Mutational Analysis of the α-1 Repeat of the Cardiac Na+-Ca2+ Exchanger*

The Na+-Ca2+ exchanger contains internal regions of sequence homology known as the α repeats. The first region (α-1 repeat) includes parts of transmembrane segments (TMSs) 2 and 3 and a linker modeled to be a reentrant loop. To determine the involvement of the reentrant loop and TMS 3 portions of the α-1 repeat in exchanger function, we generated a series of mutants and examined ion binding and transport and regulatory properties. Mutations in the reentrant loop did not substantially modify transport properties of the exchanger though the Hill coefficient for Na+ and the rate of Na+-dependent inactivation were decreased. Mutations in TMS 3 had more striking effects on exchanger activity. Of mutations at 10 positions, 3 behaved like the wild-type exchanger (V137C, A141C, M144C). Mutants at two other positions expressed no activity (Ser139) or very low activity (Gly138). Six different mutations were made at position 143; only N143D was active, and it displayed wild-type characteristics. The highly specific requirement for an asparagine or aspartate residue at this position may indicate a key role for Asn143 in the transport mechanism. Mutations at residues Ala140 and Ile147 decreased affinity for intracellular Na+, whereas mutations at Phe145 increased Na+ affinity. The cooperativity of Na+ binding was also altered. In no case was Ca2+ affinity changed. TMS 3 may form part of a site that binds Na+ but not Ca2+. We conclude that TMS 3 is involved in Na+ binding and transport, but previously proposed roles for the reentrant loop need to be reevaluated.

Na ϩ -Ca 2ϩ exchangers are found in a wide variety of tissues. These plasma membrane proteins are electrogenic transporters that utilize the electrochemical gradient of Na ϩ to exchange three extracellular Na ϩ ions for one intracellular Ca 2ϩ . As a Ca 2ϩ efflux mechanism, the exchanger helps maintain intracellular Ca 2ϩ homeostasis. Cloning of the Na ϩ -Ca 2ϩ exchanger (1) revealed a protein composed of 938 amino acids organized in nine membrane-spanning regions with the first five transmembrane segments (TMSs) 1
Much progress has been made in understanding exchanger topology, function, and regulation. Modulation by intracellular factors such as Na ϩ and Ca 2ϩ ions, ATP, and PIP 2 , have been extensively studied, and the domains involved in regulation have been localized to the large cytoplasmic loop (6). Less is known about the regions of the Na ϩ -Ca 2ϩ exchanger involved in the translocation of Na ϩ and Ca 2ϩ . Two regions with similar sequences, called the ␣-1 and ␣-2 repeats, likely have key roles. The ␣-1 repeat consists of residues spanning TMSs 2 and 3 and an extracellular loop connecting the two TMSs. The homologous ␣-2 repeat encompasses TMS 7 and part of the following intracellular loop (Fig. 1). Several observations emphasize the functional importance of the ␣ repeats. First, they are highly conserved among all Na ϩ -Ca 2ϩ exchangers and are the only regions of sequence conservation between members of the NCX and NCKX families (7). Second, data suggest that the portions of the ␣ repeats connecting TMS 2-3 and TMS 7-8 form reentrant membrane loops ( Fig. 1 and Ref. 2). Reentrant loops have been identified in ion channels ("P-Loops"), and in the aquaporin family of proteins and form a portion of the hydrophilic pathway for ions or other molecules to cross the membrane (8,9). By analogy, the reentrant loops within the ␣ repeats may be involved in the translocation pathway for Na ϩ and Ca 2ϩ . This hypothesis is supported by helix packing studies, which demonstrate that the two ␣ repeats are in proximity (10). In addition, site-directed mutagenesis reveals that mutations in the ␣ repeats drastically alter the transport of Na ϩ and Ca 2ϩ (4,11,12), supporting a role in ion translocation.
The aim of this study was to further define the role of the ␣-1 repeat in ion translocation by combining mutational and biophysical analysis. Mutations were made in the first reentrant loop and in TMS 3. We show that mutations within the reentrant loop do not cause major perturbations in ion translocation; all mutations lead to active exchangers with little alteration in ion affinities or regulatory properties. These data seem to exclude the reentrant loop of the ␣-1 repeat as a key component of ion translocation. In contrast, residues within TMS 3 are important in determining Na ϩ affinity. Mutations of residues Ala 140 , Phe 145 , and Ile 147 drastically alter the apparent affinity for Na ϩ , without affecting the binding of transported Ca 2ϩ . Thus, TMS 3 is important in determining the ion dependence of the Na ϩ -Ca 2ϩ exchanger.

Molecular Biology and 45 Ca 2ϩ Uptake Measurements-Mutations
were generated in 200 -400-bp cassettes from the full-length exchanger using QuikChange Mutagenesis (Stratagene). Cassettes were then se-quenced and subcloned back into the full-length exchanger. RNA was synthesized using T3 mMessage mMachine (Ambion) after linearization with HindIII. RNA was injected into Xenopus oocytes and 45 Ca 2ϩ uptake measured as described previously (13). To measure extracellular Ca 2ϩ dependence, mutants were transiently expressed in BHK or HEK cells, and 45 Ca 2ϩ uptake was measured at different [Ca 2ϩ ]. Data were fit to the Michaelis-Menten equation using GraphPad Prism version 4.0 for Macintosh.
The dog/squid NCX chimera was generated by a combination of mutagenesis and subcloning. The squid exchanger has a BamHI site at the Gly 108 position and an XcmI site at the Ile 136 position. The dog exchanger also has an equivalent XcmI site. A BamHI site was introduced at the equivalent site in the dog NCX1 by mutagenesis. The mutant dog exchanger was then digested with BamHI and XcmI, and the 80-base pair BamHI-XcmI fragment from the squid exchanger was placed into the dog exchanger. This resulted in a chimeric exchanger containing the dog background with the squid reentrant loop between TMSs 2 and 3.
Na ϩ and Ca 2ϩ activation curves were obtained by perfusing solutions with different ion concentrations. Currents were fitted to a Hill function and normalized to extrapolated maximum values. Values are mean Ϯ S.E. Values were considered significantly different at a level of p Ͻ 0.05. pCLAMP (Axon Instruments, Burlingame, CA) software was used for acquisition and analysis. Data were acquired on-line at 4 ms/point and filtered at 50 Hz using an 8-pole Bessel filter. Experiments were performed at 35°C and at a holding potential of 0 mV.

Electrophysiological Characterization of Reentrant Loop
Mutants-We generated point mutants H124N, N125C, F126C, T127H, and D130C of the first reentrant loop of the Na ϩ -Ca 2ϩ exchanger and investigated biophysical properties. Mutants were expressed in Xenopus oocytes and analyzed by both 45 Ca 2ϩ uptake assays and by electrophysiology using the giant excised patch technique. In a typical excised patch experiment, Ca 2ϩ is within the patch pipette at the extracellular surface. An outward Na ϩ -Ca 2ϩ exchange current is activated by the rapid application of Na ϩ to the intracellular surface. In this configuration, the excised patch technique allows assessment of two different regulatory properties (15). First, although Na ϩ provides substrate to activate transport, Na ϩ at the intracellular surface also induces a slow inactivation process known as I 1 (e.g. see Fig. 4 and Ref. 16). Second, intracellular Ca 2ϩ binds to a regulatory site on the large intracellular loop to activate transport function (17,18). To measure apparent ion affinities at transport sites without confounding effects of these regulatory processes, the exchanger can be first "deregulated" by treatment with chymotrypsin at the intracellular surface.
We first determined apparent affinities for Na ϩ , measured at the cytoplasmic surface of deregulated mutants. Our data (see Figs. 2, left panels, and 8) show that the concentration of Na ϩ needed to activate 50% of the mutated exchangers slightly decreased in H124N and D130C and increased for mutant F126C. Replacement of Asn 125 did not produce any apparent effect. Exchangers H124N, F126C, T127H, and D130C also showed a significant decrease in the Hill coefficient with the effects being most substantial for H124N and F126C (Fig. 8). The shallower course of the Na ϩ dependence for H124N and F126C allows these mutants to transport more efficiently at lower Na ϩ concentration than the wild type. For example, Fig. 2 shows that at 5 mM intracellular Na ϩ , H124N and wild-type exchangers had about 23 and 6% of maximal activities, respectively.

FIG. 2.
Mutations within the reentrant loop of the ␣-1 repeat of the Na ؉ -Ca 2؉ exchanger do not impede ion transport. Point mutants were created within the first reentrant loop of the canine cardiac exchanger. Outward currents were generated by perfusing the intracellular surface of the patch with 100 mM Na ϩ in the presence of 8 mM Ca 2ϩ within the pipette. Before recording, the patches were treated with chymotrypsin to remove ionic regulation. Point mutations at residues 124, 125, 126, 127, or 130 generated active exchangers. The modest changes in the Na ϩ affinity measured in H124N, F126C, and D130C were significant as well the change in slope of the H124N, F126C, T127H (not shown), and D130C Na ϩ dependence curves. See Fig. 8 for values. Open and closed squares indicate wild-type and mutant exchangers, respectively.
To determine the affinity of the mutated exchangers for transported Ca 2ϩ , inward exchange currents were generated by application of different cytoplasmic Ca 2ϩ concentrations with 100 mM extracellular Na ϩ within the pipette. We found no changes in the apparent Ca 2ϩ affinity values for any reentrant loop mutant. However, F126C showed a shallower course of the Ca 2ϩ dependence curve (Figs. 2 and 8).
Positions His 124 , Asn 125 , and Asp 130 are of particular interest since a previous study showed that replacement of each of these residues with a cysteine decreased the apparent affinity of the external binding site for transported Ca 2ϩ (4). In our hands, however, neither the N125C nor the D130C exchangers had altered apparent affinities for external Ca 2ϩ . The activities of N125C, D130C, and wild-type exchangers were measured at different extracellular Ca 2ϩ concentrations as Na ϩ gradientdependent 45 Ca 2ϩ uptake into intact cells. N125C and D130C behaved similarly to wild type (Fig. 3). This is consistent with the electrophysiological measurements on the apparent affinities for intracellular Ca 2ϩ .
We also examined the Na ϩ -dependent inactivation, I 1 , of the mutated exchangers (Fig. 4). Mutations H124N and D130C substantially slowed the time constant () of Na ϩ -dependent inactivation. Current decays were fitted to a single exponential to calculate time constants of 8.8 Ϯ 0.7 (n ϭ 8) for H124N, 5.5 Ϯ 1.1 s (n ϭ 6) for D130C and 3.3 Ϯ 0.2 s (n ϭ 23) for the wild-type exchanger. Preliminary experiments also indicate that the mutant exchanger F126C had slowed inactivation (ϳ5 s, n ϭ 2), while mutations at position 125 (n ϭ 2) and 127 (n ϭ 2) caused no change in the inactivation rate. We have previously reported on the inactivation of mutant H124N (3).
Overall, our results indicate that residues forming the putative reentrant loop connecting TMSs 2 and 3 do not set cytoplasmic affinities for transported ions but can alter the cooperativity of Na ϩ binding and inactivation rates.
Functional Activity of the Squid Chimera Exchanger-As shown in Fig. 1, the portion of the ␣-1 repeat that invaginates the membrane is highly conserved in the three mammalian Na ϩ -Ca 2ϩ exchangers (NCX1, NCX2, and NCX3) but diverges in non-mammalian species such as the squid neuronal exchanger (NCX-SQ1). NCX-SQ1 has the following non-conservative changes: C122V, H124Q, N125K, T127E, and D130Q. Interestingly, NCX-SQ1 has biophysical properties similar to the mammalian NCX1 (19). If the reentrant loop is involved in ion transport, then perhaps other parts of the NCX-SQ1 protein compensate for the amino acid divergence within the reentrant loop. To explore this possibility, we replaced the ␣-1 reentrant loop of the mammalian Na ϩ -Ca 2ϩ exchanger with that from NCX-SQ1 (squid chimera).
Previously we have shown that the outward current from the squid chimeric exchanger possesses both Na ϩ -and Ca 2ϩdependent inactivation processes (3). We now performed kinetic analysis of the I 1 regulatory properties of the squid chimera. At a cytoplasmic Ca 2ϩ concentration of 1 M, the time constant for inactivation of squid chimera exchangers was modestly increased as compared with the wild-type cardiac Na ϩ -Ca 2ϩ exchanger (4. 1 Ϯ 0.2 s, n ϭ 3 versus 3.3 Ϯ  0.2 s, n ϭ 23; Fig. 4). The affinities of the squid chimera exchanger for transported ions (measured at the cytoplasmic surface) were determined after the removal of the Na ϩ -and Ca 2ϩ -dependent inactivation processes (I 1 and I 2 ) using chymotrypsin. As shown in Figs. 5A and 8, the squid chimera exchanger displayed a slight but significant increase in Na ϩ affinity and a decrease in cooperativity (see left bottom panel of Fig. 5). The fit of the Na ϩ dependence curves to a Hill function yielded apparent affinities values of 16.6 Ϯ 1.2 mM (n ϭ 6) for the wild type and 13.4 Ϯ 0.7 mM (n ϭ 6) for the squid chimera with corresponding Hill coefficients of 1.9 Ϯ 0.1 and 1.3 Ϯ 0.1. The Ca 2ϩ dependence for the transport site on the cytoplasmic surface was also investigated and no significant differences were found between the canine cardiac Na ϩ -Ca 2ϩ exchanger (wild type) and the exchanger containing residues 122 to 130 of the squid isoform. Panel B of Fig.  5 shows examples of inward currents and the corresponding Ca 2ϩ concentration response curve (see Fig. 8 for values).
The results indicate that replacement of the extracellular portion of the ␣-1 repeat does not cause major perturbations of Na ϩ or Ca 2ϩ transport. There were minor effects on the apparent Na ϩ affinity and more substantial effects on the cooperativity of Na ϩ binding.
Biophysical Properties of Cardiac Na ϩ -Ca 2ϩ Exchangers Mutated within TMS 3-We made point mutations at residues 137-145 and 147. Effects of some mutations at three of the residues (Gly 138 , Ser 139 , and Asn 143 ) have been reported previously (11). Conservative mutations of Gly 138 to an alanine, serine, or cysteine have resulted in low activity and were not reexamined here. Mutants S139A and N143V are inactive (11), and we examined the effects of additional mutations at position Asn 143 . Replacement of Asn 143 with a cysteine, alanine, glutamate, or serine also all produced inactive exchangers. Surprisingly, however, the mutant N143D produced an exchanger with normal Na ϩ affinity and activity level (Fig. 6). The N143D exchanger also displayed normal regulatory properties. Both Na ϩ -and Ca 2ϩ -dependent inactivations are demonstrated in Fig. 6C. Na ϩ -dependent inactivation is removed by high intracellular Ca 2ϩ , and activity disappears upon the removal of regulatory Ca 2ϩ as also occurs with the wild-type exchanger.
Mutation of residues Ala 140 , Phe 145 , and Ile 147 in the ␣-1 region of TMS 3 caused substantial alterations in the apparent affinity for intracellular Na ϩ . As shown in Fig. 7A, mutation of alanine at position 140 or isoleucine at position 147 decreased the Na ϩ affinity of the exchanger. In contrast, mutant exchanger F145C had an increased Na ϩ affinity. The apparent Na ϩ affinities were: 37.1 Ϯ 2.6 mM (n ϭ 5) for A140C, 9.9 Ϯ 0.9 mM (n ϭ 6) for F145C and 88.5 Ϯ 2.3 mM (n ϭ 4) for I147S. The value for the wild-type exchanger was 16.6 Ϯ 1.2 mM (n ϭ 6).
The extent of the decrease in the Na ϩ affinity caused by the replacement of alanine 140 correlated with the size of the residue inserted (Figs. 7B and 8). The smallest residue, glycine, yielded the highest affinity for Na ϩ (13.7 mM) and the largest, asparagine, yielded the lowest affinity (60.2 mM).
Replacement of phenylalanine 145 produced two effects: an increase in the apparent Na ϩ affinity and a decrease in the Hill coefficient for both Na ϩ and Ca 2ϩ . The increase in Na ϩ affinity was prominent in mutants F145C and F145L and did not correlate with either hydrophobicity or size of the introduced residue (see Figs. 7C and 8). The shift in the shape of the Na ϩ dependence curve was prominent in mutants F145C and F145S and may correlate with either hydrophobicity or size. Serine   FIG. 6. Role of TMS 3 in ion transport. Intramembrane residues located in TMS 3 were mutated to determine their function in ion transport. As shown in panel A, mutations V137C, A141C, N143D, and M144C did not modify the cytoplasmic Na ϩ dependence of the Na ϩ -Ca 2ϩ exchanger. Substitution of asparagine at position 143 with alanine (n ϭ 2), cysteine (n ϭ 2), glutamate (n ϭ 1), serine (n ϭ 1), or valine (n ϭ 10) abolished ion transport (B) when measured as 45 Ca 2ϩ uptake and only substitution of an aspartate residue maintained exchange activity. Mutant N143D generated outward exchange current with regulatory properties similar to wild type (C) .   FIG. 7. Replacement of residues Ala 140 , Phe 145 , or Ile 147 of TMS 3 affects cytoplasmic Na ؉ dependence. Panel A shows dose response curves for cytoplasmic Na ϩ for the wild-type exchanger and the mutants A140C, F145C, and I147S. Mutations A140C and I147S decreased cytoplasmic Na ϩ affinity while mutation F145C increased the affinity for Na ϩ and modified the slope of the curve. Panel B shows the Na ϩ dependence curves for mutants A140G, A140C, and A140N. Panel C illustrates Na ϩ dependence curves for mutations at position 145. Normalized currents versus cytoplasmic [Ca 2ϩ ] for wild-type, A140C, F145C, and I147C exchangers are shown in panel D. Small shifts in Ca 2ϩ titration curves are not significant. and cysteine are of comparable size and are smaller than leucine or phenylalanine and are also both less hydrophobic than leucine or phenylalanine. Thus, the slope of the Na ϩ dependence may be set by either the size or the hydrophobicity of the amino acid at position 145. A mutation at a nearby position, F142C, also had a decreased Hill coefficient for Na ϩ activation though, in this case, apparent Na ϩ affinity was unchanged (Fig. 8).
We also investigated the response of mutant exchangers A140C, F145C, and I147S to cytoplasmic Ca 2ϩ (Fig. 7D). We observed no significant shift in the apparent affinities for Ca 2ϩ . However, mutation of phenylalanine 145 to cysteine modestly reduced the Hill coefficient for the binding of cytoplasmic Ca 2ϩ ; Hill coefficients were 1.3 Ϯ 0.1 and 1.1 Ϯ 0.1 (n ϭ 5) for wild-type and F145C exchangers, respectively.

DISCUSSION
Since the cloning of the Na ϩ -Ca 2ϩ exchanger in 1990 (1), there has been much progress in understanding exchanger structure and function. Important regulatory regions have been identified (18,20), and a new topology has been proposed (2,3,5). The latest studies indicate a protein comprised of nine transmembrane segments with the initial five TMSs separated from the last four TMSs by a large cytoplasmic loop. The hydrophilic linkers connecting TMSs 2 and 3 and TMSs 7 and 8 have been proposed to form reentrant loops as some residues are accessible to sulfhydryl reagents applied to either surface of the membrane (2,5). These reentrant loops are on opposite sides of the membrane (Fig. 1) and are in proximity to one another (10). These two linkers are parts of the ␣ repeats, homologous sequences within the exchanger of about 40 amino acids each, including parts of TMSs 2 and 3 (␣-1 repeat) and a portion of TMS 7 (␣-2 repeat). Reentrant membrane loops (Ploops) were first described in ion channels where they form a portion of the channel pore. By analogy, the reentrant loops found in the Na ϩ -Ca 2ϩ exchanger family may be important for ion transport. In support of this, a previous report described four point mutations within the hydrophilic portion of the ␣-1 reentrant loop (H124C, N125C, G129C, and D130C) that caused up to a 4-fold reduction in the apparent affinity for external Ca 2ϩ (5). In this study, we further investigated the role of the first reentrant loop in ion translocation and extended our studies to residues forming the first portion of TMS 3.
Our data demonstrate that amino acids encompassing the extracellular linker between TMSs 2 and 3 are replaceable and that substitutions do not obstruct the transport of ions. Mutant exchangers H124N, F126C, and D130C and the squid chimera exchanger (containing the changes C122V, H124Q, N125K, T127E, and D130Q) showed modest change in intracellular Na ϩ affinity, though intracellular Ca 2ϩ affinities were unchanged. These data tend to rule out the first reentrant loop as a key component of the ion binding and translocation pathway. This contrasts with ion channel P loops that form the ion selectivity filter within the pore pathway. Consistent with our results is the relatively poor conservation of residues within the reentrant loop. One would expect a critically important reentrant loop to demonstrate the high degree of conservation seen in the adjoining TMSs 2 and 3 (Fig. 1). The lack of effect of the nonconservative changes in the squid chimera (e.g. H124Q, N125K, and D130Q) argues against a key role for the reentrant loop.
The most prominent effect of reentrant loop mutations was on the cooperativity of Na ϩ dependence. With the exception of the N125C, all mutations within the reentrant loop decreased the Hill coefficient of the Na ϩ dependence, indicating a change in cooperativity. Indeed, the Hill coefficient decreased from the wild-type value of 2 to a value much closer to 1 for the H124N, F126C, and squid chimera exchangers. The data suggest that these residues are involved in the interactions between Na ϩ binding sites.
We found that reentrant loop mutations had little effect on either the cytoplasmic or extracellular Ca 2ϩ dependencies. This is surprising since others have reported that mutations at position 124, 125, and 130 in the reentrant loop decreased the affinity for external Ca 2ϩ (5). We do not have an explanation for this apparent discrepancy.
Cytoplasmic Na ϩ is transported by the Na ϩ -Ca 2ϩ exchanger but also regulates exchange activity. High concentrations of internal Na ϩ force the exchanger into a nonconductive state (Na ϩ -dependent inactivation) (16). The molecular mechanisms that lead to the Na ϩ -dependent inactivation are not known.
However, it appears that Na ϩ bound to the internal transport sites triggers the process (16). Interestingly, two mutants within the first reentrant loop of the exchanger (H124N and D130C) show a slower decay of exchange current in the presence of 100 mM cytoplasmic Na ϩ . The data suggest that residues located within the first reentrant loop contribute to regulation and their replacement slows down conformational changes involved in the Na ϩ -dependent inactivation process. This behavior is reminiscent of C-type inactivation found in some voltage-gated ion channels (21,22). C-type inactivation involves a slow movement of the protein, leading to closure of the outer pore over several seconds. Mutations within the outer pore of the channel alter inactivation (23)(24)(25). By analogy, cytoplasmic Na ϩ may trigger a slow conformational change of the reentrant-loop region of the exchanger and modifications of this region slow down decay of transport function.
The functional importance of TMS 2 and the linker between TMS 2 and 3 has been extensively investigated through mutagenesis studies (2,4,5,11,13) with results emphasizing the critical role of TMS 2 in ion translocation. In contrast, TMS 3 has received relatively little attention. Nicoll et al. (11) showed that exchange activity is highly sensitive to mutation of Gly 138 , Ser 139 , or Asn 143 within TMS 3. Also, recent data place this portion of the exchanger in contact with TMSs 2, 7, and 8 to form a three-dimensional structure that may define a pathway for ion movement (10). These data and the high conservation of this region suggest an involvement in the movement of ions. To examine whether TMS 3 plays a role in ion selectivity and transport, we investigated the effects of mutations within TMS 3. We mutated each of the residues belonging to the ␣ repeat portion of TMS 3 and measured the effects of these mutations on Na ϩ and Ca 2ϩ translocation. Several mutations produced prominent effects. First, as reported previously (11), mutants G138A and G138S have low activity and S139A has no activity. Of the active mutant exchangers, replacement of alanine 140, phenylalanine 145, and isoleucine 147 produced the most prominent changes in the biophysical properties of the Na ϩ -Ca 2ϩ exchanger. Mutations at these sites altered the apparent affinity of the exchanger for Na ϩ , with the most prominent effect at position 147. The effects of mutations at position 140 may be sensitive to the size of the replacement residue. Substitution with a large amino acid (asparagine) produced the largest decrease in the Na ϩ dependence, while insertion of a small residue (glycine) did not modify transport. There is no simple pattern to explain the effects of mutations at position 145.
Since Ca 2ϩ and Na ϩ are thought to bind at the same sites, mutations that alter Na ϩ dependence might also be expected to modify the apparent affinity for transported Ca 2ϩ . However, the apparent affinity for Ca 2ϩ of mutant exchangers A140C, F145C, and I147S was not significantly different from that measured for the wild type exchanger. It has previously been proposed (26,27) that the Na ϩ -Ca 2ϩ exchanger has two classes of cation binding sites. One class of sites binds either Ca 2ϩ or two Na ϩ and the second class binds one Na ϩ . Residues 140, 145, and 147 may be involved in the second class of sites, and thus mutations affect Na ϩ , but not Ca 2ϩ , affinity. There is little previous information on the location of Na ϩ binding sites within the exchanger. The only TMS mutation previously reported to alter Na ϩ affinity is T103V, near the intracellular surface of TMS2 (13). In this case, the cooperativity of Na ϩ transport was also reduced.
Particular attention must be given to the asparagine at position 143 in TMS 3. Replacement of this residue with an alanine, cysteine, serine, valine, or glutamate residue totally abolished exchange activity. However, replacement of Asn 143 with an aspartate residue produced an exchanger indistin-guishable from wild type. It is quite striking that introduction of an aspartate leaves properties unaltered while introduction of any of five other residues abolishes activity. The sensitivity of this position to mutations may indicate the direct involvement of this residue in ion transport and also indicates a strict requirement for specific amino acids (asparagine or aspartate) at this site. Aspartate, asparagine, and valine are all of similar size but only aspartate and asparagine are hydrophilic. Thus, not only size but also degree of hydrophilicity at this position is essential in maintaining exchanger activity. It may seem surprising that introduction of a negatively charged aspartate residue into the center of a transmembrane segment does not disrupt transport function. However, in a hydrophobic environment this aspartate may remain in an undissociated and uncharged state.
The three Na ϩ -Ca 2ϩ exchanger mutations in TMS 3 that do not perturb wild type properties (A141C, M144C, and V137A) can all be modeled to be on the same face of a transmembrane ␣-helix. All the remaining mutations that alter exchanger function are excluded from this face. These sites may face TMSs 2, 7, and 8 that have been modeled to interact with TMS 3 and form an ion translocation pathway (10).
Taken together, our data indicate that residues forming the first reentrant membrane loop of the Na ϩ -Ca 2ϩ exchanger do not have a key role in ion transport but could be involved in movements associated with Na ϩ -dependent inactivation and in cooperativity between Na ϩ binding sites. In contrast, at least six amino acid residues within the extracellular half of TMS 3 are important components of exchanger function and some may be involved in the binding of Na ϩ . The Na ϩ binding site associated with TMS 3 may be a unique site that does not participate in Ca 2ϩ binding. The data confirm the importance of TMS 3 and the ␣-1 repeat in exchanger function.