Residues Contributing to the Ca2+ and K+ Binding Pocket of the NCKX2 Na+/Ca2+-K+ Exchanger*

The Na+/Ca2+-K+ exchanger (NCKX) extrudes Ca2+ from cells utilizing both the inward Na+ gradient and the outward K+ gradient. NCKX is thought to operate by a consecutive mechanism in which a cation binding pocket accommodates both Ca2+ and K+ and alternates between inward and outward facing conformations. Here we developed a simple fluorometric method to analyze changes in K+ and Ca2+ dependences of mutant NCKX2 proteins in which candidate residues within membrane-spanning domains were substituted. The largest shifts in both K+ and Ca2+ dependences compared with wild-type NCKX2 were observed for the charge-conservative substitutions of Glu188 and Asp548, whereas the size-conservative substitutions resulted in nonfunctional proteins. Substitution of several other residues including two proline residues (Pro187 and Pro547), three additional acidic residues (Asp258, Glu265, Glu533), and two hydroxyl-containing residues (Ser185 and Ser545) showed smaller shifts, but shifts in Ca2+ dependence were invariably accompanied by shifts in K+ dependence. We conclude that Glu188 and Asp548 are the central residues of a single cation binding pocket that can accommodate both K+ and Ca2+. Furthermore, a single set of residues lines a transport pathway for both K+ and Ca2+.

The gene families of the SLC8 Na ϩ /Ca 2ϩ exchangers (NCX) 1 and the SLC24 Na ϩ /Ca 2ϩ -K ϩ exchangers (NCKX) comprise proteins that use the inward Na ϩ gradient (NCX) or both the inward Na ϩ gradient and the outward K ϩ gradient (NCKX) to drive Ca 2ϩ extrusion from cells that show dynamic Ca 2ϩ fluxes (1,2). The Na ϩ /Ca 2ϩ -K ϩ exchanger was originally described as a plasma membrane protein in the outer segments of retinal rod photoreceptors (3,4), where it is the only export mechanism present to extrude Ca 2ϩ that enters this organelle via the cGMP-gated channels (5,6). The rod Na ϩ /Ca 2ϩ -K ϩ exchanger (NCKX1) was first cloned from bovine retina (7), and since then another five members of the mammalian NCKX gene family (NCKX2 through NCKX6) have been cloned (2,8). In addition, several NCKX cDNAs have been cloned from lower organisms on the basis of sequence homology within the two sets of transmembrane-spanning segments found in all NCKX sequences (9 -12). The first four mammalian NCKX isoforms and those cloned from lower organisms have been shown to code for K ϩ -dependent Na ϩ /Ca 2ϩ exchangers when expressed in cell lines and transcripts are found in a wide variety of tissues (2). NCKX1 and NCKX2 carry out K ϩ and Ca 2ϩ cotransport and Na ϩ and Ca 2ϩ countertransport at a stoichiometry of 4Na ϩ : 1Ca 2ϩ ϩ 1K ϩ (3,13,14).
This study concerns identification of residues involved in cation transport via the NCKX Na ϩ /Ca 2ϩ -K ϩ exchanger. In situ studies on the properties of the NCKX cation transport sites have only been carried out on the NCKX1 exchanger in retinal rod outer segments (reviewed in Refs. 4,15,16), while several studies have reported on the cation concentration dependence of heterologously expressed NCKX1 (dolphin, chicken), NCKX2 (chicken, human, rat), Caenorhabditis elegans NCKX, and a NCKX1 deletion mutant from which the two large hydrophilic loops (ϳ63% of the total sequence) were removed (9,13,14,17,18). The values obtained in situ for NCKX1 and in heterologous systems for the various NCKX isoforms mentioned above were quite similar. The apparent K m for internal and external Ca 2ϩ were both found to be 1-3 M, the K m for external K ϩ was reported to be 2-10 mM, whereas the K m for external Na ϩ was found to be 35-45 mM with a Hill coefficient of ϳ2. However, in one study a much higher value of Ͼ35 mM was reported for the external K ϩ K m of rat NCKX2 (14). Comparing the sequences of the various NCKX proteins used in the above-cited studies suggests that two distinct and highly conserved stretches of NCKX sequence contribute the residues important to cation binding and transport. These socalled ␣1 and ␣2 repeats are thought to have arisen from an ancient gene duplication event and are the only domains that show significant sequence similarity between members of the SLC24 NCKX gene family and members of the SLC8 NCX gene family (19). Scanning mutagenesis of the ␣1 and ␣2 repeats in human NCKX2 identified 20 residues, which, when substituted individually, resulted in a strong reduction (Ͼ80%) of the maximal transport rate of the mutant NCKX2 protein; this includes eight residues conserved between members of both NCKX and NCX gene families (20). The position of these eight residues is illustrated in Fig. 1B. Individual substitution of the eight corresponding residues in NCX1 greatly lowered the maximal transport rate of the mutant NCX1 proteins precluding detailed kinetic measurements (21,22). The objective of this study was to identify residues within the two ␣ repeats of NCKX2 that control the K ϩ and Ca 2ϩ concentration depen-dences of the NCKX2 protein. We developed a sensitive fluorometric assay for NCKX function in mammalian cell lines permitting measurements of cation dependence of NCKX2 mutant proteins with greatly reduced maximal transport rates. Our results show that substitution of several of the residues highlighted in Fig. 1B resulted in parallel changes in K ϩ and Ca 2ϩ concentration dependences of the mutant NCKX2 proteins expressed in HEK293 cells.

MATERIALS AND METHODS
Mutagenesis and Protein Analysis-Codons for selected residues were mutated in the Myc-tagged short splice variant of human NCKX2 cDNA (AAF25811), and NCKX function of the mutant NCKX2 proteins was analyzed after heterologous expression with the pEIA vector in an insect cell line (High Five cells) as described before (20). Wild-type and mutant NCKX2 proteins expressed in cell lines were detected by Western blotting with the mononclonal Myc antibody (NEB Biolabs, Pickering, Ontario) after separation on SDS-PAGE as described before (20).
Functional Analysis with 45 Ca Uptake-Insect High Five cells were transfected with the various NCKX2 mutants, Na ϩ loaded using the ionophore monensin and washed as described before (9). The final cell pellet was resuspended in 150 mM choline chloride, 80 mM sucrose, 20 mM Hepes, pH 7.4, and 0.05 mM EDTA, and was stored at 25°C. 45 Ca (0.5-1.0 Ci per experiment) (Amersham Biosciences) uptake experiments were carried out in a medium containing 150 mM KCl, 80 mM sucrose, 20 mM Hepes (adjusted to pH 7.4 with arginine), 1 M carbonylcyanide p-trifluoromethoxyphenylhydrazone (FCCP), and 0.036 mM CaCl 2 in addition to 45 Ca. Nonspecific 45 Ca uptake was measured when the uptake medium contained NaCl instead of KCl, which results in complete inhibition of 45 Ca uptake via reverse Na ϩ /Ca 2ϩ -K ϩ exchange. External 45 Ca was removed from 45 Ca taken up into cells by a rapid filtration and washing procedure with the use of borosilicate glass fiber filters; the ice-cold washing medium contained 140 mM KCl, 80 mM sucrose, 20 mM Hepes, pH 7.4, 5 mM MgCl 2 , and 1 mM EGTA. NaCl, KCl, LiCl, and choline chloride were all SigmaUltra grade (Sigma-Aldrich). Further details of the 45 Ca uptake method have been described before (9). Protein content of cell samples was determined with the Bio-Rad protein assay.
Functional Analysis with Fluo-3-Human embryonic kidney (HEK293) cells were subcultured in DMEM (Invitrogen, Burlington, Ontario) supplemented with 10% fetal bovine serum (Invitrogen). Transient expression of wild-type and mutant human NCKX2 cDNAs was carried out using the calcium phosphate precipitation method. Briefly, 1 ml of transfection solution with calcium phosphate-cDNA precipitates was prepared by adding 0.5 ml of DNA solution (10 -30 g of plasmid DNA in 500 mM CaCl 2 solution) to 0.5 ml of phosphate-containing solution (274 mM NaCl, 1.8 mM Na 2 HPO 4 , and 50 mM HEPES, pH 7.07); the mixture was applied to 10-cm plates containing 50% confluent HEK293 cells for 16 -20 h. Next, the medium containing the transfection solution was replaced with fresh DMEM. Two days after transfection, the cells were washed on the plate and collected in suspension. The cell suspension (in 0.5 ml of DMEM) was loaded with 12 M Fluo-3AM for 20 min. Nonhydrolyzed Fluo-3AM was removed by washing the cells twice with a buffer containing 150 mM NaCl, 3 mM KCl, 2 mM CaCl 2 , 6 mM D-glucose, 0.25 mM sulfinpyrazone, and 20 mM HEPES pH 7.4. Sulfinpyrazone is an anion transporter inhibitor that may reduce leakage of Fluo-3 from the cells. Even in the presence of sulfinpyrazone, dye leakage limited the amount of time available to sequential measurements of a series of four experimental conditions (e.g. testing four different KCl concentrations) on different aliquots of the same HEK293 cell suspension. In the second wash step, KCl was omitted from the above solution. About 100,000 Fluo-loaded cells (50 l) were added to a cuvette containing 1.95 ml of 150 mM LiCl, 6 mM D-glucose, 0. 25 (1-5 mM); to obtain the signal for Ca 2ϩ -free Fluo-3, HEK293 cells were permeabilized by the addition of 0.02% saponin in the presence 100 M EDTA. Further details of the Fluo-3 assay as applied to cell suspensions have been published before (23).

Functional Analysis of NCKX2 Mutant Proteins in High Five Cells by 45 Ca Uptake
In a previous study we carried out scanning mutagenesis of the two ␣ repeats of NCKX2 and identified several residues that are important for NCKX transport function and are thought to be located well within the plane of the lipid bilayer: these include four acidic residues, four hydroxyl containing residues, and two proline residues (20). The results obtained previously for nine of these ten residues (20) are summarized in Fig. 1A. For each residue substitution a set of 2-5 new experiments was added. In addition to the mutant NCKX2 proteins reported before, four new substitutions were analyzed that represent conservative Ser to Thr and Thr to Ser changes. Fig.  1B illustrates the location of these residues within our current model of the NCKX2 topology; the eight residues located in hydrophobic segments H2 and H8 (part of the ␣1 and ␣2 repeat, respectively) are conserved between the NCKX and NCX gene families, whereas members of the NCX gene family do not have acidic residues in positions equivalent to the NCKX2 residues Asp 258 and Asp 575 . The four new mutant NCKX2 proteins represent conservative substitutions of the four critical hydroxyl-containing residues found in all NCX and NCKX proteins. Interestingly, in all three cases (Ser 185 , Ser 545 , Ser 552 ) the conservative Ser to Thr substitution resulted in a very strong reduction of mutant NCKX2 transport function compared with wild-type NCKX2, whereas the conservative T544S substitution had much less effect on NCKX2 transport compared with the less conservative T544A and T544C substitutions. When mutant NCKX2 protein samples were analyzed by Western blotting, most showed the 2-band pattern observed for wildtype NCKX2, in which the lower band represents NCKX2 protein from which the signal peptide was cleaved. The S545T mutant NCKX2 protein consistently showed mostly the upper band representing the uncleaved NCKX2 (Fig. 2). This is a highly uncommon phenotype among the ϳ150 single residue NCKX2 mutants we have analyzed in the High Five cell system (also observed for the S185C and S185T mutant NCKX2 proteins in some, but not all experiments). As signal peptide cleavage appears essential for plasma membrane targeting (24), strongly reduced transport observed for the S545T NCKX2 mutant proteins may be caused by the lack of cleaved mutant NCKX2 protein present in the plasma membrane. Surface biotinylation of the S545T mutant NCKX2 protein (e.g. Ref. 24) confirmed that only the small amount of mutant NCKX2 protein represented by the lower band was present in the plasma membrane (data not shown).

K ϩ Concentration Dependence of Mutant NCKX2 Proteins
Determined in High Five Cells by 45 Ca Uptake 45 Ca uptake via reverse Na ϩ /Ca 2ϩ -K ϩ exchange was used as a simple quantitative assay for NCKX2 function as illustrated in Fig. 3. Analysis of shifts in K ϩ and Ca 2ϩ concentration dependences of NCKX2 mutants requires a sensitive and consistent assay for NCKX function in view of our observation that maximal NCKX transport activity was strongly reduced upon substitution of residues hypothesized to be involved in cation binding. Only two of the mutant NCKX2 proteins illustrated in Fig. 1 showed sufficient NCKX2 transport activity to allow consistent measurements of the K ϩ concentration dependence of 45 Ca uptake via reverse Na ϩ /Ca 2ϩ -K ϩ exchange (Fig. 3). Lowering extracellular K ϩ concentration affected the initial rate and the total amount of 45 Ca uptake in comparable fashion, similar as observed before for a variety of different NCKX clones as well as for NCX1 (9,13,17). Much higher K ϩ concentrations were required to activate 45 Ca uptake via reverse Na ϩ /Ca 2ϩ -K ϩ exchange in the case of the D548E mutant (apparent K m ϳ 10 mM), whereas the K ϩ dependence of the P187A mutant (apparent K m ϳ 0.8 mM) appeared unchanged when compared with wild-type NCKX2 (apparent K m ϳ0.7 mM) (apparent K m values were estimated from an analysis of the initial rates of 45 Ca uptake). The ability to resolve low NCKX2 activity in these experiments is limited by the nonspecific 45 Ca uptake observed in high NaCl medium when NCKX2-mediated uptake is completely inhibited. (Note: uptake in NaCl medium was very similar to that observed in cells transfected with empty vector.) In cells expressing wild-type NCKX2, the ratio of specific NCKX2-mediated uptake/nonspecific uptake (i.e. that observed in NaCl medium) was about 10 in this particular experiment, but this ratio had dropped to 2 or less in the case of most of the NCKX2 mutants listed in Fig. 1.

Fluorometric Analysis of NCKX2 Mutant Function in HEK293 Cells
Here we describe a more sensitive assay for NCKX function that measures NCKX2-mediated changes in intracellular free Ca 2ϩ in HEK293 cells transfected with (mutant) NCKX2 cDNAs and loaded with the fluorescent Ca 2ϩ -indicating dye Fluo-3. High Five cells proved to be not suited for this method caused by the fast leakage of Fluo-3 from these cells. The Fluo-3 assay, illustrated in Fig. 4, enabled us to analyze cation dependence for most of the NCKX2 mutant proteins even when maximal transport capacity was severely reduced. HEK293 cells were loaded with Fluo-3 after transfection with either empty vector or vector containing wild-type or mutant human NCKX2 cDNA. About 100,000 of the Fluo-loaded cells were FIG. 1. Mutagenesis of residues in the ␣1 and ␣2 repeats. A, indicated mutants were analyzed for NCKX2 function in High Five cells and compared with wild-type NCKX2 function. The 45 Ca uptake assay was used (e.g. see Fig. 3). Data reported in Ref. 20 were supplemented with 2-5 additional observations and four new mutants were tested (S185T, T544S, S545T, and S552T). Data (average of 45 Ca uptake at the 60, 90, and 120 s time points) are normalized with respect to wild-type NCKX2 function and present average Ϯ S.E. Temperature: 25°C. B, NCKX2 protein contains 11 hydrophobic segments (H1-H11), in addition to the cleavable signal peptide (H0). In our current model of NCKX2 topology (31), 10 of 11 hydrophobic segments form membrane-spanning helices, and the location of the residues subject of this study is indicated within these helices. The 8 residues conserved between NCX and NCKX are represented by black ovals with white lettering (located in H2 and H8). Two additional acidic residues found within membrane helices of NCKX, but not NCX, are represented by black squares with white lettering. Six additional acidic residues found at the periphery of membrane helices are represented by white ovals and black lettering. The shaded parts of the sequence represent the two ␣ repeats, thought to have originated from an ancient gene duplication event (19). placed in a cuvette containing 2 ml of a buffered LiCl solution with a final concentration of 100 M EDTA present. Fluo-3 fluorescence was continuously measured while various additions were made to the cuvette from concentrated solutions under constant stirring (Fig. 4A). First, CaCl 2 was added to a final concentration of 250 M as indicated by the arrow and by the spike in the trace; the spike was caused by lowering the pipette through the light beam and served as a convenient event marker. The CaCl 2 addition caused a small and instantaneous increase in fluorescence caused by saturation of the small amount of Fluo-3 dye that had leaked out of the cells. No time-dependent increase in Fluo-3 fluorescence was observed indicating the lack of any Ca 2ϩ influx in either control or NCKX2-transfected cells under this condition. The second arrow indicates the addition of KCl to a final concentration of 20 mM. This KCl addition caused no change in Fluo-3 fluorescence in control cells, but led to a rapid and large rise in Fluo-3 fluorescence in NCKX2-transfected cells. The fast rise in fluorescence indicates a rapid K ϩ -dependent Ca 2ϩ influx representing reverse Na ϩ /Ca 2ϩ -K ϩ exchange. The third arrow indicates addition of saponin (final concentration 0.02%) to permeabilize the plasma membrane. Saponin addition caused the release of all Fluo-3 present in the cells, thereby saturating Fluo-3 fluorescence by the high (ϳ150 M) Ca 2ϩ concentration in the medium. Minimum Fluo-3 fluorescence (obtained by addition of saponin in the presence 0.1 mM EDTA) was only slightly lower than fluorescence representing resting Ca 2ϩ levels observed in HEK293 cells in the LiCl-EDTA medium at the start of the experiment (i.e. before activating Ca 2ϩ influx via the exchanger). The K ϩ -and Ca 2ϩ -dependent increase in Fluo-3 fluorescence observed in cells transfected with wild-type NCKX2 ranged from 30 -50% of the saponin-induced increase in Fluo-3 fluorescence in different transfection experiments. When NCKX2 cDNA was co-transfected with cDNA encoding green fluorescent protein (GFP) a similar percentage of the transfected cells showed GFP staining indicating the percentage of successfully transfected cells (data not shown). Combined these results suggest that reverse Na ϩ /Ca 2ϩ -K ϩ exchange via wild-type NCKX2 leads to a rapid Ca 2ϩ influx that competes effectively with Ca 2ϩ clearance mechanisms endogenous to HEK cells and results in a significantly elevated free Ca 2ϩ concentration and increased Fluo-3 fluorescence, but only in those cells transfected with NCKX2. Since the fluorescence measurements represent a mixed population of transfected and untransfected cells, we have not attempted to convert fluorescence into free cytosolic [Ca 2ϩ ] for those cells successfully transfected with NCKX2.

K ϩ and Ca 2ϩ Concentration Dependences of NCKX2 Activity Measured with the Fluo-3 Assay
Using the above Fluo-3 assay, we measured the K ϩ -induced rise in Fluo-3 fluorescence as a function of the external K ϩ concentration (top panels of Fig. 4B). Similar to the pattern observed in the 45 Ca uptake assay, lowering the external K ϩ concentration below saturating values lowered the initial rate as well as the amplitude of the changes in Fluo-3 fluorescence observed in cells transfected with wild-type NCKX2. The smaller amplitudes of the Fluo-3 signals indicate a lower free [Ca 2ϩ ] in the HEK293 cells caused by a reduced rate of Ca 2ϩ influx via NCKX2, while Ca 2ϩ clearance via the plasma membrane and endoplasmic reticulum Ca 2ϩ pumps remained constant. Thus, the plateau fluorescence reflects the equilibrium free Ca 2ϩ concentration that arises from the balance of Ca 2ϩ influx via NCKX2 and Ca 2ϩ efflux via the Ca 2ϩ pump. The K ϩ dependence of the plateau fluorescence levels was very similar FIG. 2. Western blot of serine and threonine NCKX2 mutants. The indicated NCKX2 mutants were expressed in High Five cells as described before (20). Protein was extracted in a buffer containing 140 mM NaCl, 25 mM Tris (pH 7.5), 1% Triton X-100, 0.5% sodium deoxycholate, 0.1 mM EDTA, and a protease inhibitor tablet (Roche Applied Science). The extract was spun down (5 min at 20,000 ϫ g), and the supernatant was used. Samples (20 g protein/lane) were run on SDS-PAGE and analyzed by Western blotting with the monoclonal anti-Myc antibody. control represents cells transfected with empty vector. In most blots wild-type NCKX2 displays a doublet, the lower band representing NCKX2 from which the signal peptide is cleaved (24). In some cases, an additional ghost band is observed at a slightly lower molecular size (e.g. the S545C mutant NCKX2), probably representing cleaved NCKX2 that was not completely unfolded in the sample buffer. NCKX2containing samples were only heated for 10 min at 37°C to avoid aggregation. to the K ϩ dependence of the initial rate of the changes in Fluo-3 fluorescence. The observed pattern suggests that both the initial rate and amplitude of the Fluo-3 signal can be used as a semiquantitative indicator of NCKX2 transport function. To ensure that changes in cation dependence were not caused by lower maximal transport activities per se, we compared wildtype NCKX2 with the E651D NCKX2 mutant, previously reported to have about 13% of wild-type activity in the High Five cell system (20). Glu 651 is located at the C terminus of NCKX2 (Fig. 1B) and is unlikely to be directly involved in cation trans- port. Consistent with the results obtained with the High Five cell system (20), the maximal amplitude of the Fluo-3 signal observed for the E651D NCKX2 mutant was much smaller than that observed for wild-type NCKX2, indicative of a much reduced maximal transport rate compared with wild-type NCKX2. However, the K ϩ concentration dependence of the Fluo-3 signals observed in cells expressing the E651D mutant NCKX2 protein appeared very similar to that observed for wild-type NCKX2 (Fig. 4B), demonstrating that reduced maximal transport rate did not itself result in a change in K ϩ concentration dependence. The patterns illustrated in Fig. 4B were highly reproducible, and for a number of experiments the initial rate of the K ϩ -induced fluorescence changes (change observed after 5 s) and the amplitude of the change in fluorescence change (fluorescence 40 s after addition of KCl) were averaged and normalized with respect to those observed for the highest K ϩ concentration of 50 mM used. Both analyses yielded very similar K ϩ concentration dependence for wild-type NCKX2 and the E651D NCKX2 mutant with a half-maximal rate or amplitude observed at ϳ1 mM (bottom panels of Fig. 4B). This value is very similar to that observed in Fig. 3 using the 45 Ca uptake method. In the same set of experiments, the Ca 2ϩ concentration dependence was measured for wild-type NCKX2 and the E651D NCKX2 mutant at a fixed KCl concentration of 20 mM; the results were analyzed in the same way as shown for the K ϩ dependence (Fig. 4C). As observed for the K ϩ dependence, very similar Ca 2ϩ dependence was observed for wild-type NCKX2 and the E651D NCKX2 mutant, both when analyzing the initial rate or the amplitude of the fluorescence signals. Half-maximal rate or amplitude was observed at ϳ1 M free [Ca 2ϩ ]. Fluo-3 fluorescence does not represent a linear [Ca 2ϩ ] scale, and some compression should occur at higher fluorescence levels because of saturation of the dye. This could result in an underestimate of apparent K m values. The compression effect should be more obvious for the maximal amplitude of the fluorescence signals and for NCKX2 (mutant) proteins that show high transport rates (e.g. wild-type NCKX2). Two observations presented here suggest that compression is not a major problem. 1) The concentration dependence obtained from the initial rates and from the amplitudes of the fluorescence signals, respectively, was very similar. 2) The K ϩ and Ca 2ϩ dependences were very similar for the two different NCKX2 proteins illustrated in Fig. 4, although maximal transport activities differed greatly. In previous studies we reported external K m values of ϳ2 M for Ca 2ϩ and ϳ2 mM for K ϩ for the wild-type NCKX2 (13,17,18).
Four of the NCKX2 mutants listed in Fig. 1A showed no detectable function when analyzed in the HEK293 cell system with Fluo-3 (E188Q, S545C, S545T, D548N), and this was not caused by reduced mutant protein expression or altered plasma membrane targeting as judged by signal peptide cleavage (not shown). The E188Q, S545C, S545T, and D548N mutant NCKX2 proteins appeared to have small, but measurable activity in the High Five cell system (Fig. 1A). Two possible explanations for this discrepancy are as follows. 1) Several other NCXK clones we have examined show strong function when expressed in High Five cells, but show no (e.g. Drosophila NCKX-X) or no consistent (e.g. chicken rod NCKX1) function when expressed in HEK293 cells (12,24). 2) High Five cells used in our experiments have high internal Na ϩ levels after Na ϩ loading with the ionophore monensin.

Shifts in External K ϩ and Ca 2ϩ Dependences of Reverse
Na ϩ /Ca 2ϩ -K ϩ Exchange in NCKX2 Mutants Acidic Residues within Membrane Helices: Glu 188 , Asp 258 , and Asp 548 -First we focused on the two acidic residues con-served among members of the NCX and NCKX gene families: Glu 188 and Asp 548 are thought to be located well within the plane of the lipid bilayer in strategic positions in helices H2 and H8, respectively (Fig. 1B). The external Ca 2ϩ and K ϩ concentration dependences of mutant NCKX2 proteins carrying the charge-conservative substitutions E188D or D548E, respectively, were compared with those of wild-type NCKX2 as illustrated in Fig. 5. When the initial rates of the fluorescence signals were analyzed, the E188D and D548E mutant NCKX2 proteins showed considerable shifts to lower affinities for both K ϩ and Ca 2ϩ dependences when compared with wild-type NCKX2 (Fig. 5), and these shifts were the largest observed for the various NCKX2 mutants examined in this study. Very similar results were obtained when the amplitudes of the fluorescence signals were analyzed (not shown). The shift in apparent Ca 2ϩ K m was from a value of ϳ1 M for wild-type NCKX2 to a value of close to 100 M for both the E188D and D548E mutant NCKX2. The shift in apparent K ϩ K m was from a value of ϳ1 mM for wild-type NCKX2 to a value of Ͼ10 mM for both the E188D and D548E mutant NCKX2. No NCKX activity was observed in HEK293 cells transfected with either the E188Q or the D548N NCKX2 mutants in which the charge was removed (not shown), consistent with their importance as key residues in Ca 2ϩ binding.
Asp 258 and Asp 575 are the only other acidic residues that are thought to be located deep within membrane helices of NCKX2 and important for its function (Fig. 1A). Mutagenesis of the Asp 575 residue results in a complex phenotype that has been reported elsewhere (32). Both D258N and D258E mutant NCKX2 proteins showed K ϩ -dependent Na ϩ /Ca 2ϩ exchange function when expressed in HEK293 cells. Similar to the results observed in the High Five cell system (Fig. 1A), the size-conservative D258N showed higher maximal activity than the charge-conservative D258E substitution as judged by the amplitudes of the fluorescence signals observed (data not shown, see also Fig. 9). When the initial rates of the fluorescence signals were analyzed, the D258N mutant showed K ϩ and Ca 2ϩ concentration dependences that were not or only slightly shifted compared with those of wild-type NCKX2, whereas the D258E mutants consistently showed shifts in both K ϩ and Ca 2ϩ dependences although the shifts observed were considerably smaller than those observed in Fig. 5 for the E188D and D548E mutants (apparent K m values of ϳ10 M for Ca 2ϩ and ϳ3 mM for K ϩ ) (Fig. 6, two top panels). Very similar results were obtained when the amplitudes of the fluorescence signals were analyzed (not shown).
Proline Residues Preceding Glu 188 and Asp 548 -The two ␣ repeats contain two highly conserved proline residues, Pro 187 and Pro 547 , that precede Glu 188 and Asp 548 , respectively. The bottom panels of Fig. 6 illustrate the K ϩ and Ca 2ϩ concentration dependences of mutant NCKX2 proteins in which Pro 187 or Pro 547 is replaced by either alanine or cysteine. When the initial rates of the fluorescence signals were analyzed, the P187A and P547C NCKX2 mutants showed a small shift in both K ϩ and Ca 2ϩ dependences, whereas the P187C and P547A mutants showed larger shifts to lower apparent affinities for both K ϩ and Ca 2ϩ . Very similar results were obtained when the amplitudes of the fluorescence signals were analyzed (not shown).
Acidic Residues Located in Short Connecting Loops-The two sets of transmembrane regions of NCKX2 contain a number of acidic residues, thought to be located at the membrane/water interface or in the short connecting loops rather than well within the membrane (20). We examined K ϩ and Ca 2ϩ dependences for substitutions of most of these acidic residues. For example, Glu 265 is located in the very short (3 residues) H4-H5 linker (Fig. 1) and is conserved in all NCKX sequences. The E265D and in particular the E265Q mutant NCKX2 proteins showed small shifts in both K ϩ and Ca 2ϩ dependences when compared with wild-type NCKX2 (Fig. 7). Likewise, Glu 651 is conserved in all NCKX sequences, but the E651D mutation showed wild-type K ϩ and Ca 2ϩ dependences even though the maximal transport rate was strongly reduced (Fig. 4). Four additional acidic residues were examined as they are located near membrane helices H2 and H8, which contain the two critical acidic residues Glu 188 and Asp 548 discussed above (Figs. 1 and 5). Asp 172 and Asp 173 are thought to be located at the cytoplasmic interface of helix H2, while Glu 532 and Glu 533 are thought to be located at the extracellular interface of helix H8. These could be strategic positions controlling access to the cation binding site(s) located within the membrane, and, hence, the effect of the charge-eliminating but size-conservative substitutions for these four acidic residues were examined. We previously showed that these size-conservative substitutions had only minor effects on maximal transport activity (20). Fig.  7 shows that three of these four mutant NCKX2 proteins showed wild-type cation dependence, while only the E533Q NCKX2 mutant protein showed significant shifts in both Ca 2ϩ (apparent K m ϳ10 M) and K ϩ (apparent K m ϳ10 mM) dependences when compared with wild-type NCKX2. Fig. 7 shows an analysis of the initial rates of the fluorescence signals. Very similar results were obtained when the amplitudes of the fluorescence signals were analyzed (not shown).
Thus, 7 of 13 NCKX2 mutants, illustrated in Figs. 5-7, show changes in Ca 2ϩ concentration dependence, which were always accompanied by changes in K ϩ concentration dependence. The other six NCKX2 mutants showed no clear shifts in either K ϩ or Ca 2ϩ dependence. For the two mutants (D548E and P187A) that could be analyzed with the 45 Ca uptake assay (Fig. 3), the results were consistent with those obtained with the Fluo-3 assay.
Serine and Threonine Residues-Membrane-spanning helices H2 and H8 contain four hydroxyl-containing residues, substitutions of which had a strong effect on NCKX2 function (Fig.  1). The S185A, S185T, S552C, S552T, T544A, T544S, and S545A (the latter not in all experiments) mutant NCKX2 proteins all gave measurable NCKX2 function in HEK293 cells, whereas the S545T and S545C mutant NCKX2 proteins showed no NCKX2 function when analyzed with the Fluo-3 assay. Compared with the other amino acid substitutions studied here, the substitution of the serine and threonine residues resulted in mutant NCKX2 proteins that showed more variable cation dependence as evidenced by the often larger error bars. Shifts in both K ϩ and Ca 2ϩ concentration dependences were observed for the S545A and S185T NCKX2 mutant (Fig. 8, top  panel), although it should be pointed out that the S545A mutant showed very low transport activity as judged by the very small amplitudes of the Fluo-3 signals observed, and no activity was observed in some experiments. The shift in K ϩ concentration dependence observed for the S185T NCKX2 mutant was small but a significant difference in the normalized rate was observed at a K ϩ concentration of 1 mM (p Ͻ 0.005 in the Student's t test). More anomalous patterns were observed for the Thr 544 and Ser 552 mutants. The two Thr 544 mutants did not show a clear correlation between shifts in K ϩ dependence and shifts in Ca 2ϩ dependence (Fig. 9, middle panel); the two Ser 552 mutants showed no shift in K ϩ dependence, but consistently showed an anomalous Ca 2ϩ dependence in which the higher Ca 2ϩ concentrations of 100 and 500 M appeared to be some- what inhibitory (Fig. 8, bottom panel). This was not observed for any of the other NCKX2 mutants reported here.

Summary of the Analysis of NCKX2 Mutant Proteins with the Fluo-3 Assay
The maximal amplitude of the fluorescence signals observed for the various NCKX2 mutant proteins in different transfection experiments was normalized with respect to that observed for wild-type NCKX2 in each experiment as summarized in Fig.  9. The results are broadly consistent with those described before using an assay based on 45 Ca uptake in the High Five cell system (20), although some differences should be noted. For example, the E265Q, D258E, and D258N mutant NCKX2 proteins appeared to have higher activity when expressed in HEK293 cells compared with expression in High Five cells. It also should be noted that NCKX2 mutant protein expression levels in HEK293 cells were not quite as uniform as observed in the High Five cell system. The High Five cell system would appear to be a better system to evaluate changes in maximal transport velocity for the various NCKX2 mutants, while the HEK293 cell system allows the evaluation of cation dependence for each individual mutant.

DISCUSSION
This is the first study to identify residues of the NCKX Na ϩ /Ca 2ϩ -K ϩ exchanger that contribute to the Ca 2ϩ and K ϩ binding pocket as assessed by changes in Ca 2ϩ and K ϩ concentration dependences of transport via NCKX mutant proteins in which selected residues were replaced. A simple and sensitive fluorometric assay using the fluorescent Ca 2ϩ -indicating dye Fluo-3 was developed to measure the Ca 2ϩ or K ϩ concentration dependence of reverse Na ϩ /Ca 2ϩ -K ϩ exchange observed in HEK293 cells after transient transfection with cDNA encoding wild-type human NCKX2 or different NCKX2 mutants. The fluorescence signals had sufficient resolution to permit analysis of cation dependence of NCKX mutant proteins, even for those with greatly reduced maximal transport rates. Twentysix different NCKX2 mutants proteins were analyzed as summarized in Fig. 9: four showed no measurable function, nine showed no clear changes in Ca 2ϩ or K ϩ dependence, while thirteen of the NCKX2 mutant proteins showed clear changes in Ca 2ϩ and/or K ϩ dependence. There was no correlation between a change in maximal transport rate (as judged by the maximal amplitude of the fluorescence signals) and a change in cation dependence for the various NCKX2 mutant proteins, showing that a reduction in maximal transport rate is not per se accompanied by a change in cation affinities.
Two groups of residues in the human NCKX2 protein were targeted. First, eight residues were selected that are conserved between all members of the mammalian NCX and NCKX gene families; substitution of these eight residues was previously found to greatly reduce maximal transport activity in both NCX1 (21,22) and NCKX2 (20) precluding a detailed kinetic analysis. In the human NCKX2 clone used here, these residues are located in membrane helices H2 (Ser 185 , Pro 187 , Glu 188 ) and H8 (Thr 544 , Ser 545 , Pro 547 , Asp 548 , Ser 552 ) as illustrated in Fig.  1B. Second, as Ca 2ϩ binding sites in proteins invariably contain negatively charged residues, an additional seven acidic residues were selected, based on their location in the two sets of membrane-spanning segments (Fig. 1B) as well as on the high degree of conservation of these residues in different mammalian NCKX isoforms and in NCKX from lower organisms. Here, we propose different roles for these residues based on their location and based on the effect of substitution on transport properties: (1) three acidic residues (Glu 188 , Asp 258 , Asp 548 ) within the horizontal mid-plane of the membrane; (2) acidic residues at the interface between membrane helices and the extracellular medium; (3) hydroxyl-containing residues that line the hydrophilic surface of membrane helices H2 and H8, and which are located one helical turn away from residues E188 and D548, respectively. Two novel findings from this study are: (1) substitutions in the NCKX2 protein that caused changes in Ca 2ϩ concentration dependence invariably caused changes in K ϩ concentration dependence; if the shift in Ca 2ϩ concentration dependence was large, the shift in K ϩ dependence was large as well. As the residues involved are thought to be located at quite different positions relative to the membrane surface, this suggests strongly that Ca 2ϩ and K ϩ share a common transport pathway; (2) Glu 188 and Asp 548 are critical residues lining a common Ca 2ϩ and K ϩ binding pocket.
Glu 188 and Asp 548 Are Key Residues in the NCKX Cation Binding Pocket-Our results suggest strongly that Glu 188 , Asp 548 , and perhaps Asp 258 are pivotal residues lining a single K ϩ and Ca 2ϩ binding pocket: 1) Eliminating the charge on either Glu 188 or Asp 548 by size-conservative substitutions resulted in nonfunctional NCKX2 mutant proteins; Glu 188 and Asp 548 are the only two acidic residues for which this is true among the twenty found within the transmembrane spanning 2) The charge-conservative E188D and D548E substitutions resulted in the largest shifts in the dependences for both K ϩ and Ca 2ϩ observed here for the various NCKX2 mutant proteins (Fig. 5). 3) The pH and Mg 2ϩ dependences of Ca 2ϩ transport via NCKX1 have features similar to Ca 2ϩ binding to EGTA, suggesting the involvement of multiple carboxylate moieties (25). 4) Proline residues are often found in the middle of membrane helices where they produce a kink in the helix and a region of increased flexibility (26). Substitutions of the highly conserved proline residues Pro 187 and Pro 547 that precede Glu 188 and Asp 548 , respectively, resulted in marked shifts in both K ϩ and Ca 2ϩ dependences (Fig. 6). 5) Glu 188 and Asp 548 were found to be in close proximity as judged by sitedirected disulfide mapping. 2 6) The D258E mutation showed clear shifts to lower affinities for both K ϩ and Ca 2ϩ , suggesting it could be a contributing residue of the K ϩ and Ca 2ϩ binding pocket (Fig. 6). However, the importance of Asp 258 is clearly less established than that of Glu 188 and Asp 548 .
Iwamoto et al. (22) reported that substitution of several aspartate residues in the transmembrane spanning segments of NCX1 affected the affinity toward Ca 2ϩ and suggested these residues are important in Ca 2ϩ liganding, whereas substitution of Glu 113 and Asp 814 , the conserved equivalent of NCKX2 residues Glu 188 and Asp 548 in NCX1, resulted in mutant NCX1 proteins with insufficient activity to assess Ca 2ϩ affinity. Distinct from our results with the NCKX2 residues Glu 188 and Asp 548 , charge-eliminating substitutions (e.g. to cysteine) of the above aspartate residues in NCX1 did not result in nonfunctional NCX1 proteins. Therefore, we believe that Glu 188 and Asp 548 are the central Ca 2ϩ liganding residues in NCKX2, and we believe the equivalent residues Glu 113 and Asp 814 play this role in NCX1. It should be pointed out that the external K m of wild-type NCX1 is about 200 M Ca 2ϩ (22), while the external K m of the various NCKX isoforms is 1-3 M Ca 2ϩ (13,14,18). A peculiar aspect then is that large shifts in K ϩ concentration dependence are observed in NCKX2 for the charge-conservative substitutions of Glu 188 and Asp 548 , residues conserved in NCX1 that does not transport K ϩ . Earlier experiments showed that alkali cations such as Li ϩ and K ϩ could stimulate Ca 2ϩ : Ca 2ϩ self exchange mediated by NCX1 in cardiac sarcolemma vesicles and could inhibit Na ϩ :Ca 2ϩ exchange via NCKX1 suggesting that alkali cations can bind near the cation binding pocket and affect transport without being transported themselves (reviewed in Refs. 4 and 27). Perhaps other residues in NCKX, not found in NCX1, enable K ϩ not only to bind to the cation binding pocket, but also to be transported. We have identified Asp 575 as one such residue (32). Clearly, other residues influence the apparent K m of Ca 2ϩ transport, perhaps by increasing the local Ca 2ϩ and K ϩ concentration through electrostatic effects. For example, the aspartate residues identified by Iwamoto et al. in NCX1 (see above) may function this way, and in NCKX2 residues Glu 265 and Glu 533 could play this role.
A Single Set of Residues of NCKX2 Lines a Common Transport Pathway for K ϩ and Ca 2ϩ -Previous kinetic studies on in situ NCKX1 (28 -30) and wild-type NCKX1 or NCKX2 expressed in cell lines (13,18) suggest that Ca 2ϩ and K ϩ each bind to a separate single occupancy binding site. The NCKX Na ϩ /Ca 2ϩ -K ϩ exchanger found in the outer segments of rod photoreceptors has been shown to carry out both countertransport of 4 Na ϩ :(1 Ca 2ϩ ϩ 1 K ϩ ) and self-exchange of (Ca 2ϩ ϩ K ϩ ):(Ca 2ϩ ϩ K ϩ ) dependent on the type of cations present on both sides of the outer segment plasma membrane (4). This is strongly suggestive of a consecutive model of transport in which the NCKX protein must simultaneously bind both 1 Ca 2ϩ and 1 K ϩ to a single (set of) binding site(s) that can alternate between an inward facing and an outward facing conformation. This raises the question whether Ca 2ϩ and K ϩ use the same transport pathway or separate pathways. In this study, shifts in cation dependence were found not only for residues thought to be lining a binding pocket in the center of membrane helices (e.g. Glu 188 , Asp 258 , Asp 548 , see above), but also for 2 (of a total of 16) residues thought to be in a more peripheral location at the membrane/external solution interface (Glu 265 and Glu 533 , Figs. 1 and 7). Furthermore, substitution of hydroxyl residues Thr 544 and Ser 545 along helix H8 could affect cation dependence as well (Fig. 8). As the boundaries of helix H8 appear well defined by a pair of acidic residues at the extracellular interface (Glu 532 and Glu 533 ) and by a pair of basic residues (Arg 557 and Lys 558 ) at the cytosolic interface, it appears rather unlikely that H8 unfolds into a binding pocket in which all the above residues are in close proximity. We suggest that residues lining a common pathway for Ca 2ϩ and K ϩ to traverse the membrane can contribute to the apparent K m values. In a previous study we suggested a role for helices H2 and H8 in cation transport as cysteine residues inserted near the extracellular surface of these two helices resulted in strong inhibition of NCKX transport by thiol modification with the positively charged MTSET reagent, in particular for G536C and L540C in H8 (31). This suggests that these inserted cysteine residues line an aqueous pathway that provides access to the cation binding pocket located further inside the membrane. Fig. 10 shows a helical wheel representation of helix H8 that contains all the residues modification of which could affect cation dependence, starting with Glu 533 at the extracellular interface, followed by Gly 536 , Leu 540 , the Thr 544 and Ser 545 pair until one reaches Asp 548 . Our results strongly suggest a common transport pathway for Ca 2ϩ and K ϩ as changes in residues that caused a shift in Ca 2ϩ concentration dependence with respect to wild-type NCKX2 invariably caused a shift in K ϩ concentration dependence as well. Thus, large shifts in both K ϩ and Ca 2ϩ dependences were observed for the E188D and D548E mutants (Fig. 5), while smaller shifts were observed for some of the Pro 187 , Pro 547 , Glu 265 , Glu 533 , and Asp 258 mutant NCKX2 proteins (Figs. 6 and 7). For example, the P187C mutation resulted in a marked shift in K ϩ and Ca 2ϩ concentration dependences, while this dependence were shifted much less in the P187A mutant (Fig. 6). Conversely, with the Pro 547 residue FIG. 10. Helical wheel representation of helix H8. A helical wheel representation of the first 18 residues of helix H8 was obtained (cti.itc.virginia.edu/ϳcmg/Demo/wheel/wheelApp.html) using a pitch of 3.6 residues per turn (an angle of 100 degrees between two residues). The helix starts at residue Glu 533 , proposed to be located at the extracellular membrane interface and ends at Ile 550 . The size of the lettering decreases as the helix descends from the extracellular surface further into the plane of the phospholipid bilayer. the P547A mutant resulted in a marked shift in both K ϩ and Ca 2ϩ concentration dependences, while those dependences were shifted much less in the P547C mutant.
The Importance of the Hydroxyl-containing Residues in Helices H2 and H8 -Helices H2 and H8 contain three serine residues (Ser 185 , Ser 545 , Ser 552 ) and one threonine residue (Thr 544 ) that are important for efficient NCKX2 transport (Fig.  1), probably by lining a hydrophilic surface of the helices that extends from the cation binding pocket in the middle of the membrane to charged residues at the membrane/aqueous environment interface (e.g. Asp 173 and Glu 533 , see above). Three additional observations underline the importance of the three serine residues, in particular. 1) These residues may also provide coordination sites for Ca 2ϩ and K ϩ binding, as several replacements of the hydroxyl-containing residues resulted in altered cation dependence (e.g. S185T, S545A, and T544S, Fig.  8) although the results showed greater variance when compared with the other residues tested. 2) The serine residues Ser 185 , Ser 545 , and Ser 552 share with the acidic residues Asp 575 and Asp 258 (20) the property that conservative replacement with a residue that merely extends the side chain by a single methylene group was poorly tolerated (i.e. Ser to Thr and Asp to Glu, respectively) ( Fig. 1). This suggests severe steric constraints at these residue positions. The latter appears particularly true for the S545T substitution (but not the S545A and S545C substitutions in which the size of the side chain is not increased), which displays trafficking defects, a rather uncommon phenotype for the many NCKX2 mutants generated in our laboratory (Fig. 2). 3) The serine residues are located in an area of increased flexibility as site-directed disulfide mapping showed that these serine residues are in close proximity to the key Ca 2ϩ liganding residues Glu 188 and Asp 548 , and to each other. 2 In addition to contributing to the cation binding site, the placement of the serine and threonine residues one helical turn away from the plane defined by residues Glu 188 and Asp 548 suggests they may control access to Glu 188 and Asp 548 in the inward and outward facing conformations of NCKX. In the outward facing conformation H2 and H8 might be tilted in such a way that Glu 188 , Asp 548 , Ser 545 /Thr 544 and perhaps Ser 192, 2 contribute to coordinating Ca 2ϩ binding, while Ser 185 and Ser 552 are tilted toward each other and impede cation passage. The subsequent conformational change that results in the inward facing conformation of NCKX changes the tilt in H2 and H8 in such a way that Ser 185 and Ser 552 move apart and now contribute with Glu 188 and Asp 548 to coordinating Ca 2ϩ and allow passage to the intracellular space. This conformational change could cause a tilt of Ser 545 /Thr 544 toward Ser 192 and block cations from exiting to the extracellular space. We propose that pairs of hydroxyl-containing residues on helices H2 and H8 (and possibly additional ones as well) function as gates that control the direction of transport and ensure the obligatory coupling of cation fluxes observed for NCKX. Future work will be directed toward addressing this issue.