Structure-Based Engineering of Lithium-Transport Capacity in an

Rapid Report pubs.acs.org/biochemistry

Structure-Based Engineering of Lithium-Transport Capacity in an Archaeal Sodium−Calcium Exchanger Bosmat Refaeli, Moshe Giladi, Reuben Hiller, and Daniel Khananshvili* Department of Physiology and Pharmacology, Sackler School of Medicine, Tel-Aviv University, Ramat-Aviv, Tel-Aviv 69978, Israel S Supporting Information *

ABSTRACT: Members of the Ca2+/cation exchanger superfamily (Ca2+/CA) share structural similarities (including highly conserved ion-coordinating residues) while exhibiting differential selectivity for Ca2+, Na+, H+, K+, and Li+. The archaeal Na+/Ca2+ exchanger (NCX_Mj) and its mammalian orthologs are highly selective for Na+, whereas the mitochondrial ortholog (NCLX) can transport either Li+ or Na+ in exchange with Ca2+. Here, structure-based replacement of ion-coordinating residues in NCX_Mj resulted in a capacity for transporting either Na+ or Li+, similar to the case for NCLX. This engineered protein may serve as a model for elucidating the mechanisms underlying ion selectivity and ion-coupled alternating access in NCX and similar proteins.

U

nderstanding the structure−dynamic determinants by which membrane proteins (channels, pumps, transporters, and receptors) recognize and transport ions remains challenging for physical,1,2 chemical,3,4 and biological sciences.5−8 The progress in basic sciences3−8 may have practical implications in the fields of bionanotechnology, pharmaceutics, drug design and delivery, among others.3,9,10 For example, improving our understanding of Li+ binding and transport selectivity in proteomimicking systems may allow the achievement of superior performance of the next-generation Li+ batteries in handy electronic devices.9,10 The gene family of Na+/Ca2+ exchangers (NCX) is one of five families belonging to the Ca2+/CA exchanger (Ca2+/cation antiporter) superfamily.11,12 Membrane-bound NCX proteins utilize the electrochemical gradient of Na+ to extrude Ca2+ from the cell and, thereby, play a key role in handling Ca2+ homeostasis in health and disease.11,12 The gene products of the Ca2+/CA superfamily share similar structural attributes13−16 but have different selectivities for Ca2+, Na+, H+, K+, and Li+ transport and exhibit large differences in ion-transport kinetics with turnover rates of 1−5000 s−1 for the transport cycle.11,12,17 NCX proteins mediate electrogenic ion exchange (3Na+:1Ca2+),11,18 in which separate steps for Na+ and Ca2+ transport (a ping-pong mechanism)19 are associated with alternating access of the ion-binding domain at opposite sides of the membrane (Figure S1).13,21 The high-resolution crystal structure (1.9 Å) of Methanococcus jannaschii NCX (NCX_Mj) depicts the outward-facing (extracellular) conformation with four ion-binding sites (Sext, Smid, Sint, and SCa), including 12 ion-coordinating residues (Figure 1A).13 The Sint and Sext sites have high selectivity for © XXXX American Chemical Society

Figure 1. Ion-coordinating residues in NCX_Mj and mitochondrial NCX capable of lithium transport (NCLX). (A) Three-Na+ ion (purple spheres) coordination as suggested by MD simulations and flux assays and Ca2+ ion (green sphere) coordination as suggested by the crystal structure (Protein Data Bank entry 3V5U). Mutated residues in the NCX_Mj/NCLX construct are colored red. (B) Sequence alignment of the α-repeats of NCX_Mj and NCLX (UniProt entries Q57556 and Q6J4K2, respectively). Mutated ioncoordinating residues are colored red.

Na+, whereas Smid and SCa lack selectivity for either ion. Consistent with the ping-pong mechanism, the simultaneous occupation of all four sites by ions is thermodynamically forbidden. According to recent findings, three Na+ ions occupy Sext, Sint, and SCa and Ca2+ occupies SCa.20,21 According to the crystal structure of NCX_Mj, four residues coordinate Ca2+ at SCa,13 whereas a systematic mutational analysis of ioncoordinating residues has identified six residues (assigned to all four ion-binding sites) contributing to Ca2+-transport catalysis.21 These findings provide a basis for a more extended investigation of ion-transport mechanisms with respect to ion selectivity and ion-coupled alternating access. Received: February 11, 2016 Revised: March 7, 2016

A

DOI: 10.1021/acs.biochem.6b00119 Biochemistry XXXX, XXX, XXX−XXX

Biochemistry

Rapid Report

Sequence alignment of the α-repeats of NCX_Mj and NCLX reveals that at matching positions the two proteins share only three residues (A47, S210, and A206) of 12 ion-coordinating residues (Figure 1B). With the goal of imitating the ion-binding pocket of NCLX, a new construct (NCX_Mj/NCLX) was generated with nine replacements in NCX_Mj (Figure 1A,B) at matching positions (T209N, E54D, S77A, S236G, S51G, E213D, T50N, N81V, and D240N), and the capacity for Li+ and Na+ transport was tested in exchange with Ca2+. In these experiments, the Li+/Ca2+ and Na+/Ca2+ exchange reactions were analyzed by assaying the Na+i- or Li+i-dependent 45Ca2+ uptake, where Na+- or Li+-loaded vesicles were rapidly diluted in an assay medium containing 45Ca2+.17,18 In this assay, Escherichia coli-derived cell membrane vesicles containing the overexpressed wild-type NCX_Mj or its mutant (10−12% of the total membrane protein content) were used.17,20,21 In these preparations of vesicles, the overexpressed proteins exhibit a uniform orientation.17 NCX_Mj-containing vesicles exhibit a typical time course for Na+i-dependent 45Ca2+ uptake, but exhibit a negligible (if any) capacity for time-dependent 45Ca2+ uptake in exchange with Li+ (Figure 2A). In contrast, the NCX_Mj/NCLX construct exhibits a similar time-course for Nai+ or Lii+-dependent 45 Ca2+ uptake. Thus, comparable rates of Li+/Ca2+ and Na+/ Ca2+ exchange reactions are observed in the NCX_Mj/NCLX construct (Figure 2A), revealing a new ion-transport feature in the NCLX prototype of NCX_Mj. For evaluating the apparent affinity (Km) for Ca2+ transport, the initial rates of Na+i- or Li+i-dependent 45Ca2+ uptake were measured in NCX_Mj- or NCX_Mj/NCLX-containing vesicles, with varying extravesicular concentrations of 45CaCl2 at a fixed (saturating) concentration (160 mM) of intravesicular Na+i or Li+i (Figure 2B). Intriguingly, the Na+/Ca2+ exchange in the NCX_Mj vesicles or the Li+/Ca2+ and Na+/Ca2+ exchanges in the NCX_Mj/NCLX vesicles exhibit comparable Km values for Ca2+ (70−120 μM), whereas the Vmax values differ by ∼2fold. Thus, NCX_Mj and its NCLX “prototype” share a similar apparent affinity for Ca2+ transport in exchange with either Na+ or Li+ (Table S1). This is interesting in light of the similar Km and Kd values observed in NCX_Mj, suggesting that ion translocation (and not ion binding) is rate-limiting throughout the transport cycle.17,21 Notably, in NCX_Mj, four residues (T50, E54, T209, and E213) coordinating the Ca2+ ion at SCa completely differ from four residues [N149 (T50), D153 (E54), D471 (E213), and N467 (T209)] located at matching positions in NCLX (Figure 1A,B). Strikingly, this “modified” rearrangement of ioncoordinating residues at the SCa site in NCLX/NCX_Mj involves the replacement of two negatively charged residues (E) with shorter side-chain residues (D), whereas two carbonylcoordinating residues (T) are replaced with bulkier (N) sidechain residues (Figure 1B). Although this “structural remodeling” at SCa has no significant effect on the apparent affinity of Ca2+ binding/transport (Figure 2B), the relevant structural rearrangements could be relevant for changing the Na+:Li+ selectivity (Table S1). This is especially interesting in terms of the capacity of SCa to bind either Na+ or Ca2+.20,21 At this end, the actual changes in the ligation chemistry and coordination radius (among others) and how these changes affect the affinity, selectivity, and stoichiometry of Na+ or Li+ binding and/or transport are unclear. High-resolution crystal structures of ion-bound proteins and detailed functional

Prokaryotic and eukaryotic NCXs contain 10 transmembrane helices (TM1−TM10), where two hubs (TM1−TM5 and TM6−TM10) with inverted 2-fold “pseudosymmetry” are linked through a cytosolic loop of a different length.11−13 Two long and tilted helices (TM1 and TM6) are loosely packed in front of a rigid eight-helix core (TM2−TM5 and TM7− TM10).13 It has been proposed that the “sliding” of the TM1/ TM6 cluster toward the rigid eight-helix core accounts for a major conformational change associated with alternating access.13 Recently elucidated structural similarities between the Ca2+/H+ exchangers (CAX)14−16 and NCX_Mj13 suggest that Ca2+/CA proteins may share a common mechanism for two-helix sliding, although how the ion interactions with specific sites induce conformational changes associated with alternating access and how these interactions are related to the ion selectivity in a given protein remain unclear. In contrast with mammalian NCXs, NCX_Mj does not contain regulatory calcium-binding domains (CBDs), which makes the NCX_Mj protein very suitable for studying the mechanisms underlying ion transport.11,12,20−24 Recent hydrogen−deuterium exchange mass spectrometry (HDX-MS) studies have identified hallmark differences in the backbone dynamics at ion-coordinating residues of apo-NCX_Mj, whereas binding of Na+ or Ca2+ to the respective sites induces relatively small, but specific, changes in backbone dynamics.21 Moreover, kinetic analyses of mutants have revealed the asymmetric contributions of ion-coordinating residues (at inverted topological positions) to ion-transport catalysis.21 Currently, it is unclear how these structure−dynamic arrangements contribute to ion selectivity and the intrinsic asymmetry of bidirectional ion movements observed in prokaryotic17 or mammalian30 NCX orthologs. Notably, the intrinsic asymmetry of bidirectional ion movements is essential for setting physiologically relevant Km values at opposite sides of the membrane.12,17,30 This principle is required for controlling the dynamic range of the transporter function at physiologically relevant concentrations of cytosolic and extracellular ions.11,12 The ion-binding pocket of NCX_Mj encompasses highly conserved α1 and α2 repeats with an inverted two-fold topology and 12 ion-coordinating residues (four in TM2 and TM7 and two in TM3 and TM8). Notably, the ion-coordinating residues are highly conserved among Ca2+/CA proteins despite the diversity of ion-transport selectivities.11−16 For example, 11 ioncoordinating residues (of 12) are conserved in prokaryotic and eukaryotic NCXs.11,12,20 The mitochondrial Na+/Ca2+ exchanger (NCLX) differs from the other NCXs and NCKXs in that it can transport either Li+ or Na+ (but not K+) in exchange with Ca2+.24,25 Moreover, in NCLX, nine ioncoordinating residues (of 12) do not match their counterparts in NCX_Mj or other NCXs (Figure 1A). Because Li+ is not a “physiological” ion, the biological relevance of Li+ transport remains enigmatic.11,12 Because the crystal structures of NCX_Mj and CAX proteins, as well as sequence alignments within the Ca2+/CA proteins, imply that the positions of ion-coordinating residues (on the α1 and α2 repeats) are highly conserved,13−16,20 it is reasonable to propose that distinct residues at matching positions may control the ion-transport selectivity in NCX and NCLX proteins. To test this hypothesis, the residues at matching positions were replaced in NCX_Mj to mimic the NCLX “ion-binding pocket” and the hybrid construct (NCX_Mj/NCLX) was analyzed for its capacity to transport Na+ or Li+ in exchange with Ca2+. B

DOI: 10.1021/acs.biochem.6b00119 Biochemistry XXXX, XXX, XXX−XXX

Biochemistry

Rapid Report

analyses of single-point back-mutations are required to resolve the underlying mechanisms of ion selectivity and transport. Whatever the structure−dynamic determinants of monovalent ion selectivity are in a given context of the structural mimicking of NCLX, the mechanistic implications for iontransport catalysis are of sizable interest. This is because Sext and Sint have high selectivity for Na+ in NCX_Mj, whereas the SCa site can bind either Ca2+ or Na+ ions.13,20,21 Consequently, this may play a critical role in triggering the ion-coupled conformational changes associated with alternating access.20−24 Interestingly, the structural similarities of the NCX_Mj and CAX proteins suggest that the “sliding” mechanism of alternating access might be a common feature not only for prokaryotic and eukaryotic NCX variants but also for many other proteins that belong to the Ca2+/CA exchanger superfamily.11−16,20,21 The remaining open question is how the Ca2+/CA proteins can generate the TM1/TM6 movement upon ion binding, which seems to be a common feature for the Ca2+/CA exchanger proteins in achieving alternating access, while having distinct ion selectivity.13−16 For evaluating the apparent affinity of Na+ or Li+ transport, the initial rates of Na+i- or Li+i-dependent 45Ca2+ uptake were measured by varying the concentration of intravesicular Na+ or Li+ at a fixed concentration of extravesicular 45CaCl2 (250 μM). As shown in Figure 2C, NCX_Mj exhibits a Km of 20 mM with a high Vmax value, whereas the capacity for Li+ transport falls below the detection limits; therefore, the observed small signals of 45Ca2+ uptake cannot be reliably quantified in the given experimental system. Nevertheless, within the limits of the analytical procedures used, it is reasonable to assume that the lower limit for the Li+:Na+ selectivity in NCX_Mj approaches 1:25 or higher values (Table S1). A similar degree of ion selectivity is observed in prokaryotic ion channels, although in contrast with transporters, the selectivity filters of the ion channels are predominantly formed by carbonyl groups, not side chains.2−6 In contrast with NCX_Mj, the NCX_Mj/NCLX protein exhibits similar Vmax and Km values for Na+ and Li+, whereas the apparent affinities for both ions (Km ≃ 1.5 mM) are at least 10fold higher than the apparent affinity observed for Na+ (Km ≃ 20 mM) in NCX_Mj (Figure 2C). There are only 2-fold differences in the Vmax values between NCX_Mj and its NCLX analogue, suggesting that the observed 10-fold difference in Km(Na) is caused by changes in equilibrium binding (Kd) rather than by altering the transport rates (kcat).17,21 NCX_Mj/ NCLX exhibits similar Km and Vmax values for Na+ and Li+, which reveals a 1:1 selectivity for Li+:Na+ (Table S1). For mitochondrial NCLX, Km values of ∼13 μM and ∼8 mM were reported for Ca2+ and Na+, respectively,26 whereas the exact Km value for Li+ remains to be established. The structure−dynamic determinants of ion selectivity were extensively studied in biological and synthetic ion channels, although the underlying dynamic mechanisms remain under debate.2−4,8 The relevant mechanisms are even less clear in ion pumps and transporters.4−7 In general, the structure−dynamic arrangement (preorganization) of the first- and second-shell residues at binding sites may differ in ion channels and transporters, because the fundamental mechanisms underlying the ion passageway basically differ in these two types of iontransport systems.27,28 Because the structural mimicking of the NCLX ion-binding pocket dramatically modifies Na+:Li+ selectivity but has a rather small effect on the ion exchange rates (kcat), the model presented here could be instrumental for

Figure 2. 45Ca2+ flux assays in E. coli-derived isolated vesicles containing overexpressed NCX_Mj or NCX_Mj/NCLX. (A) Time course of Na+i- or Li+i-dependent 45Ca2+ uptake measured for NCX_Mj (filled red circles or empty red circles for Na+/Ca2+ or Li+/Ca2+ exchange, respectively) and NCX_Mj/NCLX (filled blue triangles or empty blue triangles for Na+/Ca2+ or Li+/Ca2+ exchange, respectively). (B) Initial rates (t = 5 s) of Na+/Ca2+ and Li+/Ca2+ exchange in NCX_Mj and NCX_Mj/NCLX measured by varying the concentration of 45Ca2+ (25−1600 μM) in the assay medium at a fixed (saturating) concentration of intravesicular Na+ or Li+ (160 mM). Data for Na+/Ca2+ exchange of NCX_Mj were fit with a Vmax of 526 ± 15 pmol of Ca2+ (mg of protein)−1 s−1 and a Km of 73 ± 9 μM. Data for Na+/Ca2+ exchange of NCX_Mj/NCLX were fit with a Vmax of 272 ± 7 pmol of Ca2+ (mg of protein)−1 s−1 and a Km of 113 ± 11 μM. Data for Li+/Ca2+ exchange of NCX_Mj/NCLX were fit with a Vmax of 234 ± 8 pmol of Ca2+ (mg of protein)−1 s−1 and a Km of 102 ± 12 μM. (C) Initial rates (t = 5 s) of Na+/Ca2+ and Li+/Ca2+ exchange in NCX_Mj and NCX_Mj/NCLX measured by varying the intravesicular concentration of Na+ or Li+ (1−160 mM, in the presence of 3 mM EGTA) at a fixed concentration of 45Ca2+ (250 μM) in the assay medium. Data for NCX_Mj Na+/Ca2+ exchange were fit with a Vmax of 560 ± 34 pmol of Ca2+ (mg of protein)−1 s−1 and a Km of 23 ± 3 mM. Data for NCX_Mj/NCLX Na+/Ca2+ exchange were fit with a Vmax of 218 ± 14 pmol of Ca2+ (mg of protein)−1 s−1 and a Km of 1.4 ± 0.4 mM. Data for Li+/Ca2+ exchange of NCX_Mj/NCLX were fit with a Vmax of 234 ± 17 pmol of Ca2+ (mg of protein)−1 s−1 and a Km of 1.4 ± 0.4 mM. C

DOI: 10.1021/acs.biochem.6b00119 Biochemistry XXXX, XXX, XXX−XXX

Biochemistry

Rapid Report

(11) Khananshvili, D. (2013) Mol. Aspects Med. 34, 220−235. (12) Khananshvili, D. (2014) Pfluegers Arch. 466, 43−60. (13) Liao, J., Li, H., Zeng, W., Sauer, D. B., Belmares, R., and Jiang, Y. (2012) Science 335, 686−690. (14) Waight, A. B., Pedersen, B. P., Schlessinger, A., Bonomi, M., Chau, B. H., Roe-Zurz, Z., Risenmay, A. J., Sali, A., and Stroud, R. M. (2013) Nature 499, 107−110. (15) Nishizawa, T., Kita, S., Maturana, A. D., et al. (2013) Science 341, 168−172. (16) Wu, M., Tong, S., Waltersperger, S., Diederichs, K., Wang, M., and Zheng, L. (2013) Proc. Natl. Acad. Sci. U. S. A. 110, 11367−11372. (17) Almagor, L., Giladi, M., van Dijk, L., Buki, T., Hiller, R., and Khananshvili, D. (2014) Cell Calcium 56, 276−284. (18) Reeves, J. P., and Hale, C. C. (1984) J. Biol. Chem. 259, 7733− 7739. (19) Khananshvili, D. (1990) Biochemistry 29, 2437−2442. (20) Marinelli, F., Almagor, L., Hiller, R., Giladi, M., Khananshvili, D., and Faraldo-Gómez, J. D. (2014) Proc. Natl. Acad. Sci. U. S. A. 111, E5354−E5362. (21) Giladi, M., Almagor, L., van Dijk, L., Hiller, R., Man, P., Forest, E., and Khananshvili, D. (2016) Scientific Reports (in press). (22) Hilge, M., Aelen, J., and Vuister, G. W. (2006) Mol. Cell 22, 15− 25. (23) Khananshvili, D. (2016) Regulation of Ca2+-ATPases, VATPases and F-ATPases. In Advances in Biochemistry in Health and Disease (Chakraborti, S., and Dhalla, N. S., Eds.) Vol. 14, pp 93−116, Springer, Berlin. (24) Giladi, M., Lee, S. Y., Hiller, R., Chung, K. Y., and Khananshvili, D. (2015) Biochem. J. 465, 489−501. (25) Carafoli, E., Tiozzo, R., Lugli, G., Crovetti, F., and Kratzing, C. (1974) J. Mol. Cell. Cardiol. 6, 361−371. (26) Palty, R., Silverman, W. F., Hershfinkel, M., Caporale, T., Sensi, S. L., Parnis, J., Nolte, C., Fishman, D., Shoshan-Barmatz, V., Herrmann, S., Khananshvili, D., and Sekler, I. (2010) Proc. Natl. Acad. Sci. U. S. A. 107, 436−441. (27) Jardetzky, O. (1966) Nature 211, 969−970. (28) Forrest, L. R., Krämer, R., and Ziegler, C. (2011) Biochim. Biophys. Acta, Bioenerg. 1807, 167−188. (29) Lee, S. Y., Giladi, M., Bohbot, H., Hiller, R., Chung, K. Y., and Khananshvili, D. (2016) FASEB J. 30, 1356−1366. (30) Khananshvili, D., Weil-Maslansky, E., and Baazov, D. (1996) Ann. N. Y. Acad. Sci. 779, 217−235.

segregating the partial contributions of ion binding and transport events in governing ion selectivity. This work puts forth the possibility that the structural template of NCX_Mj can be used for engineering novel iontransport proteins with predefined ion selectivity. In the past few years, advanced approaches have been applied on NCX proteins (small-angle X-ray scattering, HDX-MS, stopped-flow, among others).21,23,24,29 Further extension of these advanced experimental approaches in conjunction with dedicated MD simulations20 may provide new opportunities for improving our understanding of the structure−dynamic determinants governing ion selectivity in NCX and similar proteins.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.biochem.6b00119. Detailed experimental procedures, a schematic representation of the ping-pong mechanism of the ion-transport cycle (Figure S1), and kinetic parameters (Table S1) (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*Department of Physiology and Pharmacology, Sackler School of Medicine, Tel-Aviv University, Ramat-Aviv, Tel-Aviv 69978, Israel. Telephone: 972-3-640-9961. Fax: 972-3-640-9113. Email: [email protected]. Author Contributions

B.R., M.G., and D.K. designed the experiments. B.R., M.G., and R.H. conducted the experiments. B.R., M.G., R.H., and D.K. analyzed and interpreted the experimental data. D.K. wrote the manuscript. Funding

This work was partially funded by Israel Science Foundation Grant 825/14 and the Fields Estate Foundation to (D.K.). Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS We appreciated the inspiring discussions with Prof. José D. Faraldo-Gómez, Prof. Youxing Jiang, and Dr. Fabrizio Marinelli. REFERENCES

(1) Åqvist, J., Alvarez, O., and Eisenman, G. (1992) J. Phys. Chem. 96, 10019−10025. (2) Yu, H., Noskov, S. Y., and Roux, B. (2010) Proc. Natl. Acad. Sci. U. S. A. 107, 20329−20334. (3) Dudev, T., and Lim, C. (2014) Chem. Rev. 114, 538−556. (4) Lockless, S. W. (2015) J. Gen. Physiol. 146, 3−13. (5) Sauer, D. B., Zeng, W., Canty, J., Lam, Y., and Jiang, Y. (2013) Nat. Commun. 4, 2721−2729. (6) Roux, B., Bernèche, S., Egwolf, B., Lev, B., Noskov, S. Y., Rowley, C. N., and Yu, H. (2011) J. Gen. Physiol. 137, 415−426. (7) Leone, V., Pogoryelov, D., Meier, T., and Faraldo-Gómez, J. (2015) Proc. Natl. Acad. Sci. U. S. A. 112, E1057−E1066. (8) Yu, H., Noskov, S. Y., and Roux, B. (2010) Proc. Natl. Acad. Sci. U. S. A. 107, 20329−20334. (9) Zhang, X., Hou, Y., He, W., Yang, G., Cui, J., Liu, S., Song, X., and Huang (2015) Nanoscale 7, 3356−3372. (10) Rahmat, F., Thamwattana, N., and Cox, B. J. (2011) Nanotechnology 22, 445707. D

DOI: 10.1021/acs.biochem.6b00119 Biochemistry XXXX, XXX, XXX−XXX