Chimeric Analysis of Na+/Ca2+ Exchangers NCX1 and NCX3 Reveals Structural Domains Important for Differential Sensitivity to External Ni2+ or Li+ *

Externally applied Ni2+, which apparently competes with Ca2+ in all three isoforms of Na+/Ca2+ exchanger, inhibits exchange activity of NCX1 or NCX2 with a 10-fold higher affinity than that of NCX3, whereas stimulation of exchange by external Li+ is significantly greater in NCX2 and NCX3 than in NCX1 (Iwamoto, T., and Shigekawa, M. (1998) Am. J. Physiol. 275, C423–C430). Here we identified structural domains in the exchanger that confer differential sensitivity to Ni2+ or Li+ by measuring intracellular Na+-dependent45Ca2+ uptake in CCL39 cells stably expressing NCX1/NCX3 chimeras or mutants. We found that two segments in the exchanger corresponding mostly to the internal α-1 and α-2 repeats are individually responsible for the alteration of Ni2+sensitivity, both together accounting for ∼80% of the difference between NCX1 and NCX3. In contrast, the segment corresponding to the α-2 repeat fully accounts for the differential Li+sensitivity between the isoforms. The Ni2+ sensitivity was mimicked, respectively, by simultaneous substitution of two amino acids in the α-1 repeat (N125G/T127I in NCX1 and G159N/I161T in NCX3) and substitution of one amino acid in the α-2 repeat (V820A in NCX1 and A809V in NCX3). On the other hand, the Li+ sensitivity was mimicked by double substitution mutation in the α-2 repeat (V820A/Q826V in NCX1 and A809V/V815Q in NCX3). Single substitution mutations at Asn125 and Val820 of NCX1 caused significant alterations in the interactions of the exchanger with Ca2+ and Ni2+, and Ni2+ and Li+, respectively, although the extent of alteration varied depending on the nature of side chains of substituted residues. Since the above four important residues are mostly in the putative loops of the α repeats, these regions might form an ion interaction domain in the exchanger.

Externally applied Ni 2؉ , which apparently competes with Ca 2؉ in all three isoforms of Na ؉ /Ca 2؉ exchanger, inhibits exchange activity of NCX1 or NCX2 with a 10fold higher affinity than that of NCX3, whereas stimulation of exchange by external Li ؉ is significantly greater in NCX2 and NCX3 than in NCX1 (Iwamoto, T., and Shigekawa, M. (1998) Am. J. Physiol. 275, C423-C430).
Here we identified structural domains in the exchanger that confer differential sensitivity to Ni 2؉ or Li ؉ by measuring intracellular Na ؉ -dependent 45 Ca 2؉ uptake in CCL39 cells stably expressing NCX1/NCX3 chimeras or mutants. We found that two segments in the exchanger corresponding mostly to the internal ␣-1 and ␣-2 repeats are individually responsible for the alteration of Ni 2؉ sensitivity, both together accounting for ϳ80% of the difference between NCX1 and NCX3. In contrast, the segment corresponding to the ␣-2 repeat fully accounts for the differential Li ؉ sensitivity between the isoforms. The Ni 2؉ sensitivity was mimicked, respectively, by simultaneous substitution of two amino acids in the ␣-1 repeat (N125G/T127I in NCX1 and G159N/I161T in NCX3) and substitution of one amino acid in the ␣-2 repeat (V820A in NCX1 and A809V in NCX3). On the other hand, the Li ؉ sensitivity was mimicked by double substitution mutation in the ␣-2 repeat (V820A/Q826V in NCX1 and A809V/V815Q in NCX3). Single substitution mutations at Asn 125 and Val 820 of NCX1 caused significant alterations in the interactions of the exchanger with Ca 2؉ and Ni 2؉ , and Ni 2؉ and Li ؉ , respectively, although the extent of alteration varied depending on the nature of side chains of substituted residues. Since the above four important residues are mostly in the putative loops of the ␣ repeats, these regions might form an ion interaction domain in the exchanger.
The Na ϩ /Ca 2ϩ exchanger is an electrogenic transporter that catalyzes exchange of 3 Na ϩ for 1 Ca 2ϩ across the plasma membrane of many cell types. Previous studies indicate that it plays a primary role in the extrusion of cytosolic Ca 2ϩ from cardiomyocytes, although its contribution in the Ca 2ϩ handling in other cell types still remains to be precisely defined (1). The mammalian Na ϩ /Ca 2ϩ exchanger forms a multigene family comprising three isoforms, NCX1, NCX2, and NCX3, which share ϳ70% identity in the overall amino acid sequences (2)(3)(4). On the basis of the hydropathy analysis, the mature Na ϩ /Ca 2ϩ exchanger proteins were modeled to consist of 11 TMs 1 and a large central hydrophilic loop between TM5 and TM6, with the N terminus localized on the extracellular side and the C terminus and the large central loop being on the intracellular side of the membrane (4,5). Recent studies on the topology of the NCX1 polypeptide have produced data that are consistent with the N-terminal half of the 11 TM model (6 -10), but these data do not support the C-terminal half of the model (9,10), indicating that the model requires revision.
Na ϩ /Ca 2ϩ exchange occurs almost normally in a mutant exchanger deleted of the large central loop (11)(12)(13), indicating that the transmembrane domain alone is sufficient to catalyze ion transport. The large central loop is localized intracellularly (8,14) and involved in the regulation of the exchanger by cytoplasmic Ca 2ϩ (15,16), cytoplasmic Na ϩ (17), ATP depletion (12,13), and protein phosphorylation by protein kinase C (13,18). In the transmembrane domain of all members of Na ϩ /Ca 2ϩ exchanger family, there are two highly conserved internal repeat sequences designated the ␣-1 and ␣-2 repeats that comprise most of TMs 2-3 and 8 -9 and the loops connecting these TMs, respectively (5,19). 2 In such repeat sequences, the putative loop regions are more variable than the transmembrane segments. These homologous sequences may be functionally important, because mutations in the putative TMs within the ␣ repeats cause a large reduction in exchange activity or a change in the I-V relationship in NCX1 (5). In addition, the Thr 103 to Val mutation at the cytoplasmic end of TM2 in NCX1 produces changes in the apparent affinity for intracellular Na ϩ and the selectivity for Li ϩ (7). Therefore these repeat sequences might form part of the structure involved in the ion translocation in the exchanger molecule.
We have recently found that three mammalian exchanger isoforms have distinct differences in their biochemical and pharmacological properties (20). Divalent cation Ni 2ϩ inhibits the reverse mode of Na ϩ /Ca 2ϩ exchange in NCX3 with a 10-fold less affinity than in NCX1 or NCX2. On the other hand, the recently identified inhibitor KB-R7943 is 3-fold more inhibitory 1  to NCX3 than to NCX1 or NCX2. Furthermore, stimulation of Na ϩ /Ca 2ϩ exchange by externally added monovalent cation Li ϩ is significantly greater in NCX2 and NCX3 than in NCX1, although these isoforms exhibit low affinity for Li ϩ . Despite these differences, however, all the NCX isoforms have similar apparent affinities for extracellular transport substrates Ca 2ϩ and Na ϩ (20,21). In this study, taking advantage of high sequence identity in the NCX isoforms, we used chimeric constructs between NCX1 and NCX3 to study the structural domain(s) responsible for the difference in their sensitivity to Ni 2ϩ or Li ϩ . We identified four amino acid residues within the ␣-1 and ␣-2 repeats of the exchanger molecule that are predominantly responsible for the observed differential effects of Ni 2ϩ and Li ϩ .

EXPERIMENTAL PROCEDURES
Cell Cultures-CCL39 cells (American Type Culture Collection) and their NCX transfectants were maintained in Dulbecco's modified Eagle's medium supplemented with 7.5% heat-inactivated fetal calf serum, 50 units/ml penicillin, and 50 g/ml streptomycin.
Substitution of amino acid residues within the internal ␣ repeat regions was performed by site-directed mutagenesis. In this procedure, DNA fragments were produced by polymerase chain reaction using pCRII-NCX1 or pCRII-NCX3 as a template and following pairs of primers: for mutation in the ␣-1 repeat, the sense primers contained an exogenous BamHI site and normal NCX sequences, whereas the antisense primers contained sequences with substituted nucleotides and an exogenous SalI site; for mutation in the ␣-2 repeat, the sense primers contained an exogenous NheI site and normal NCX sequences, while the antisense primers contained an exogenous MluI site and sequences with desired mutations. The polymerase chain reaction products were digested with either BamHI and SalI or NheI and MluI, and inserted into pCRII-NCX1Ј or pCRII-NCX3Ј, and then the full-length mutant cDNAs were transferred into pKCRH. Successful construction was verified by sequencing.
To stably express chimeric and mutant exchangers, pKCRH plasmids carrying NCX cDNAs were transfected in the presence of Lipofectin (Life Technologies, Inc.) into CCL39 fibroblasts that exhibit little endogenous Na ϩ /Ca 2ϩ exchange activity (13,20,24). Cell clones expressing high Na ϩ /Ca 2ϩ exchange activity were selected by a Ca 2ϩkilling procedure as described previously (24).
Assay of Na ϩ i -dependent 45 Ca 2ϩ Uptake-Assay of Na ϩ i -dependent 45 Ca 2ϩ uptake into cells were described in detail previously (20). Briefly, confluent NCX transfectants in 24-well dishes were loaded with Na ϩ by incubation at 37°C for 30 min in 0.5 ml of BSS (10 mM Hepes/Tris (pH 7.4), 146 mM NaCl, 4 mM KCl, 2 mM MgCl 2 , 0.1 mM CaCl 2 , 10 mM glucose, and 0.1% bovine serum albumin) containing 1 mM ouabain and 10 M monensin. 45 Ca 2ϩ uptake was then initiated by switching the medium to Na ϩ -free BSS (replacing NaCl with equimolar choline chloride) or to normal BSS, both of which contained 0.1 mM 45 CaCl 2 (1.5 Ci/ml) and 1 mM ouabain. After a 30-s incubation, 45 Ca 2ϩ uptake was terminated by washing cells four times with an ice-cold solution containing 10 mM Hepes/Tris (pH 7.4), 120 mM choline chloride, and 10 mM LaCl 3 . Cells were then solubilized with 0.1 N NaOH, and aliquots were taken for determination of radioactivity and protein.
Measurement of Whole Cell Exchange Currents-Outward and inward currents from NCX transfectants were measured using the whole cell voltage clamp technique as described previously (25). For recording the outward current, the external solution contained 150 mM NaCl, 1 mM MgCl 2 , 0 or 2 mM CaCl 2 , 20 M ouabain, 2 M nicardipine, 5 M ryanodine, and 5 mM Hepes (pH 7.2), whereas the pipette solution contained 20 mM NaCl, 90 mM CsOH, 40 mM aspartic acid, 3 mM MgCl 2 , 10 mM CaCl 2 , 5 mM MgATP, 5 mM K 2 CrP, 20 mM BAPTA, and 20 mM Hepes (pH 7.2). The ionized Ca 2ϩ concentration in the pipette solution was calculated to be 0.14 M. The outward exchange current was activated by switching the external solution from one without CaCl 2 to one with CaCl 2 . For recording the inward current, the external solution contained 140 mM choline chloride or NaCl, 1 mM MgCl 2 , 20 M ouabain, 2 M nicardipine, 10 M ryanodine, and 5 mM Hepes (pH 7.2), while the pipette solution contained 30 mM CsCl, 90 mM CsOH, 50 mM aspartic acid, 3 mM MgCl 2 , 16 mM CaCl 2 , 5 mM MgATP, 5 mM K 2 CrP, 20 mM BAPTA, and 20 mM Hepes (pH 7.2). The ionized Ca 2ϩ concentration in the pipette solution was calculated to be 1.75 M. The inward exchange current was induced by switching the choline chloride-containing external medium to the Na ϩ -containing external medium. All experiments were performed at about 35°C and the holding and test potentials were Ϫ40 mV. All data were acquired and analyzed by the pCLAMP (Axon Instrument) software.
Statistical Analysis-Data are expressed as mean Ϯ S.E. of three to five independent determinations. Differences for multiple comparisons were analyzed by unpaired t test or one-way ANOVA followed by the Dunnett's test. Values of p Ͻ 0.05 were considered statistically significant.

FIG. 1. Schematic representation of chimeric constructs between NCX1
and NCX3. Two series of chimeras (N1 and N3 series) were constructed by substituting segments of NCX1 with homologous segments of NCX3 and vice versa. In each series, chimeras are named based on the amino acid numbers for respective isoforms. The restriction enzyme cut sites are shown by broken lines and the substituted segments were indicated by black boxes. The numbered shaded boxes at the top and bottom show the positions of TMs in the exchanger predicted on the basis of the hydropathy analysis.

Inhibition of Exchange Activities of NCX1 and NCX3 by
Ni 2ϩ -We previously measured the whole cell current from cloned dog cardiac NCX1 expressed in CCL39 cells using a conventional patch-clamp technique (25). Using the same method, we measured the outward and inward currents evoked in NCX1-or NCX3-transfected CCL39 cells by the extracellular application of 2 mM Ca 2ϩ or 140 mM Na ϩ (Fig. 2, A and B). These currents were reproducibly measured; after a recovery interval, a second pulse of Ca 2ϩ or Na ϩ generated the outward or inward current whose peak value was 96 -102% of the corresponding first current. However, they were never observed in nontransfected CCL39 cells. Inclusion of Ni 2ϩ in the external medium caused inhibition of the outward and inward currents (Fig. 2, A and B). We found that the outward current in NCX1transfected cells was approximately 10-fold more sensitive to Ni 2ϩ than that in NCX3-transfected cells (IC 50 , 39 versus 310 M) and that the inhibition in these cells reached 92 and 81%, respectively, at 3 mM Ni 2ϩ (Fig. 2C). A similar difference in the sensitivity to Ni 2ϩ was observed for the inward currents in NCX1-and NCX3-transfected cells (IC 50 , 25 versus Ͼ100 mM), although its inhibition by Ni 2ϩ occurred at much greater concentrations compared with the outward current ( Fig. 2, C and D).
Our previous measurement of Ni 2ϩ dependence of Na ϩ i -dependent 45 Ca 2ϩ uptake gave IC 50 values of 33 and 343 M for Ni 2ϩ in NCX1-and NCX3-transfected CCL39 cells, respectively (Ref. 20, see also Fig. 3, inset), in agreement with the observed inhibitory potencies of Ni 2ϩ on the outward current. Thus both electrophysiological and biochemical data establish that NCX1 and NCX3 exhibit a 10-fold difference in the sensitivity to Ni 2ϩ , at least when they function in the reverse exchange mode.
NCX Structural Domains Important for Determination of Ni 2ϩ Sensitivity-Taking advantage of the fact that NCX1 and NCX3 exhibit high amino acid homology, we used chimeric constructs between the two isoforms to study the structural domain(s) responsible for the difference in their sensitivity to Ni 2ϩ . Using endogenous or newly introduced common restriction enzyme sites, we constructed two series of chimeric cDNAs in which one or two segments from NCX3 were incorporated into NCX1 in exchange for the homologous segment(s) in the latter (N1 series chimeras), and vice versa (N3 series chimeras) (see Fig. 1). We stably expressed these chimeras in CCL39 fibroblasts. All transfectants, after selection by a Ca 2ϩ -killing procedure, exhibited exchange activities similar to that of cells expressing the wild-type NCX1 or NCX3 (see the legend to Fig.  3).
We examined the effect of 0.1 mM Ni 2ϩ on the initial rate of Na ϩ i -dependent 45 Ca 2ϩ uptake into cells expressing these chimeric exchangers (Fig. 3). Under the conditions used, Ni 2ϩ at this concentration reduced the uptake rate of the wild-type NCX1 or NCX3 by 77 or 24%, respectively. N1 series chimeras N1-109/133, N1-788/829, and N1-109/133,788/829, which contained the homologous BamHI-SalI and/or NheI-MluI segments from NCX3 (Fig. 1), exhibited significantly less sensitivity to inhibition by 0.1 mM Ni 2ϩ compared with the wild-type

FIG. 3. Effects of Ni 2؉ on Na ؉
i -dependent 45 Ca 2؉ uptake into cells expressing N1 (A) and N3 series (B) chimericexchangers. TheinitialrateofNa ϩ idependent 45 Ca 2ϩ uptake was measured in the presence or absence of 0.1 mM Ni 2ϩ as described under "Experimental Procedures." The uptake rates in cells expressing these chimeras were 5-10 nmol/mg/30 s in the absence of Ni 2ϩ , which were similar to those in cells expressing the wildtype NCX1 or NCX3 (9 and 6 nmol/mg/30 s for NCX1 and NCX3, respectively). Data are presented as percentage of the values obtained in the absence of Ni 2ϩ . Data are mean Ϯ S.E. of three independent experiments. *, p Ͻ 0.05 versus NCX1 (in A); *, p Ͻ 0.05 versus NCX3 (in B). Inset, doseresponse curves for inhibition of the uptake by Ni 2ϩ in cells expressing NCX1 or NCX3. NCX1 (Fig. 3A). In contrast, N3 series chimeras N3-143/167, N3-777/818, and N3-143/167,777/818 containing the BamHI-SalI and/or NheI-MluI segments from NCX1 ( Fig. 1) were more sensitive to inhibition by Ni 2ϩ compared with the wild-type NCX3 (Fig. 3B). All other chimeras, however, were not significantly different from the parental NCX1 or NCX3. For some of those chimeras with an altered sensitivity to Ni 2ϩ , we determined dose-response profiles for Ni 2ϩ by measuring the rate of Na ϩ i -dependent 45  Intriguingly, the BamHI-SalI and NheI-MluI segments correspond, respectively, to the major portions of the phylogenetically conserved regions of the Na ϩ /Ca 2ϩ exchanger designated the ␣-1 and ␣-2 repeats (see Fig. 1) (19). Na ϩ /Ca 2ϩ exchange has been shown to be highly sensitive to mutagenesis in the putative transmembrane helices within the ␣ repeats (5), suggesting the functional importance of these regions.
Amino Acids Residues Involved in Differential Sensitivity to Ni 2ϩ - Fig. 4 shows amino acid sequences of the ␣ repeat regions from NCX1, NCX2, and NCX3. Of note, in these regions, there are only a small number of amino acids unique to each isoform, all of which are localized within the BamHI-SalI and NheI-MluI segments. To identify the residues involved in the differential Ni 2ϩ sensitivity, these unique residues in the ␣ repeats were exchanged between NCX1 and NCX3 or mutated to other amino acids (Fig. 5). Three amino acids Val, Asn, and Thr in the ␣-1 repeat of NCX1 (corresponding to Val 118 , Asn 125 , and Thr 127 ) were replaced with Leu, Gly, and Ile from the same region of NCX3 (corresponding to Leu 152 , Gly 159 , and Ile 161 ), respectively, and vice versa. We found that single substitution mutants were not different from the parental NCX1 or NCX3 in their sensitivity to 0.1 mM Ni 2ϩ (Fig. 5, A and B). On the other hand, double substitution mutants N1-N125G/T127I and its reciprocal mutant N3-G159N/I161T showed decreased and increased sensitivities to inhibition by Ni 2ϩ , respectively, which are comparable to those seen in N1-109/133 and N3-143/167 chimeras (compare Figs. 3 and 5). Double mutants N1-V118L/ N125G and N1-V118L/T127I, however, exhibited Ni 2ϩ sensitivity similar to that of NCX1 (data not shown). Thus, Asn 125 and Thr 127 in the ␣-1 repeat of NCX1 and corresponding Gly 159 and Ile 161 in NCX3 are responsible for the altered responses of N1-109/133 and N3-143/167 to Ni 2ϩ . We also examined the Ni 2ϩ sensitivity of NCX1 mutants with Asn 125 to Cys (N1-N125C) or Thr 127 to Cys substitution (N1-T127C). Interestingly, N1-N125C, but not T127C, exhibited an increased sensitivity to inhibition by Ni 2ϩ relative to that of the parental NCX1 (Fig. 5A).
We found that the Val 820 to Ile substitution in NCX1 (N1-V820I) and the corresponding Ala 809 to Ile substitution in NCX3 (N3-A809I) resulted in an increase in the sensitivity to inhibition by Ni 2ϩ , whereas substitution of these residues with Gly resulted in a decrease in the Ni 2ϩ sensitivity in N1-V820G, but not in N3-A809G (Fig. 5, A and B). Other mutants N1-Q826E and N1-Q826R and the corresponding mutants N3-V815E and N3-815R showed the Ni 2ϩ sensitivity not different from that of the parental NCX1 or NCX3.
Structural Domains and Amino Acids Residues Involved in Li ϩ -induced NCX Activation-Monovalent cation Li ϩ stimulates Na ϩ i -dependent 45 Ca 2ϩ uptake by all three NCX isoforms with low affinity, with the extent of stimulation being greater in NCX2 or NCX3 than in NCX1 (20). To obtain insight into the structural domains involved in the Li ϩ -induced NCX activation, we examined the effect of Li ϩ on the exchange activities of the chimeras and mutants described above. Consistent with the previous result, extracellular application of Li ϩ caused a dose-dependent increase in the rate of Na ϩ i -dependent Ca 2ϩ uptake into NCX1-or NCX3-transfected cells, which reached about 145 and 270% at 146 mM Li ϩ , respectively, of the control measured in the absence of Li ϩ (Fig. 6, inset). Fig. 6, A and B, show the extent of stimulation of the uptake by various chimeras and mutants at 146 mM Li ϩ . Intriguingly, N1-788/829 and N1-109/133,788/829 containing a major portion of the ␣-2 repeat from NCX3 produced NCX1 mutants exhibiting Li ϩ sensitivity almost identical to that of the wild-type NCX3. Conversely, the reciprocal mutants N3-777/818 and N3-143/ 167,777/818 exhibited Li ϩ sensitivity similar to that in the wild-type NCX1. On the other hand, other N1 and N3 chimeras were not different from the wild-type NCX1 or NCX3, respectively. Thus the ␣-2 repeat seems to be exclusively responsible for the differential stimulation by Li ϩ .
We examined the effect of substitution of four unique amino acids in the ␣-2 repeat on the response of NCX1 or NCX3 to 146 mM Li ϩ (Fig. 6). Single substitution mutants N1-L808F, N1-V820A, N1-T823L, N1-Q826V, N3-F797L, N3-A809V, N3-L812T, and N3-V815Q exhibited Li ϩ sensitivity not different from that of each parental exchanger (Fig. 6, A and B). On the other hand, double substitution mutant N1-V820A/Q826V and its reciprocal mutant N3-A809V/V815Q showed Li ϩ -dependent stimulation similar to those of the wild-type NCX3 and NCX1, respectively. Thus both Val 820 and Gln 826 of NCX1 and Ala 809 and Val 815 of NCX3 are responsible for the different response of the exchanger to Li ϩ .
Interestingly, 146 mM Li ϩ decreased Na ϩ i -dependent 45 Ca 2ϩ uptake by N1-V820I to a level below that of the parental NCX1 without Li ϩ (Fig. 6A). The corresponding NCX3 mutant N3-A809I also exhibited significantly less stimulation by Li ϩ than NCX3 (Fig. 6B). In contrast, the uptake rate was stimulated to a greater extent in N1-V820G and the corresponding N3-A809G than the parental NCX1 or NCX3 (Fig. 6, A and B).
We measured the initial rates of Na ϩ i -dependent 45 Ca 2ϩ uptake by NCX1 ␣ repeat mutants as a function of [Ca 2ϩ ] o . Double-reciprocal plots of uptake rates versus [Ca 2ϩ ] o were all linear for the wild-type NCX1, N1-N125C, N1-N125G/T127I, N1-V820I, N1-V820A/Q826V, and the wild-type NCX3 in the presence or absence of Ni 2ϩ , Na ϩ , or Li ϩ (Fig. 8A, and  V max estimated from these plots are summarized in Table I. An important finding is that in the absence of Ni 2ϩ , Na ϩ , or Li ϩ , the K m(Ca) values in these mutants were not different from that of the wild-type NCX1, except for N1-N125C. Other mutants such as N1-N125G, N1-T127C, N1-V820A, and N1-Q826V also had K m(Ca) values similar to that of the wild-type NCX1 (data not shown). In N1-N125C, however, the K m(Ca) increased 3-fold (p Ͻ 0.05) compared with NCX1 (Table I).
External Na ϩ increased K m(Ca) values in these NCX1 ␣ repeat mutants (Table I), consistent with the fact that Na ϩ is transported by the exchanger as a substrate, although like Li ϩ , it also functions as a stimulating monovalent cation at relatively low concentrations (20, 26 -28). Since N1-N125C exhibited a decreased affinity for Ca 2ϩ o compared with NCX1 (see above), we measured the inward exchange currents evoked in N1-N125C-transfected cells as a function of [Na ϩ ] o . We found a 1.4-fold increase (p Ͻ 0.05) in the K m(Na) in N1-N125C compared with the wild-type NCX1 (Fig. 9).
Application of Ni 2ϩ increased K m(Ca) without influencing V max in the NCX1 ␣ repeat mutants (Fig. 8A, Table I, and data not shown). We used the Dixon plot to identify the type of inhibition for Ni 2ϩ and determine its intrinsic inhibition constant (K i(Ni) ). The obtained profiles of Dixon plots were consistent with the view that Ni 2ϩ competes with Ca 2ϩ for the same site ( Fig. 8B and data not shown). K i(Ni) values calculated from the Dixon plots are summarized in Table II. N1-N125C and N1-V820I showed a 3-fold decrease in the K i(Ni) compared with wild-type NCX1, whereas N1-N125G/T127I and N1-V820A/ Q826V exhibited a 4-fold increase. We found that N1-V820A had a K i(Ni) value (40 M) similar to that for N1-V820A/Q826V.
Li ϩ at 146 mM, on the other hand, did not significantly affect K m(Ca) in NCX1, N1-N125C, N1-N125G/T127I, N1-V820A/ Q826V, and NCX3, whereas it increased V max for these exchangers by 35, 48, 35, 119, and 143%, respectively (Table I). In N1-V820I, however, Li ϩ increased K m(Ca) by 6-fold and V max by 44%. In this mutant, therefore, Li ϩ markedly decreased the apparent Ca 2ϩ o affinity, although the Li ϩ stimulation remained similar to that in NCX1. DISCUSSION Previous studies have demonstrated that external divalent and trivalent cations inhibit Ca 2ϩ transport by Na ϩ /Ca 2ϩ ex-  changer, whereas external monovalent cations such as Li ϩ activates it (20, 26 -31). Ni 2ϩ and Mg 2ϩ have been shown to inhibit Na ϩ i -dependent 45 Ca 2ϩ uptake into cardiomyocytes, smooth muscle cells, or CCL39 cells expressing different NCX isoforms by apparently competing with Ca 2ϩ for the extracellular transport site (20,30,31). Recently, we have found that Ni 2ϩ inhibits Ca 2ϩ uptake by NCX1 or NCX2 10-fold more potently than that by NCX3, although all three isoforms exhibit similar affinity for Ca 2ϩ o or Na ϩ o (20,21). On the other hand, stimulation by external Li ϩ is greater in NCX2 or NCX3 than in NCX1. Using a whole cell patch clamp technique and 45 Ca 2ϩ uptake measurement, we confirmed these differential effects of Ni 2ϩ and Li ϩ on NCX1 and NCX3 (Figs. 2, 3, and 6). In addition, we noted that in both NCX1 and NCX3, the inward exchange current exhibits much lower sensitivity to inhibition by Ni 2ϩ compared with the outward exchange current (Fig. 2). The reason for this difference, however, is not clear at present. In these current measurements, we used different intracellular ionic conditions (20 mM Na ϩ and 0.14 M ionized Ca 2ϩ for the outward and 0 mM Na ϩ and 1.75 M ionized Ca 2ϩ for the inward current). It is possible that the different ionic conditions alter the kinetics of the exchange reaction and could thus influence the IC 50 for Ni 2ϩ . Further studies are required to clarify this point.
In the present study, we attempted to identify the structural domains of the Na ϩ /Ca 2ϩ exchanger that are responsible for the differential response to Ni 2ϩ or Li ϩ , which may provide important information about the mechanism for the ion ex-change and its modulation. To accomplish this, a series of chimeras were constructed between NCX1 and NCX3 and the effects of Ni 2ϩ and Li ϩ on Ca 2ϩ uptake by them were compared with those on the uptake by the parental exchangers. All chimeric constructs exhibited exchange activities comparable to those of the parental exchangers, indicating that the compatibility of the segments exchanged between the two isoforms are excellent.
Analysis with N1 series chimeras, in which one or two segments in NCX1 were replaced by the corresponding segment(s) from NCX3, revealed that the BamHI-SalI (amino acids 109 -113) and NheI-MluI (amino acids 788 -829) segments of NCX1 (see Fig. 1) were individually involved in the reduction of Ni 2ϩ sensitivity and that these segments accounted for ϳ80% of the difference in Ni 2ϩ sensitivity between NCX1 and NCX3 when both were replaced simultaneously ( Fig. 3A and "Results"). We obtained exactly the reciprocal results with N3 series chimeras N3-143/167 and N3-143/167,777/818, in which the BamHI-SalI and/or NheI-MluI segments from NCX1 were transplanted into NCX3 (Fig. 3B). On the other hand, the extent of Li ϩ -induced stimulation of Ca 2ϩ uptake by N1 and N3 series chimeras was found to be determined exclusively by the NheI-MluI segment (Fig. 6, A and B; see also Fig. 1). Importantly, the BamHI-SalI and NheI-MluI segments correspond, respectively, to major portions of the internal ␣-1 and ␣-2 repeats which are conserved in all known members of the Na ϩ /Ca 2ϩ exchanger family (19). These results suggest that both ␣-1 and ␣-2 repeats are predominant determinants of the differential sensitivity of NCX1 and NCX3 to Ni 2ϩ , whereas the ␣-2 repeat alone is sufficient for generating the difference in Li ϩ sensitivity.
We found that only three and four amino acids are different in the ␣-1 and ␣-2 repeats of NCX1 and NCX3, respectively, all of them in fact being located within the BamHI-SalI or NheI-MluI segments (see Figs. 1 and 4). These amino acids were exchanged individually or in combination between NCX1 and NCX3 to evaluate their functional importance. Screening of these single and double substitution mutants of the ␣-1 repeat for a change in Ni 2ϩ sensitivity permitted us to identify the double substitution mutant N1-N125G/T127I and its reciprocal mutant N3-G159N/I161T that exhibited altered Ni 2ϩ sensitivities equivalent to those seen in chimeras N1-109/133 and N3-143/167, respectively (Figs. 3 and 5). Similarly, single substitution mutants N1-V820A and N3-A809V of the ␣-2 repeat were found to exhibit reciprocal changes in Ni 2ϩ sensitivity comparable to those in chimeras N1-788/829 and N3-777/818, respectively (Figs. 3 and 5). Double substitution mutants N1-V820A/Q826V and N3-A809V/V815Q also exhibited an altered Ni 2ϩ sensitivity, but their responses were not different from those of single mutants N1-V820A and N3-A809V. On the other hand, these double substitution mutants reproduced the differential Li ϩ responses seen in the N1-788/829 and N3-777/818 (Fig. 6, A and B). Thus, the effects of the BamHI-SalI and NheI-MluI segments on Ni 2ϩ sensitivity are mimicked, respectively, by simultaneous substitution of two amino acids in the ␣-1 repeat and substitution of one amino acid in the ␣-2 repeat, while the effect of the NheI-MluI segment on Li ϩ sensitivity was mimicked by simultaneous substitution of two amino acids in the ␣-2 repeat.
We kinetically characterized those NCX1 mutants, i.e. N1-N125G/T127I, N1-V820A, and N1-V820A/Q826V, that exhibited altered sensitivity to Ni 2ϩ or Li ϩ . In these mutants, no significant change in apparent Ca 2ϩ o affinity was observed (Table I and "Results"). Na ϩ o affinity also did not seem to be altered, as deduced from the effect of 50 mM Na ϩ o on K m(Ca) ( Table I). On the other hand, the values of intrinsic inhibition constant for Ni 2ϩ (K i(Ni) ) obtained from the Dixon plots in-  creased 3-4-fold in these mutants compared with NCX1 (Table  II and "Results"). In contrast, among these mutants, N1-V820A/Q826V alone exhibited an enhanced response to Li ϩ (Fig. 7B and "Results"). Taken together, the results indicate that in NCX1, the Asn 125 to Gly and Thr 127 to Ileu double substitution in the ␣-1 repeat and the Val 820 to Ala single substitution in the ␣-2 repeat individually cause a decrease in the apparent affinity of the exchanger for Ni 2ϩ and that such decreases occur without alteration in the apparent affinity for Ca 2ϩ o , Na ϩ o , or Li ϩ . Thus, Ni 2ϩ does not seem to compete directly with Ca 2ϩ o for the transport site, although it increased K m(Ca) without influencing V max in these mutants (Table I and "Results"). It seems likely that Ni 2ϩ influences Ca 2ϩ o binding to the transport site through an indirect interaction with the latter, consistent with the view that the sites for Ca 2ϩ o and Ni 2ϩ in the exchanger are not identical. Likewise, the sites for Ni 2ϩ and Li ϩ seem not to be identical, which is consistent with our previous results (20). Furthermore, Li ϩ does not affect the binding of Ca 2ϩ to these NCX1 mutants, since application of Li ϩ (146 mM) did not alter their K m(Ca) values as in the wildtype NCX1 (Table I).
We wish to discuss the possible functions of the four amino acids identified in the ␣-1 and ␣-2 repeats of NCX1. The Asn 125 to Cys mutation (N1-N125C) caused 3-and 1.4-fold increases in the values of K m(Ca) and K m(Na) , respectively, compared with those of the wild-type NCX1 (Fig. 8, A and B, Table I, and Fig.  9), while it caused a 2.5-fold decrease in the K i(Ni) relative to NCX1 (Table II). However, this mutant showed Li ϩ concentration dependence of a NCX1 type (Fig. 7B) and its K m(Ca) was not affected by 146 mM Li ϩ (Table I). Thus, the Asn 125 to Cys mutation significantly alter the apparent affinities of the exchanger for Ca 2ϩ o , Na ϩ o , and Ni 2ϩ , but not the response to Li ϩ . The side chain carbonyl of Asn 125 does not ligand any of these cations directly, since the Asn 125 to Gly substitution produced no effect on the interaction of the exchanger with all these cations. When both Asn 125 and nearby Thr 127 were mutated to Gly and Ile (N1-N125G/T127I), the apparent Ni 2ϩ affinity decreased 3-fold without influencing both the affinity for Ca 2ϩ o and the response to Li ϩ (Tables I and II, and Fig. 7B). However, their single mutants N1-N125G, N1-T127I, and N1-T127C showed no change in their response to Ca 2ϩ o , Ni 2ϩ , or Li ϩ . It appears therefore that the effect of mutation in Thr 127 is relatively minor, producing a large decrease in the apparent Ni 2ϩ affinity only when Asn 125 is simultaneously mutated. We obtained no evidence for the involvement of Asn 125 and Thr 127 in the Li ϩ -dependent stimulation, although mutation studies involving these residues are still incomplete.
Changing Val 820 in the ␣-2 repeat of NCX1 to Ala, Gly, or Ile altered the Ni 2ϩ sensitivity without influencing the K m(Ca) (Figs. 5A and 7A, Tables I and II, and "Results"). The Val 820 to Gly mutation also caused a significant increase in the extent of stimulation by Li ϩ (Fig. 6A). Intriguingly, in the Val 820 to Ile mutant, 146 mM Li ϩ inhibited Ca 2ϩ uptake in 0.1 mM Ca 2ϩ (Figs. 6A and 7B). This mutant does not appear to have altered interaction with Li ϩ , as the increase of its V max by 146 mM Li ϩ was similar to that in the wild-type NCX1 (Table I). We found that K m(Ca) of this mutant was 6-fold larger than that of NCX1 in the presence of 146 mM Li ϩ (Table I). Thus the observed inhibition of Ca 2ϩ uptake by N1-V820I in 0.1 mM Ca 2ϩ was due to a marked decrease in the apparent Ca 2ϩ o affinity caused by high Li ϩ . Since the Ca 2ϩ o affinity of the same mutant in the absence of Li ϩ was not different from that of NCX1, it is likely that Li ϩ binding to its site exerts a large indirect effect on Ca 2ϩ o binding in this mutant. On the other hand, single mutation of Gln 826 to Val, Glu, or Arg did not affect the response of NCX1 to Ca 2ϩ o , Ni 2ϩ , or Li ϩ , suggesting that mutation in Gln 826 seems to exert a relatively minor effect. However, the Gln 826 to Val mutation was able to enhance Li ϩ stimulation when Val 820 was simultaneously mutated to Ala (Figs. 6A and 7B). In addition, Gln 826 could potentially influence Ni 2ϩ affinity, since the NCX3 single mutant N3-V815Q exhibited a slightly but significantly higher sensitivity to inhibition by Ni 2ϩ relative to the wild-type NCX3 (Fig. 5B). Val 815 in NCX3 is equivalent to Gln 826 in NCX1.
All these results show that in NCX1, mutations of Asn 125 and Thr 127 in the ␣-1 repeat and of Val 820 and Gln 826 in the ␣-2 repeat are able to cause alteration in the interaction of the exchanger with Ca 2ϩ and Ni 2ϩ , and Ni 2ϩ and Li ϩ , respectively. Since all these residues do not appear to serve as direct ligands for cations (see above) and since the nature of the side chains of substituted residues appears to determine the type of the cation affected, it is likely that mutations at these residues are able to induce conformational changes in the neighboring cation-binding sites, thereby influencing the interaction of the latter with externally applied cations. However, involvement of a long-range allosteric effect also cannot be ruled out.
At present, we have little information about the location of the binding sites for Ca 2ϩ , Ni 2ϩ , and Li ϩ , and the amino acids directly liganding these cations. Interestingly, the four amino acid residues identified here are mostly localized in the putative loops in the ␣ repeats. By substituted cysteine accessibility mapping of the NCX1 polypeptide using the inhibition of exchange activity of the mutants by membrane-impermeant SHlabeling reagent as a marker, we have recently provided evidence suggesting that the loop in the ␣-1 repeat forms a reentrant membrane loop with both ends facing the extracellular side and at least one of the residues (Asn 125 ) being accessible from the inside, while the corresponding region in the ␣-2 repeat is mostly exposed on the cytoplasmic side (10). The opposite membrane orientation of the ␣-1 and ␣-2 repeats has also been suggested by Nicoll et al. (9). The same latter group has recently reported that Cys 151 and Thr 815 (or Ala 821 ) within the ␣-1 and ␣-2 repeats are localized close to each other in space in NCX1 (32). These structural data appear to be consistent with a hypothesis that the putative loop region of each ␣ repeat has an opposite orientation and is relatively close to each other within the membrane bilayer, possibly forming an ion interaction domain within the ion transport pathway of the exchanger molecule.