Conformational free-energy landscapes of a Na+/Ca2+ exchanger explain its alternating-access mechanism and functional specificity

Abstract

Secondary-active transporters catalyze the movement of myriad substances across all cellular membranes, typically against opposing concentration gradients, and without consuming any ATP. To do so, these proteins employ an intriguing structural mechanism evolved to be activated only upon recognition or release of the transported species. We examine this self-regulated mechanism using a homolog of the cardiac Na⁺/Ca²⁺ exchanger as a model system. Using advanced computer simulations, we map out the complete functional cycle of this transporter, including unknown conformations that we validate against existing experimental data. Calculated free-energy landscapes reveal why this transporter functions as an antiporter rather than a symporter, why it specifically exchanges Na⁺ and Ca²⁺, and why the stoichiometry of this exchange is exactly 3:1. We also rationalize why the protein does not exchange H⁺ for either Ca²⁺ or Na⁺, despite being able to bind H⁺ and its high similarity with H⁺/Ca²⁺ exchangers. Interestingly, the nature of this transporter is not explained by its primary structural states, known as inward- and outward-open conformations; instead, the defining factor is the feasibility of conformational intermediates between those states, wherein access pathways leading to the substrate binding sites become simultaneously occluded from both sides of the membrane. This analysis offers a physically coherent, broadly transferable route to understand the emergence of function from structure among secondary-active membrane transporters.

Fig. 1.

Topological repeats within the architecture of NCX_Mj, the Na⁺/Ca²⁺ exchanger from M. jannaschii. (A) Crystal structure of NCX_Mj in the outward-facing state (7). The structure contains two inverted topological repeats of five helices each, namely TM1 to TM5 and TM6 to TM10, shown in orange and marine, respectively. The protein is bound to three Na⁺ ions, shown in green. (B and C) Overlays of the internal structures of the two repeats reveals they are in part similar, but clearly distinct. The angle formed between TM2a and TM2b differs from that observed between TM7a and TM7b; the relative position of TM2a and TM1 also differs from that between TM7a and TM6. Taken together, these two differences explain why the structure shown in (A) features an access pathway into the Na⁺ binding site from the extracellular side. (D) A hypothetical structural model wherein the repeats swap conformations (see SI Appendix for more details), with no further adjustment, shows an intracellular access pathway analogous to that in panel (A).

Fig. 2.

Free-energy landscapes underlying the alternating-access mechanism of NCX_Mj. (A) Landscapes are shown for the transporter loaded with 3 Na⁺ ions (Left) or with 1 Ca²⁺ ion (Right). For clarity, the maps represent the conformational free energy as a function of the degree of opening or closing of intra- and extracellular access pathways into the ion binding site. To objectively quantify accessibility, we use the number of contacts between two sets of protein residues lining those pathways. On the extracellular side, these contacts are between TM6 or TM7 and TM2 or TM3; on the intracellular side, between TM1 or TM2 and TM7 or TM8. Note that these selections are topologically symmetric with respect to the membrane plane. The two free-energy minima featured in each map correspond to the outward and inward-facing states (OF and IF) of the transporter. Red circles mark the positions in these maps of the outward-facing crystal structures of NCX_Mj obtained at high Na⁺ (Left) or high Ca²⁺ (Right) concentrations (PDB entries 3V5U/5HXE and 5HXR, respectively), the repeat-swap model shown in Fig. 1D, and the recently determined cryo-EM structure of NCX1 (20). Contours are shown in intervals of 1 kcal/mol. Error estimates for each map are provided in SI Appendix, Fig. S4B. (B) Water density maps for each of the free-energy minima revealed in the maps in panel (A) are overlaid onto representative configurations. For clarity only water molecules within 12 Å of the ion binding sites are mapped. Note OF and IF states have opposing water accessibility patterns, though in all cases the binding sites are occluded from the solvent. (C) Close-up of the ion binding sites, highlighting the mode of ion coordination in each case. Note the binding site geometries for OF and IF states are nearly identical.

Fig. 3.

Validation of the predicted IF structure of NCX_Mj against existing experimental data. (A) Graphical representation of published HDX-MS data (23) for two forms of NCX_Mj (WT and 5L6-8) with distinct propensities to adopt the OF of IF state (see text). Differences in measured deuteration levels for these two forms (after 15 s and for saturating [Na⁺] conditions) are indicated with a color scale, for the regions of the protein that were examined experimentally (TM2, TM7, and TM8). Data on overlapping protein fragments has been broken down into smaller regions by linear combination of the deuterated fractions. Residues G42 and G201, whose accessibility was probed through complementary labeling assays, are also highlighted. For reference, the figure also shows density isosurfaces for water molecules within 12 Å of the protein binding sites (gray mesh), based on our simulation data. (B) Solvent accessibility of G42 and G201 in either the OF or IF states identified in the free-energy maps in Fig. 2, as well as in the intermediate regions. This accessibility is quantified by the number of number water molecules found, on average, within 4.2 Å of either residue. (C) For both WT and 5L6-8, the degree of deuteration measured for a collection of protein fragments in TM2, TM7 and TM8 (after 15 s and for saturating [Na⁺]) is contrasted with deuteration levels calculated for an ensemble of OF and IF structures extracted from the free-energy basins in Fig. 2 (SI Appendix). Calculated and experimental data are compared when OF and IF are equally weighted, and for alternative weights that result in optimal agreement with measured data, for either WT or 5L6-8 (SI Appendix, Fig. S6). These population shifts are consistent with the known conformational propensities of these forms. (D) Comparison between the IF state of NCX_Mj identified in the free-energy landscapes shown in Fig. 2 with a recently reported cryo-EM structure of human NCX1 (20), coincidentally captured in the IF state. The RMS difference between the Cα traces is 1.9 Å.

Fig. 4.

Mechanism of alternating-access in NCX_Mj inferred from analysis of simulation data. (A and B) Comparison between the OF and IF states, using representative configurations of the free-energy minima in Fig. 2, for the Ca²⁺- and Na⁺-bound transporter, respectively. Note the displacement of the TM1 to TM6 unit across the membrane midplane, alongside localized changes in the intracellular and extracellular halves of TM2 (TM2ab) and TM7 (TM7ab), respectively. Close-ups of the ion binding sites are also shown, for the OF and IF states as well as for a doubly occluded intermediate (see below). (C and D) Change in free-energy along the minimum free-energy (most probable) multidimensional path connecting the OF and IF states, for the Ca²⁺- and Na⁺-bound transporter, respectively. The profiles reveal a metastable intermediate about halfway through the transition. These profiles were calculated in a four-dimensional space (Methods); additional dimensions did not significantly alter their features. (E and F) OF, intermediate and IF states are compared side by side, highlighting water molecules in proximity to the ion-binding sites, filling access pathways into the protein interior from either side of the membrane. Note the displacement of TM1-TM6 in panels (A and B) in fact entails two distinct movements in different directions, as indicated.

Fig. 5.

Free-energy landscapes dictate the functional specificity of NCX_Mj. (A) Free-energy landscapes analogous to those shown in Fig. 2, but for different ion-occupancy states, namely with no ions bound, with 2 H⁺ bound to E54 and E213, and with 2 Na⁺ bound only. Red circles mark the positions in these maps of the outward-facing crystal structures of NCX_Mj obtained at low pH and at low and high [Na⁺] (PDB entries 5HXH, 5HWY and 5HXE, respectively), and of the recently determined cryo-EM structure of NCX1 (20). The landscapes show the alternating-access transition is energetically unfeasible, and that the transporters are trapped either in OF or IF conformations that are distinctly more open to the solvent that those favored when either 3 Na⁺ or 1 Ca²⁺ ion are bound; Insets show an overlay of these different conformations, for both the OF (Left) and IF (Right) states. Note that we did not quantitate the precise magnitude of the free-energy barrier between the OF and IF states, as it appears to be in the order of tens of kcal/mol. The free-energy maps for OF and IF states are shown on the same scale starting at a zero value solely for conciseness. The actual value of the relative free-energy between OF and IF states is unknown; this quantity is however irrelevant, as these states are not in equilibrium with each other in these particular ion-occupancy configurations. Error estimates for each map are provided in SI Appendix, Fig. S4C. (B) Water density maps for each of the free-energy minima revealed in the maps in panel (A) are overlaid onto representative configurations. For clarity only water molecules within 12 Å of the ion binding sites are mapped. Note OF and IF states have opposing water accessibility patterns, and that in all cases the ion binding sites are readily exposed to the solvent, but only on one side of the membrane. (C) Close-up of the ion binding sites, highlighting the configuration of the ion coordination shell in each case.

Fig. 6.

Specificity in the mechanism of alternating-access of the Na⁺/Ca²⁺ exchanger. This study demonstrates that the functionality of NCX_Mj, namely antiport of 3 Na⁺ and 1 Ca⁺, owes to the fact that the alternating-access transition is not viable for any other substrate-occupancy states, including the apo protein. The scheme highlights the essence of the conformational changes that the transporter undergoes when transitioning between the OF and IF states, namely a sliding motion of TM1 and TM6 (blue and brown empty cylinders, respectively) across the membrane mid-plane, and changes in secondary structure in TM2 and TM7 (orange and marine full cylinders, respectively). Also highlighted are the side chains of E54 (on TM2c) and E213 (on TM7c) and the carbonyl groups of A47 (on TM2b) and A206 (on TM7b), which coordinate Na⁺, Ca²⁺ and H⁺. The scheme shows only the states and connectivity deduced from the calculated free-energy landscapes in Figs. 2 and 5A; additional open or semiopen states with only 1 Na⁺ or 1 H⁺ bound, or a combination of Na⁺, Ca²⁺ and H⁺, are also conceivable. Asterisks mark the states that have been also determined by X-ray crystallography.