Structure, mechanism and lipid-mediated remodeling of the mammalian Na+/H+ exchanger NHA2

The Na+/H+ exchanger SLC9B2, also known as NHA2, correlates with the long-sought-after Na+/Li+ exchanger linked to the pathogenesis of diabetes mellitus and essential hypertension in humans. Despite the functional importance of NHA2, structural information and the molecular basis for its ion-exchange mechanism have been lacking. Here we report the cryo-EM structures of bison NHA2 in detergent and in nanodiscs, at 3.0 and 3.5 Å resolution, respectively. The bison NHA2 structure, together with solid-state membrane-based electrophysiology, establishes the molecular basis for electroneutral ion exchange. NHA2 consists of 14 transmembrane (TM) segments, rather than the 13 TMs previously observed in mammalian Na+/H+ exchangers (NHEs) and related bacterial antiporters. The additional N-terminal helix in NHA2 forms a unique homodimer interface with a large intracellular gap between the protomers, which closes in the presence of phosphoinositol lipids. We propose that the additional N-terminal helix has evolved as a lipid-mediated remodeling switch for the regulation of NHA2 activity.

(1) Spontaneous binding from 349.9 ns onwards until the end of the simulation.
(2) Spontaneous partial binding from 327.0 ns onwards until the end of the simulation.Bar represent normalized expression level compared to bison NHA2 DN for various NHA2 constructs and mutants as measured by GFP fluorescence in yeast whole-cells.Errors bars represent the: mean values ± s.d. of n = 4 biologically independent samples for biNHA2∆N, D277C-D278C, biNHA2 WT, ∆TM-1, E214A, R431A, T461E, E214R-R431E, hsNHA2∆N; mean values ± s.d. of n = 3 biologically independent samples for H169A, S244D, W456A, W456F, K459A, K459R; range of n = 2 biologically independent samples for R176A, E214R, R431E, T461A, D330A-Q331A; note only the full-length bison NHA2 construct is poorly expressed (10-fold less than bison NHA2 DN) and has a large fraction of aggregates (see b.), yet still shows significant complementation at 10 mM LiCl and some complementation at 25 mM LiCl (Supplementary Fig. 3b).b.FSEC traces of listed NHA2 constructs extracted by DDM/CHS from ABc11 yeast membranes.The peak at 11.5 ml corresponds to the NHA2 homodimer and all the constructs shown here have a main FSEC peak at 11.5ml; the only constructs with a shifted predominant peak at 12.2 ml for the monomer are NHA2TM-1 and NHA2 DN (Q330A, D331A) shown in Fig. 3b.All constructs showing a minor monomer peak at 12.2 ml show poor folding as apparent by the aggregation peak in the void.e.g., E214R, R431A, T461E, and bison NHA2 full-length (WT).c.Thermal shift of purified monomeric NHA2  Supplementary Fig. 3 a.

d. c.
Ion-binding site Oligomerization   In most cases Li + was used to asses NHA2 complementation as Li + is more toxic to yeast cells and NHA2 DN has a higher apparent affinity (KD) for Li + vs. Na + .Nevertheless, for those ion-binding site mutants that retained some Li + -complementation, they were further assessed for Na + sensitivity.above: Yeast were grown as outlined in Methods in -URA media supplemented with 2% galactose and NaCl and growth was determined by optical density of the culture at 600 nm (OD600) after 72 at 30°C and normalised against growth of bison NHA2 DN. below: To facilitate comparison the above data is shown as bar graphs for 300 and 400 mM NaCl.Errors bars represent the: mean values ± s.d. of n = 3 biologically independent samples for biNHA2∆N, D277C-D278C, W456F, T461E and E214R-R431E; range of n = 2 biologically independent samples for E214R, R431E, T461A.
Expression and quality of NHA2-GFP fusions assessed by fluorescence-detection size exclusion chromatography (FSEC) in the ABc11 yeast strain.a.
Multiple sequence alignment of bison NHA2 and eukaryotic homologues.NHA2 selected sequences of mammalian vertebrates, Xenopus laevis (African clawed frog) and Human (SLC9B2) were aligned.Residues with over 70% sequence identity are indicated by purple background.Conserved ion-binding site residues are highlighted with a red border.Positions which have been identified as part regions for lipid binding are highlighted with yellow border, and proline and glycine clusters associated with TM -1 mobility with orange border.Helix breakpoints (zigzag-line), and helices of the core 6-TM transport domain TMs (pink), dimerization domain (green) and domain-swapped helix TM -1 (blue) are shown and enumerated.
Functional complementation of Li + sensitivity in the yeast strain AB11c by heterologous expression of NHA2 constructs.a. Schematic showing the yeast saltstress assay used for screening for NHA2 functional activity.b.Ab11c strain was transformed with bison NHA2 WT, bison NHA2 DN, human NHA2 DN or bison NHA2 DN with both Asp277 and Asp278 have been substituted with cysteine (D277C-D278C).Yeast were grown as outlined in Methods in -URA media supplemented with 2% galactose and LiCl at listed concentrations and growth was determined by optical density of the culture at 600 nm (OD600) after 48 at 30°C and normalised against growth of bison NHA2 DN.Errors bars represent the mean values ± s.d. of n = 4 biologically independent samples c.As in a. for all bison NHA2 DN structural and oligomerization mutants d.As in a, for bison NHA2 DN ion-binding site mutants.Errors bars represent the mean values ± s.d. of n = 4 biologically independent samples.e.

Supplementary Data Table 1: MD simulations.
TM-1-GFP in the presence of DDM/CHS.Data presented are normalized mean fluorescence as mean values ± s.e.m of n = 3 technical repeats; the apparent TM was calculated with a sigmoidal 4-parameter logistic regression function.