKCC2A was also found to have severalfold greater sensitivity to the loop diuretic drug bumetanide, a difference that may be advantageous from the standpoint of drug efficacy.
We have used a chimeric construct in order to obtain sufficient expression to measure NKCC2-mediated flux. In addition to the 5Ј-UTR from NKCC1, the h 1 r 2A 0.7 chimera includes hNKCC1 residues replacing the first 104 residues of the rabbit NKCC2A sequence. Most of the N terminus is very poorly conserved within the CCC family, and where we have tested, we have seen no effect of N-terminal modifications on function.
(a) A c-myc epitope tag appended to the KCC1 N terminus does not affect function (3,4). (b) Interchanging N termini between sNKCC1 and hNKCC1 has no effect on ion affinity differences; rather the dissimilarities are conferred by the large central hydrophobic domains (21).
A significant finding of this report is the activation of h 1 r 2A 0.7 in response to a decrease in [Cl]. This suggests that NKCC2 is modulated by [Cl] i in the renal epithelial cell, providing a way for the apical cotransporter to respond to changes in the rate of Cl exit across the basolateral membrane. In many species vasopressin regulates NaCl and water reabsorption in the medullary thick ascending limb via a cAMP-dependent mechanism. By demonstrating a route for basolateral 3 apical communication, the present result supports a model in which the initial point of cAMP regulation is at the basolateral exit pathways (33), although it does not rule out a model in which the cotransporter is the primary site of regulation (34).
Compared with NKCC1, h 1 r 2A 0.7 displays a higher level of constitutive activity and exhibits a faster and smaller response to changes in intracellular Cl (see Fig. 6). Together, the results suggest that the [Cl] i set point is higher for NKCC2. In addition, the h 1 r 2A 0.7 response to [Cl] i appears to be more easily overridden by a response to a change in cell volume.
Our previous results with sNKCC1/hNKCC1 chimeras demonstrate that the N-and C termini of NKCC are not important in determining ion and bumetanide affinity differences, and it is reasonable to expect that the chimera h 1 r 2A 0.7 is fully representative of NKCC2A with regard to these properties. On the other hand, the N terminus of sNKCC1 has been shown to be involved in regulation of transport by a mechanism that involves phosphorylation of T 184 and T 189 (17,35). The 15-residue region surrounding these phosphoacceptors is well conserved, with a single amino acid change in the h 1 r 2A 0.7 chimera (Q 98(NKCC2) 3 R 183(NKCC1) ). We do not know if the region upstream of the phosphoacceptors plays a role in regulation and, if so, whether the introduction of NKCC1 residues in this region makes the regulatory behavior of h 1 r 2A 0.7 different from that of native NKCC2A.
The HEK cell cotransporter has functional characteristics that are quite different from both NKCC1 and h 1 r 2A 0.7. In particular, the affinity of the endogenous transporter for Rb is 8-and 2.5-fold lower than that of NKCC1 and h 1 r 2A 0.7, respec-tively, and the affinity for Na is Ͼ1.5-fold lower than that of both of the described isoforms. The results of RT-PCR experiments demonstrate that the HEK cell does not express a detectable message for NKCC2 and that NKCC1, or a form homologous to NKCC1, is present (Fig. 7).
We consider three possible explanations for the uniqueness of the HEK cell cotransporter. (a) Assuming that the endogenous HEK cell cotransporter is in fact hNKCC1, why is the kinetic and regulatory behavior different from hNKCC1? One possibility is that the HEK cell cotransporter contains an accessory subunit that is not available in the amounts necessary to accompany overexpressed hNKCC1 in transfected cells. A related idea is that the hNKCC1 protein may be post-translationally modified and that the modification machinery is inadequate for the overexpressed protein. (b) It is possible that the HEK cell cotransporter is a splice variant of hNKCC1, with functional characteristics that are different from the transporter encoded by the cDNA isolated from T-84 cells (6). We have reported three splice variants of NKCC2 that differ in the sequence of the predicted second transmembrane domain. Although the hypothesis has not yet been tested, alternative splicing of this region might be expected to result in transporters with different ion affinities (9). In a search for splice variants of NKCC1, Delpire and co-workers (36) found no evidence of alternative splicing in this region but did report that 16 residues in the C terminus are sometimes removed by splicing. In light of our recent finding that sNKCC1/hNKCC1 ion affinity differences are due only to differences within the central hydrophobic domain (21), it seems unlikely that this alteration would result in the discrepancy between HEK cell affinities and those of hNKCC1. (c) It is possible that the HEK cell cotransporter is an isoform of NKCC that has not yet been identified. At present there is no direct evidence for or against this hypothesis. However, in a general sense, the possibility of additional isoforms is supported by the broad distribution of the Na-K-Cl cotransporter in cells with very different physiological function. In epithelial tissues, the function of the Na-K-Cl cotransporter is transepithelial transport, and regulation of the transporter is presumably optimized for hormonal control and/or apical-basolateral communication. In non-polarized tissues, the Na-K-Cl cotransporter plays a role in regulation of intracellular volume (1) and may also be important in its effect on extracellular [K] (37). Thus, we propose that the HEK cell cotransporter may represent a unique NKCC isoform that operates with low cation affinities.