Crystal structure of the sodium-proton antiporter NhaA dimer and new mechanistic insights

Abstract

Sodium-proton antiporters rapidly exchange protons and sodium ions across the membrane to regulate intracellular pH, cell volume, and sodium concentration. How ion binding and release is coupled to the conformational changes associated with transport is not clear. Here, we report a crystal form of the prototypical sodium-proton antiporter NhaA from Escherichia coli in which the protein is seen as a dimer. In this new structure, we observe a salt bridge between an essential aspartic acid (Asp163) and a conserved lysine (Lys300). An equivalent salt bridge is present in the homologous transporter NapA, but not in the only other known crystal structure of NhaA, which provides the foundation of most existing structural models of electrogenic sodium-proton antiport. Molecular dynamics simulations show that the stability of the salt bridge is weakened by sodium ions binding to Asp164 and the neighboring Asp163. This suggests that the transport mechanism involves Asp163 switching between forming a salt bridge with Lys300 and interacting with the sodium ion. pKa calculations suggest that Asp163 is highly unlikely to be protonated when involved in the salt bridge. As it has been previously suggested that Asp163 is one of the two residues through which proton transport occurs, these results have clear implications to the current mechanistic models of sodium-proton antiport in NhaA.

Figure 1.

Schematic diagram of the NhaA structure. The core and dimer domains are illustrated with blue and beige shadowing, respectively. TM 4 (red) and TM 11 (yellow) are discontinuous and cross over in the center of the protein. Asp133 and Lys300 have been proposed to neutralize the positively and negatively charged helix dipoles of the discontinuous helices. Asp163 and Asp164 are thought to interact with the sodium ion. Coordinates are from Protein Data Bank accession number 1ZCD (Hunte et al., 2005).

Figure 2.

Electron density. (A) Electron density of TM 10. The 2mFo-DFc map shown in blue (contoured at 1.5 σ) was calculated with phases derived from the model before reassigning the sequence of TM 10 (shown) and averaged over the four molecules of the asymmetric unit. The anomalous difference map shown in red has been calculated from the selenomethionine-derivatized triple mutant and contoured at 3.6 σ. (B) The same maps as in A but with the structure of the wild-type protein (refined before the data of the triple mutant were collected). (C) Stereo view of the superposition of the final refined structure of the triple mutant (green) on that of the published monomeric structure (gray) (Hunte et al., 2005). The anomalous difference map is shown as in A. Met296 is in cyan, and Leu296 from the published structure is in wine red. Overall, there is good correspondence in the position of the helices and methionines between the two structures, in agreement with the similar activity of the mutant to wild type (Fig. 4). An enlarged view of TM 10 is shown in Fig. S1.

Figure 3.

Position of Lys300 on TM 10. (A) Cartoon representation of the structure viewed from the periplasmic side of the membrane. The charged residues, Asp133 (TM 4b), Asp163 (TM 5) and Asp164 (TM 5), and Lys300 (TM 10), in the new structure are shown as colored sticks. The position of Lys300 in the previously published structure (Protein Data Bank accession no. 1ZCD) is shown in gray (red text). The structure has been colored from red at the N terminus through to blue at the C terminus. Loop regions except for the breaks between the discontinuous helices have been omitted for clarity. (B) As A, but from the side and zoomed in. The dashed line represents the salt bridge between Asp163 and Lys300.

Figure 4.

Stabilization, characterization, and crystallization of the NhaA mutant. (A) The NhaA double mutant (A109T and Q277G) and NhaA triple mutant (A109T, Q277G, L296M) are more stable in detergent as shown by the longer unfolding half-life (_t1/2) in LDAO at 40°C. (B) The ATP synthase and NhaA wild-type and mutants were co-reconstituted in liposomes. ATP-driven proton pumping establishes a ΔpH (acidic inside) as monitored by the quenching of 9-amino-6-chloro-2-fluorescence (ACMA). Proton efflux is initiated by the addition of increasing concentrations of NaCl/LiCl, and apparent ion-binding affinities for NhaA wild type (closed circle) and mutant (open circle) at pH 8.5 were calculated: K_MNa⁺ wild type (mean ± SD): 1.8 ± 0.2; K_MNa⁺ mutant: 1.6 ± 0.1; K_MLi⁺ wild type: 4.1 ± 0.5; K_MLi⁺ mutant: 4.1 ± 0.3. (C) pH dependence of NhaA Na⁺(Li⁺)–H⁺ antiporter activity for wild type (closed circle) and mutant (open circle) were measured in proteoliposomes by the level of ACMA dequenching as in B at the indicated pH values after the addition of saturating NaCl/LiCl at pH 8.5; all experiments were repeated in triplicate and representative traces are shown.

Figure 5.

Position of β hairpins in the NhaA dimer. (A) 2mFo-DFc electron density averaged over the four molecules of the asymmetric unit (contoured at 1.5 σ). The map was calculated directly after molecular replacement using a search model where the β hairpins and loops had been omitted. (B) Cartoon representation of the NhaA dimer viewed from the periplasmic-facing side of the membrane, with the two protomers shown in light brown and green, respectively. (C) The 14–amino acid β hairpins from each promoter, shown in light brown and green, form a four-stranded antiparallel β sheet, as viewed from the cytoplasmic side of the membrane. The residues facing the membrane are hydrophobic except for Lys57. (D) Comparison of the positions of the β hairpins in each of the determined NhaA structures: crystal structure of the dimer (this work; light brown and green), monomeric crystal structure (Hunte et al., 2005; pale green), and dimer from cryo-EM (Appel et al., 2009; yellow).

Figure 6.

MD simulation of the NhaA dimer in a model membrane bilayer. (A) The NhaA dimer is stable in a POPC membrane. Snapshots show the structure after an ∼1-µs MD (simulation S2/1). The periplasmic space is at the top. Protomers are colored wheat and green, and POPC lipids are white. A bound sodium ion is shown as a cyan sphere. Residues Asp163, Asp164, and Lys300 are partially visible in a stick representation. (B) View from the periplasm.

Figure 7.

Sodium ion binding and salt-bridge stability in MD simulations with different protonation states of Asp163, Asp164, and Lys300. (Left column; A and B) Asp163 deprotonated and Asp164 and Lys300 protonated (simulation S1/1, protomer B). (Middle column; C and D) Asp163 and Asp164 deprotonated and Lys300 protonated (simulation S2/1, protomer B). (Right column; E and F) Asp163, Asp164, and Lys300 deprotonated (simulation S4/1, protomer B). (Top row; A, C, and E) Distances between the closest sodium ion and Asp163 or Asp164 are plotted as a function of time. Spontaneous Na⁺ binding to Asp164 was observed when both aspartates were deprotonated. (C) Continuation of the simulation with Lys300 deprotonated (a 3-ns equilibration simulation with position restraints on all heavy protein atoms is symbolized by dashed lines between panels) leads to a rapid change in the Na⁺-binding mode toward closer interaction with Asp163. (Bottom row; B, D, and F) Distance of the closest Asp163 carboxyl group from the N-amino group of Lys300. Distances <4 Å are indicative of a stable salt-bridge interaction (yellow shaded area), whereas those ≥4 Å are considered a weak or broken salt bridge. Binding of Na⁺ to Asp164 destabilizes the salt bridge. (D) Lines show data averaged over blocks of 10 ns, with fluctuations in the data indicated as shaded regions encompassing the lower 5 and upper 95% percentile. The snapshots show the Na⁺ binding event with subsequent rupture of the salt bridge (cytoplasmic view along the axis of helix TM 5). Videos 1 and 2 show this simulation. Repeat simulations (see Table 1) are shown in Figs. S3–S7.

Figure 8.

Effect of breaking the Asp163–Lys300 salt bridge on the pK_a of conserved residues. (A) Distributions of pK_a values estimated with PROPKA 3.1 (Søndergaard et al., 2011) from all MD simulations shown as violin plots. pK_a values were calculated from snapshots extracted at 1-ns intervals, with sodium ions included within 6 Å of the protein. For reference, a dotted line has been drawn at pH 7.6. The percentiles of the data are shown as broken lines inside the distributions, which were computed as Gaussian kernel density estimates (see Materials and methods). Data were split depending on the state of the salt bridge (formed if the distance is <4 Å, and broken if it is ≥4 Å; the salt-bridge distance distribution is also shown in yellow). The multimodal distribution of Asp164 when the salt bridge is formed is a result of data from simulations S1 (Fig. S9 A) during which its solvent accessibility is much smaller than during other simulations (see also Fig. S3, M–O). (B) Distributions of pK_a shifts (ΔpK_a) when the salt bridge is broken (data aggregated over all MD simulations). The pK_a of Lys300 is downshifted by −2.5 ± 1.4, and thus the charged form is destabilized. Shifts of the other residues are not significantly different from 0 (also see Table 3).

Figure 9.

Comparison of NhaA with NapA and ASBT_NM. (A) Structural comparison of NhaA and NapA. Cartoon representation of TMs 4, 5, 10, and 11 of NhaA superimposed on the same helices of NapA (Lee et al., 2013). Residue numbering is shown for NhaA. K300 in NhaA corresponds to K305 in NapA, and D163 in NhaA corresponds to D156 in NapA. The helices of the TMs in NhaA have been colored as in Fig. 3, and NapA is colored salmon. (B) Structural comparison of NhaA and ASBT_NM. Cartoon representation of TMs 4, 10, and 11 of NhaA superimposed on TMs 4, 8, and 9 of ASBT_NM (Hu et al., 2011). The helices of the TMs in NhaA have been colored as in A, and ASBT_NM is colored gray. The sodium Na2 site in ASBT_NM is depicted as a purple sphere, and Lys300 residue in NhaA is shown as a stick model.

Figure 10.

Proposed schematic model of NhaA transport. The transport mechanism occurs through a reorientation of the protein, alternately exposing a cavity containing Asp164 to the intracellular or the periplasmic space. The core domain is depicted as a blue square, and the panel domain is in beige. (1) The probable situation in the solved NhaA structure at low pH where the protein will be protonated. One of the protons (in orange) is likely to be bound to Asp164, and we hypothesize that the second proton (orange hashed) will interact with Lys300. Sodium ions (blue) will compete with protons for binding to Asp164, causing the Asp163–Lys300 salt bridge to break and possibly Lys300 to lose a proton (2). The sodium ion–bound protein will then switch to the outward-facing state (3). Upon release of the sodium ion, the protein will be reprotonated and the salt bridge between Lys300 and Asp163 reformed (4). The protein will then switch back to the inward-facing conformation.