Lithium-calcium exchange is mediated by a distinct potassium-independent sodium-calcium exchanger.

Sodium-calcium exchangers have long been considered inert with respect to monovalent cations such as lithium, choline, and N-methyl-d-glucamine. A key question that has remained unsolved is how despite this, Li(+) catalyzes calcium exchange in mammalian tissues. Here we report that a Na(+)/Ca(2+) exchanger, NCLX cloned from human cells (known as FLJ22233), is distinct from both known forms of the exchanger, NCX and NCKX in structure and kinetics. Surprisingly, NCLX catalyzes active Li(+)/Ca(2+) exchange, thereby explaining the exchange of these ions in mammalian tissues. The NCLX protein, detected as both 70- and 55-KDa polypeptides, is highly expressed in rat pancreas, skeletal muscle, and stomach. We demonstrate, moreover, that NCLX is a K(+)-independent exchanger that catalyzes Ca(2+) flux at a rate comparable with NCX1 but without promoting Na(+)/Ba(2+) exchange. The activity of NCLX is strongly inhibited by zinc, although it does not transport this cation. NCLX activity is only partially inhibited by the NCX inhibitor, KB-R7943. Our results provide a cogent explanation for a fundamental question. How can Li(+) promote Ca(2+) exchange whereas the known exchangers are inert to Li(+) ions? Identification of this novel member of the Na(+)/Ca(2+) superfamily, with distinct characteristics, including the ability to transport Li(+), may provide an explanation for this phenomenon.

Plasma membrane Na ϩ /Ca 2ϩ exchange is an important element in cellular Ca 2ϩ homeostasis. It has been extensively investigated in mammals, especially in cardiac and neuronal tissues (1), where this mechanism is essential for regulation of Ca 2ϩ homeostasis. Na ϩ /Ca 2ϩ exchange also plays a key role in many other organs and tissues by modulating the intracellular Ca 2ϩ response (2). The Na ϩ /Ca 2ϩ exchangers described to date are members of four major families (1). Of these, only NCX1-3 and NCKX1-4 are found in mammalian cells, whereas the other two are expressed in plants, yeast, and bacteria.
The stoichiometry of NCX exchangers is based on 3-4Na ϩ / 1Ca 2ϩ , and the protein is structurally organized as nine trans-membrane helixes divided by a large cytoplasmic loop (3). NCKX family proteins have a stoichiometry of 4Na/1Ca/1K, and their topology is thought to consist of two sets of five membrane helixes divided by a cytoplasmic loop and NH 2terminal extracellular domain (4). The two families share several important functional and structural motifs: a structural hallmark of the Na ϩ /Ca 2ϩ exchanger superfamily is the presence of two regions of sequence similarity, called ␣1 and ␣2, which are considered necessary and sufficient for generating exchange activity (1,5). Functionally, these exchangers are considered catalytically inert to inorganic and organic monovalent cations such as Li ϩ , NMG ϩ , 1 and choline ϩ , which are often used in "sodium-free" solutions to reverse Na ϩ /Ca 2ϩ exchange activity (2).
Although Li ϩ ions are not transported by either NCX or NCKX, numerous physiological studies have reported a significant effect of this ion on Ca 2ϩ exchange (6 -8). Earlier studies, conducted on squid axons and barnacle muscle, observed an inhibitory effect on intracellular Li ϩ , whereas extracellular Li ϩ accelerated 45 Ca efflux (9,10). These studies, therefore, suggested that Li ϩ may interact with a regulatory site on Na ϩ / Ca 2ϩ exchangers (2). More recent studies, conducted on striated muscle, have suggested that Li ϩ may catalytically participate in Li ϩ /Ca 2ϩ exchange mediated by an unidentified exchanger that is distinct from NCX and NCKX (11).
During our search for candidates for the Na ϩ /Zn 2ϩ exchanger gene (12), we cloned the full open reading frame of the FLJ22233 gene from HEK293 cells and analyzed its activity with respect to ion transport. While this work was in progress, it was reported (13) that the full-length FLJ22233 protein, heterologously expressed in HEK293 cells, is retained in the endoplasmic reticulum, and therefore, is non-functional. A mouse spliced isoform with a disrupted ␣2 repeat was shown to mediate active K ϩ -dependent Na ϩ /Ca 2ϩ exchange. Our findings, however, indicate that the full-length protein encoded by the human FLJ22233 gene, is in fact, functional, mediating Na ϩ /Ca 2ϩ exchange independent of K ϩ . We further demonstrate that FLJ22233 is distinct from both NCX and NCKX in its ability to catalyze Li ϩ /Ca 2ϩ exchange. The protein is highly selective for Ca 2ϩ and does not catalyze either Ba 2ϩ or Zn 2ϩ transport. We therefore propose that FLJ22333, named here NCLX, is a novel member of the Na ϩ /Ca 2ϩ exchanger family with distinct ion selectivity.

EXPERIMENTAL PROCEDURES
Cloning of the Human NCLX-The conserved sequences of ␣1 and ␣2 repeats from AtMHX1 and from six different members of the human Na ϩ /Ca 2ϩ exchanger superfamily (NCX1-3, NCKX1-3) were used to * This work was supported by German-Israeli Foundation for Scientific Research and Development Grant 10588099.01/98, ISF Grant 456/ 02.1 (to I. S.), and Israel Science Foundation equipment Grant 456/02.2 (to I. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AY601759 (NCLX) and AY601760 (S-NCLX).
screen the human genome sequence protein data base using BLAST software (14). This search revealed a hypothetical protein, FLJ22233 (GenBank TM accession number NP_079235). Based on the NCBI assembly sequence annotation (using RefSeq, at www.ncbi.nlm.nih.gov/IEB/Research/Acembly), one set of primers was designed to enhance the longest deduced open reading frame of this gene using reverse transcriptase-PCR. Total RNA was extracted from HEK293 cells using RNeasy Midi kit (Qiagen), and mRNA was further purified using Poly Tract mRNA isolation system III (Promega). Reverse transcriptase-PCR was performed with a one-step reverse transcriptase-PCR Kit (Qiagen) using 200 ng of mRNA and the following primers: (forward primer) 5Ј-CCAGCTGTTTGGAACTGAGC-3Ј and (reverse primer) 5Ј-TCACATGCTTTTCAGGTGAATCACTCCAA-3Ј.
Two transcripts of 1656 and 1824 bp were amplified and ligated into a pGEM-T cloning vector. Both strands of the two clones were sequenced (Sequencing Unit, Biological Core Facility at the Institute for Applied Sciences at Ben-Gurion University). The long and short transcripts were excised from pGEM-T vector (Promega), ligated into a pCDNA3.1ϩ vector using EcoRI endonuclease, and termed NCLX and s-NCLX for the full-length and the shorter isoform, respectively. Two opposite orientations for each transcript were selected using the reverse-oriented transcript as control. Plasmids were purified using the Hi-Speed plasmid purification kit (Qiagen).
Cell Cultures and Plasmid Transfection-HEK293-T cells (human embryonic kidney cell line) were cultured in Dulbecco's modified Eagle's medium as described previously (15). Briefly, cells grown on glass coverslips were transfected with 8 g of NCLX, control, or rat NCX1 (kindly provided by Dr. Hanna Rahamimoff, The Hebrew University, Jerusalem, Israel) using standard calcium phosphate (CaPO 4 ) precipitation as described previously (12). The EYFP (0.35 g) plasmid (Clontech) was added as a fluorescent reporter for the identification of transfected cells. Transport experiments or cell harvesting were carried out 25-35 h following transfection.
Fluorescent Imaging of Ion Transport-Two imaging systems were used. The first consisted of a Zeiss Axiovert 100 microscope (Zeiss) and the second consisted of an Olympus IX (Olympus) inverted microscope, Polychrome II monochromator (T.I.L.L. Photonics) equipped with a cooled CCD camera (PCO Imaging) as described previously (15). Fluorescent imaging measurements were acquired using Axon Imaging Workbench 2 software (Axon Instruments, Foster City, CA). HEK293-T cells grown on coverslips were incubated for 30 min with 5 M Fura-2 AM (TEF Labs, Austin, TX) in 0.1% bovine serum albumin in sodium Ringer's solution (140 mM NaCl, 10 mM HEPES, 20 mM glucose, 0.8 mM MgCl, 0.5 mM CaCl, at pH 7.4). Following dye loading, the cells were washed in sodium Ringer's solution, and the coverslips were mounted in a chamber that allowed the perfusion of cells. Fura-2 was excited at 340-and 380-nm wavelength light and imaged with a 510-nm-long pass filter.
Influx assays were performed by perfusing the cells with Na ϩ -free Ringer's solution such that Na ϩ was iso-osmotically replaced by NMG ϩ with or without 5 mM KCl. Active calcium efflux was monitored by first loading the cells with Ca 2ϩ by perfusion with the NMG ϩ or choline ϩreversal medium and then switching to a Na ϩ -or Na ϩ -free (NMG ϩ -, Li ϩ -, or choline ϩ -containing) Ringer's solution. Zinc and barium transport mediated by the Na ϩ /Ca 2ϩ exchanger was monitored using the influx assay using NMG ϩ -reversal medium in which Ca 2ϩ was replaced with 400 M Zn 2ϩ or 1 mM Ba 2ϩ . The effect of inhibitors on calcium influx rate mediated by the Na ϩ /Ca 2ϩ exchanger was assessed in the same manner while adding different inhibitors to Ca 2ϩ -containing NMG ϩ Ringer's solution. The results shown are the means of 5-6 independent experiments, with averaged responses from 20 -40 cells in each experiment.
Generation of NCLX Antibody-The peptide CPVTPEILSDSEEDR, located at the start of the putative cytoplasmic loop of NCLX, was synthesized in the Peptide Core Facility at the Weizmann Institute of Science. The peptide (4 mg) was conjugated to maleimide KLH (Pierce) according to the manufacturer's instructions and used to immunize New Zealand White rabbits according to standard procedures (16).
Sample Preparation and Immunoblot Analysis-Rats were deeply anesthetized and killed according to the protocol approved by the Committee for the Ethical Care and Use of Animal in Experiments at the Faculty of Health Sciences at Ben-Gurion University. Appropriate tissues were placed in cold lysis buffer (20 mM Tris-Cl (pH 7.5), 0.32 M sucrose, 0.2 mM EDTA, and 0.5 mM EGTA with a protease inhibitor mixture (Roche Applied Science) and briefly sonicated three times using a tip-sonicator for 30 s (17). After brief centrifugation, supernatants were collected and frozen at Ϫ70°C until use.
HEK293-T cells expressing NCLX or control were washed once with ice-cold phosphate-buffered saline solution, scraped from the plate using a rubber policeman, and placed into radioimmune precipitation buffer. Protein was quantified by the Bradford dye binding procedure (Bio-Rad). Equal amounts of protein were resolved by SDS-PAGE and transferred onto nitrocellulose membranes. Immunoblot analysis was carried as described previously (17), using the NCLX-antiserum (or the pre-immune serum) at a 1/2500 dilution.

RESULTS
Na ϩ /Ca 2ϩ Exchange Mediated by NCLX-We have cloned the FLJ22233 gene during our search for cation transporters expressed in HEK293 cells (12). The complete open reading frame was obtained and inserted into a pCDNA-3 mammalian expression vector. The activity of this putative transporter was determined by monitoring intracellular Ca 2ϩ changes in HEK293-T cells heterologously expressing the FLJ22233 gene product, loaded with Fura-2 AM. Cells were superfused with Ringer's solution, which was replaced with NMG ϩ Ringer's solution (NMG ϩ iso-osmotically substituting for Na ϩ ), as shown in Fig. 1. The removal of Na ϩ was followed by robust Ca 2ϩ influx. The subsequent addition of Na ϩ into the Ringer's solution induced a rapid efflux of Ca 2ϩ . This Na ϩ /Ca 2ϩ exchange activity was not monitored in control HEK293-T cells (transfected with vector containing a reverse-oriented insert). The exchange activity was compared with the activity mediated by NCX1. For both exchangers, a significant Ca 2ϩ influx was monitored at the absence of Na ϩ , which was replaced by NMG ϩ , and active Ca 2ϩ efflux was subsequently monitored when Na ϩ was reintroduced. Our results indicate that the full-length NCLX codes for a functional Na ϩ /Ca 2ϩ exchanger.
K ϩ Dependence of NCLX-To determine the K ϩ dependence of NCLX, we applied the paradigm described in Fig. 1 while using either choline ϩ or NMG ϩ (Fig. 2) in the reversal solution to verify that the exchange activity is not affected non-specifically by either NMG ϩ or choline ϩ . If the exchanger is K ϩ -dependent, a robust reduction in the Ca 2ϩ influx rate in the absence of K ϩ would be expected using the reversal mode. As shown in Fig. 2, rates of Ca 2ϩ influx were similar in the presence or absence of K ϩ , using either choline ϩ or NMG ϩ . Our results, therefore, indicate that the Na ϩ /Ca 2ϩ exchange mediated by NCLX is not K ϩ -dependent.
Li ϩ /Ca 2ϩ Exchange Mediated by NCLX-Lithium has often been used as an inert ion to study the reversal of Na ϩ /Ca 2ϩ exchangers. We have, therefore, employed Li ϩ as an additional control for the activity of the NCLX as compared with NCX1, again employing the same paradigm as shown in Fig. 1. In NCX1-expressing cells, an enhanced Ca 2ϩ influx was observed, consistent with the reversal of the exchanger activity. Unex- HEK293-T cells transfected with either NCLX or rat NCX1 plasmids were loaded with Fura-2 and superfused with the indicated Ringer's solution. The influx and efflux of Ca 2ϩ in cells expressing either NCX1 or NCLX were monitored, whereas Ca 2ϩ transport in the control cells was only residual.
pectedly, NCLX-expressing cells showed a much slower rate of Ca 2ϩ influx in the presence of Li ϩ (Fig. 3). This striking difference between the two exchangers suggests that although NCX1 is inert to Li ϩ transport, this ion is either inhibitory to Na ϩ / Ca 2ϩ exchange or transported by NCLX.
To distinguish between the two mechanisms, Ca 2ϩ was loaded into cells expressing NCLX using NMG ϩ -Ringer's, and the rate of active Ca 2ϩ efflux was compared in the presence of Na ϩ , NMG ϩ , or Li ϩ . As shown in Fig. 4, whereas rates of Ca 2ϩ efflux were somewhat slower in the presence of Li ϩ as compared with Na ϩ (1.6 Ϯ 0.5-fold), Li ϩ ions were strikingly more effective (8 Ϯ 1-fold) as compared with NMG ϩ in promoting active Ca 2ϩ efflux. The small, residual efflux activity observed in the presence of NMG ϩ may be related to the activity of the sarco(endo)plasmic reticulum Ca 2ϩ -ATPase (SERCA). Taken together, the failure of Li ϩ to promote Ca 2ϩ influx in the reverse mode on the one hand and the acceleration of efflux in cells loaded with Ca 2ϩ on the other indicate that Li ϩ is transported by NCLX. Thus, this exchanger can mediate either Li ϩ or Na ϩ calcium exchange.
Cation Dose Dependence of the Exchange Mediated by NCLX-The Ca 2ϩ influx dose dependence of NCLX was determined using the reversal mode of the exchanger while changing FIG. 2. NCLX is mediating K ؉ -independent Na ؉ /Ca 2؉ exchange. The same experimental paradigm shown in Fig. 1 was applied using K ϩ -free Ringer's. Reversal activity was monitored in the presence of NMG ϩ (a) or choline ϩ (b) to verify that the effect is independent of the cation replacing Na ϩ . Similar activity is also observed with 5 mM K ϩ in the presence of NMG ϩ (as shown on Fig. 1) or choline ϩ (c). Thus, NCLX activity in the absence or presence of K ϩ is similar.  4. NCLX is mediating Li ؉ /Ca 2؉ exchange. Cells expressing NCLX were loaded with Ca 2ϩ using NMG ϩ reversal Ringer's and then were superfused with NMG ϩ -, Li ϩ -, or Na ϩ -containing Ringer's. Calcium efflux was greatly enhanced following replacement of NMG ϩ with either Li ϩ or Na ϩ . The similarity between the efflux rates in the presence of Li ϩ or Na ϩ indicates that NCLX catalyzes Li ϩ /Ca 2ϩ exchange. extracellular Ca 2ϩ concentration and yielded an apparent Hill coefficient of 1.5 ϩ-0.4 and an apparent affinity of 0.8 Ϯ 0.2 mM for Ca 2ϩ , well within the physiological range (Fig. 5a). To determine the Na ϩ dependence of Ca 2ϩ efflux mediated by NCLX-expressing cells, the cells were first loaded with Ca 2ϩ , using the reversal of NCLX, and then extracellular Na ϩ was gradually iso-osmotically replaced with NMG ϩ , and the rates of Ca 2ϩ efflux were monitored. As shown in Fig. 5b, Ca 2ϩ efflux rates increased with external Na ϩ concentrations, fitting a Michaelis-Menten equation with an apparent half-maximal rate at 60 Ϯ 5 mM Na ϩ and a 3.5 Ϯ 1 Hill coefficient. Similar to other members of the NCX and NCKX family, the NCLX exchanges 3-4 Na ϩ ions per 1 Ca 2ϩ ion. The apparent affinities of both extracellular Ca 2ϩ (influx mode) and Na ϩ (efflux mode) for NCLX are also within the range previously monitored for NCX1 (2).
Since Li ϩ was transported by NCLX in the absence of Na ϩ , we further studied whether Li ϩ will functionally replace Na ϩ at physiological concentrations. The same experimental paradigm described for Na ϩ /NMG ϩ replacement was applied for Na ϩ /Li ϩ . Remarkably, only a slight change in Ca 2ϩ efflux rate is observed over the whole range, beginning with 140 mM Na ϩ and gradually changing to 140 mM Li ϩ . Our results clearly indicate that Li ϩ and Na ϩ act additively in catalyzing Ca 2ϩ efflux mediated by NCLX.
As the Na ϩ -dependent exchange activity of other cations such as Ba 2ϩ , Mg 2ϩ , and Zn 2ϩ has been previously functionally described (12,18), we sought to determine the cation selectivity of the NCLX in its reversal mode. Rates of Ba 2ϩ or Zn 2ϩ transport by NCLX are negligible (not shown), indicating that at least with respect to these cations, NCLX activity is specific to Ca 2ϩ .
Inhibitors for NCLX Activity-The use of inhibitors is an important tool for the functional identification and character-  b, cells were loaded with Ca 2ϩ , yielding a similar fluorescence level, using the reversal of NCLX. Subsequently, extracellular Na ϩ was gradually iso-osmotically replaced with NMG ϩ or Li ϩ (the top axis is Li ϩ concentration), and the rates of Ca 2ϩ efflux are presented depending on the Na ϩ concentration. Ca 2ϩ efflux rates decreased when external Na ϩ was replaced with NMG ϩ , and the line is a fit to a Michaelis-Menten equation yielding an apparent half-maximal rate at 60 Ϯ 5 mM Na ϩ and a 3.5 Ϯ 1 Hill coefficient. Only minor changes in Ca 2ϩ efflux rate are monitored when Na ϩ is replaced with Li ϩ , indicating that Li ϩ and Na ϩ act additively in catalyzing Ca 2ϩ efflux.
ization of an ion exchanger. We have, therefore, studied the sensitivity of the NCLX to various inhibitors of cation transporters. As shown in Fig. 6, a-c, the application of KB-R7943, a potent inhibitor of NCX with K1 ⁄2 of 10 M (19), resulted only in partial inhibition of NCLX activity to ϳ40% at a KB-R7943 concentration of 50 M, and a further increase in KB-R7943 concentration to 100 M did not result in additional inhibition of the exchange activity. Zinc ions, which are known to inhibit NCX1 non-selectively (using 30 M (20)), were slightly more effective in inhibiting NCLX activity with maximal inhibition of ϳ75% at 200 M and an apparent half-maximal inhibition at ϳ10 M. In contrast, even a high concentration of verapamil (100 M), a Ca 2ϩ antagonist that efficiently blocks Ca 2ϩ channels, had a weak effect on NCLX activity, reducing Ca 2ϩ influx rate by only ϳ20% as compared with that of control.
NCLX Short Splice Isoform-We cloned an additional splice isoform of the FLJ22233 gene from HEK293 cells, a humanderived cell line, referred to here as s-NCLX (short-NCLX isoform, see Fig. 7, inset). This splice isoform lacks exon 7, resulting in a deletion of 54 amino acids in the primary human sequence and a predicted molecular mass based on the primary amino sequence of 58 KDa, whereas the full-length NCLX gene has a predicted molecular mass of 64 KDa. The region of deletion encompasses the putative transmembrane domains 3 and 4 in NCLX. Since the domains truncated in this splice isoform have been shown to play a role in ion transport (13), we compared the activity, K ϩ dependence, and Li ϩ transport of this splice isoform to the full-length NCLX. The deletion did not change s-NCLX activity as compared with that of the fulllength NCLX in all of these aspects (Fig. 7).
NCLX Tissue Distribution-To determine the tissue distribution of NCLX, we generated a polyclonal antibody using a peptide derived from the putative intracellular domain specific to NCLX and not shared by known members of the NCKX or NCX protein families. Western blot analysis was performed using protein extracts from HEK293-T cells, heterologously expressing s-NCLX or NCLX (full-length NCLX). The NCLX antibody specifically stained a major band of ϳ70 KDa and a weaker one at ϳ55 KDa (Fig. 8a). Protein extract from control cells did not react with the antibody. These results indicate that the antibody specifically recognizes the full-length NCLX and s-NCLX. The small shift in the molecular weight of s-NCLX is consistent with the small truncation of ϳ50 amino acids.
Finally, as our antibody was designed to cross-react with the rat full-length and short isoforms of NCLX, we analyzed tissues from rat for the presence of the exchanger. Again, the NCLX antibody specifically stained two major bands, at ϳ70 and ϳ55 KDa (Fig. 8b). In their study, Cai et al. (13) reported that NCLX transcripts are ubiquitously expressed in rat tissues; our findings indicate that NCLX protein expression is more restricted. We show here that the reactivity is most intense on proteins from rat pancreas, skeletal muscle, and stomach. As a control, we also studied mouse heart extracts. In this tissue, both forms (ϳ70 and ϳ55-KDa) are observed, as has been noted previously (13). DISCUSSION Our findings indicate that NCLX is kinetically distinct from both the NCX and NCKX. Although both NCX and NCKX are The reversal of Ca 2ϩ exchange in control, NCLX (the same as the graph for NCLX in Figs. 1 and 2), or s-NCLX-expressing cells was monitored by substituting Na ϩ with choline ϩ without (b) or with (c) 5 mM K ϩ . Substituting Na ϩ with choline ϩ was followed by Ca 2ϩ influx into cells expressing s-NCLX (short isoform). Superfusion with Na ϩ -containing Ringer's solution was followed by Ca 2ϩ efflux. The short isoform of NCLX exhibits similar transport characteristics, in the presence or absence of K ϩ . These results indicate that, as is true of the full-length NCLX, the s-NCLX is a K ϩ -independent Na ϩ /Ca 2ϩ exchanger. d, cells expressing s-NCLX were loaded with Ca 2ϩ using NMG ϩ reversal Ringer's and then were superfused with NMG ϩ -, Li ϩ -, or Na ϩ -containing Ringer's while monitoring calcium efflux. The s-NCLX Li ϩ dependence is similar to that of the full-length NCLX. totally inert to Li ϩ transport, NCLX catalyzes Li ϩ /Ca 2ϩ exchange at a rate that is only slightly slower than Na ϩ /Ca 2ϩ exchange. Moreover, our data suggest that Li ϩ and Na ϩ act additively in catalyzing Ca 2ϩ exchange. In further contrast to NCX proteins, NCLX does not significantly catalyze Na ϩ /Ba 2ϩ exchange, indicating that in this aspect, it is similar to NCKX (4,21,22). However, although NCLX shares somewhat greater sequence similarity with NCKX (21% for NCKX1) than with NCX (11% for NCX1), Na ϩ /Ca 2ϩ exchange catalyzed by NCLX is not, as suggested previously, K ϩ -dependent (13). NCLX, unlike NCKX, is also partially sensitive to KB-R7943 (4,23). Key residues, highly conserved in both NCX and NCKX, are surprisingly different in the NCLX primary sequence. Most notable are changes in the ␣-repeat domains, suggesting alterations in the cation selectivity of this exchanger (1,13). Indeed, phylogenetic analysis of NCLX has revealed that it is somewhat more related to NCKX, although it has also diverged at an early stage from these proteins (13). Taken together with our functional analysis, we propose that NCLX is a novel Na ϩ /Ca 2ϩ exchanger with characteristics distinct from either the NCX or the NCKX families.
The full-length FLJ22233 gene product tagged with FLAG epitope at the NH 2 -terminal domain was described by Cai et al. (13) as non-functional when expressed in HEK293 cells. The use of Li ϩ as an inert cation for the reversal of the exchanger activity may have yielded an apparent non-functional phenotype. Immunoblot and histochemical analysis by Cai et al. (13) suggested that FLJ22233 is expressed in HEK293 cells as a 55-KDa protein that is retained in the endoplasmic reticulum.
We found, in contrast, that heterologous expression of NCLX in HEK293-T cells resulted in a fully functional protein. It is possible that the tagging by the FLAG epitope interfered with the proper sorting of the mouse gene described by Cai et al. (13). Immunoblot analysis indicated that NCLX migrated as a major band of 70 KDa in addition to a much weaker 55-KDa protein. In native tissue, however, both the study of Cai et al. (13) and our study have identified both forms. It may be possible, therefore, that post-translational modifications have yielded the differences in molecular weight and functionality between our respective studies (13). Another major difference between our findings is the K ϩ dependence of Ca 2ϩ exchange. Thus, we found that NCLX is K ϩ -independent; the spliced short mouse isoform studied by Cai et al. (13) was K ϩ -dependent and was inert to Li ϩ . This discrepancy may be linked to structural differences between the full-length human gene and the short mouse isoform, which has a putatively disrupted second ␣-repeat and a long intracellular tail. We did not find a similar isoform to the mouse in human. It remains to be seen whether these changes in the primary sequence of the mouse gene are converting this isoform to a K ϩ -dependent and Li ϩinert Na ϩ /Ca 2ϩ exchanger.
We were able to clone a human isoform lacking exon 7, which results in the deletion of the transmembrane domains 3 and 4. These domains are highly conserved in NCX and NCKX families. Although in NCKX2, mutations of several amino acids located in this region have resulted in the loss of activity (24), we find that the truncation of this region from NCLX, manifested in s-NCLX, had no effect on activity.
It was previously shown that although NCX1 does not transport Li ϩ , a single point mutation of Thr-103/Val-103 did activate Li ϩ /Ca 2ϩ exchange that was mediated by canine NCX1 (25). This threonine residue is conserved among all members of NCX, NCKX, and NCLX. Interestingly however, and only in NCLX, the amino acid preceding threonine is valine. Whether this specific residue or its domain is responsible for Li ϩ /Ca 2ϩ exchange, or whether other regions participate in Li ϩ /Ca 2ϩ exchange mediated by NCLX, is an intriguing and open question.
Although mRNAs for FLJ22233 are found in many tissues (13), the expression level of the protein varies markedly in these tissues. In rat, the protein is highly expressed in the pancreas, skeletal muscle, and stomach, and to lesser extent, in cardiac tissue, brain, spleen, and skin. In human, the protein is highly expressed in the brain and skeletal muscle, and low expression levels are observed in the pancreas (data not shown). The lower expression of NCLX in human tissues, especially in the pancreas, should be taken with a degree of caution, considering the high content of proteolytic enzymes that exist there and that may be activated in a cadaver. In the rat pancreas, we have found that NCLX is primarily found in the exocrine part (data not shown). In this regard, Na ϩ /Ca 2ϩ exchange in pancreatic ducts was reported previously to be more intense than in acini where the plasma membrane Ca 2ϩ pump is the primary transporter for Ca 2ϩ efflux (26). In mouse brain, NCX2 was recently shown to play a role in synaptic transmission, and specifically, in the context of learning and memory (27). The importance of Na ϩ /Ca 2ϩ exchange in skeletal muscle was recently demonstrated by the effect of NCX3 knockout on muscle activity (28).
The potential physiological importance of NCLX is highlighted by numerous reports on observed effects of Li ϩ on Ca 2ϩ exchange. Both voltage sensitivity and magnitude of Na ϩ /Ca 2ϩ exchange have been described as depending on the presence of Li ϩ as opposed to NMG ϩ (replacing Na ϩ ) (6, 7). A growing body of evidence, particularly in skeletal muscle, suggests that Li ϩ FIG. 8. Tissue distribution of NCLX. Immunoblot analysis using the NCLX antibody at a dilution of 1/2500 (see "Experimental Procedures") was conducted on HEK293-T cells expressing NCLX isoforms or control (a). In contrast to control cells, in cells heterologously expressing either isoform of NCLX, a major 70-KDa band was apparent, as well as a weaker one of 55 KDa. b, a similar analysis of the indicated (20 g) protein tissue extracts, in mouse (M) and rat (R) tissue, using NCLX antibody. No stain was monitored when the pre-immunserum was used (not shown). NCLX was expressed in two distinct molecular masses of 70 and 55 KDa. The expression of NCLX in the rat tissue was high in the pancreas, skeletal muscle, and stomach. Relative medium to low expression levels are present in the heart, brain, skin, smooth muscle, and spleen tissues.
ions may directly participate in Ca 2ϩ exchange (11). Since neither NCX nor NCXK family members transport Li ϩ , it has been conjectured that an as yet unidentified exchanger may be involved in this mode of transport. The Li ϩ transport mediated by NCLX, taken together with its high abundance in skeletal muscle, is consistent with the conclusion that NCLX is that mediator of Li ϩ /Ca 2ϩ exchange. Since Li ϩ /Ca 2ϩ exchange is slower than Na ϩ /Ca 2ϩ exchange mediated by NCLX, it may well be that the exchange activity of NCLX regulating intracellular Ca 2ϩ homeostasis is underestimated. Expression of NCLX in cardiac muscle may also explain the significant residual activity of Na ϩ /Ca 2ϩ exchange found in cardiomyocytes cultured from NCX1-null embryos (29). Future studies combining the silencing of specific members of the Na ϩ /Ca 2ϩ exchangers superfamily, together with functional studies comparing the role of Li ϩ or Na ϩ in Ca 2ϩ exchange, can now be undertaken to determine whether the activity of NCLX is related to this intriguing effect.