Structure and electromechanical coupling of a voltage-gated Na+/H+ exchanger – Nature

Expression and purification of SLC9C1

The gene encoding the S. purpuratus SLC9C1 protein (NP_001091927.1), referred to as SLC9C1, was synthesized (Thermo Fisher Scientific) and subcloned into the pcDNA3.1(+) vector followed by a TEV (Tobacco Etch Virus nuclear-inclusion-a endopeptidase) recognition site, eGFP and a TwinStrep tag at the C terminus. An Asp238Ala variant of SLC9C1 was further constructed using the Quickchange kit (Life Technologies). The recombinant vectors were verified by DNA sequencing and purified using the PureLink Giga Plasmid Purification Kit (Invitrogen). The serum-free suspension-adapted Freestyle HEK293F cells (Invitrogen) were used for expression and cultured in FreeStyle 293 expression medium (Life Technologies), shaken at 140 rpm in a humidified incubator at 37 °C and under a 8% CO₂ atmosphere. Cell growth and viability was assessed using a Countess II FL automated cell counter (Invitrogen). The cells were transiently transfected using 20 kDa linear polyethylenimine (PEI) (PolySciences). Transfections were performed using a final DNA concentration of 1 μg ml⁻¹. In brief, cloned Slc9c1-gfp containing pcDNA3.1(+) vector was diluted in FreeStyle medium and vortexed for 10 s. PEI-20KDa was diluted using the same medium at a 1:2 (w/w) DNA:PEI ratio and vortexed for 10 s before being added to the DNA-containing medium. The DNA:PEI mixture was incubated for 10 min at room temperature and added to the cells at a cell density of 1–1.5 × 10⁶ cells per ml. Then, 24 h after transfection, sodium butyrate was added to a final concentration of 5 mM. Typically, 48 h later 10 l of cell culture was collected by centrifugation (3,000g at 4 °C for 10 min). The cells were subsequently resuspended in 100 ml cell resuspension buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl) supplemented with 1 tablet of cOmplete, EDTA-free protease inhibitor cocktail (Roche) and subsequently lysed using a homogenizer. The membranes were isolated from the resulting pellet after ultracentrifugation at 195,000g and 4 °C for 1 h.

For structural studies using cryo-EM, the isolated membranes were solubilized in solubilization buffer containing 2% (w/v) glyco-diosgenin (GDN, Anatrace), 50  mM Tris-HCl pH 7.5, 150 mM NaCl, 0.003% brain fraction 7 (Sigma-Aldrich) and 10% glycerol for 2 h at 4 °C with mild agitation. The solution was subsequently cleared by ultracentrifugation 195,000g and 4 °C for 1 h. The supernatant was applied to GFP nanobody-coupled CNBr-activated Sepharose 4B resin (Cytiva) and incubated for 1 h at 4 °C, and subsequently washed with 10 column volumes (CV) of wash buffer consisting of 0.02% GDN (w/v), 20 mM Tris-HCl pH 7.5, 150 mM NaCl and 0.003% brain fraction 7 (w/v). The resin-bound GFP-TwinStrep-His₈ residue was cleaved by TEV-His₈ protease present in excess amounts during mild agitation overnight at 4 °C. The cleaved protein was collected and concentrated using 100 kDa MW cut-off spin concentrators (Amicon, Merck-Millipore). The concentrated SLC9C1 protein was subsequently injected onto a pre-equilibrated Superose 6 increase 10/300 column (GE Healthcare) in GDN SEC buffer consisting of 50 mM Tris-HCl pH 7.5, 150 mM NaCl and 0.02% (w/v) GDN, and the peak fraction at 13.6 ml was collected using the Äkta system. Owing to the low levels of produced protein, typically around 0.3 mg of purified protein could be isolated from 10 l of HEK293 cell culture.

For reconstitution of SLC9C1 into nanodiscs, SLC9C1 was purified as described above and concentrated to approximately 1 mg ml⁻¹. Concentrated SLC9C1 was subsequently mixed with yeast polar lipids (Sigma-Aldrich) and MSP1E2 nanodiscs at a molar ratio of 1:4:200 for 30 min at room temperature. To this solution, 50 mg of preactivated SM-2 biobeads (Bio-Rad) was added and allowed to incubate overnight at 4 °C. The next day, 50 mg of fresh biobeads were added and further incubated for 1 h. The biobeads were discarded and the solution was centrifuged at 14,000 rpm for 5 min. The reconstituted nanodiscs were further processed by size-exclusion chromatography on a Superose 6 increase column in a buffer containing 50 mM Tris pH 7.5 and 150 mM NaCl. Peak fractions corresponding at 13.6 ml were collected.

SSM-based electrophysiology

For SSM-based electrophysiology measurements, purified protein in GDN SEC buffer was reconstituted into liposomes. In brief, yeast polar lipids (Avanti) were solubilized in a 2:1 (v/v) chloroform:methanol solution and thin-layered using a rotary evaporator (Hei-Vap Core, Heidolph Instruments). The lipids were thoroughly resuspended in 10 mM MES-Tris pH 7.5, 5 mM MgCl₂ buffer at a final concentration of 10 mg ml⁻¹. Unilamellar vesicles were prepared by extruding the resuspended lipids using an extruder (Avestin) with 400 nm polycarbonate filters (Whatman). The 500 μg of extruded lipids were destabilized by adding sodium cholate to a final concentration of 0.65% (w/v) (Sigma-Aldrich). Then, 20 μg of either the SEC-purified SLC9C1 or D238A variant was added to the destabilized liposomes at a desired LPR of 125:1 (w/w) (or, as stated in Supplementary Fig. 3, the protein ratio was tested between LPR 5 to 500) and incubated for 5 min at room temperature. Purified rat GLUT5 and the horse NHE9 ion-binding site mutant D244A/N243A were prepared as previously described^18,51 and added to destabilized liposomes at an LPR of 5:1 (w/w) and incubated for 5 min at room temperature. The samples were individually applied to the PD SpinTrap G-25 desalting column (Cytiva) to remove excess detergent and the reconstituted proteoliposomes were collected in a final volume of 500 μl. The sample was diluted to final lipid concentration of 1 mg ml⁻¹ in 10 mM MES-Tris at the desired pH, ranging from 6.5, 7.5 to 8.5 in a buffer otherwise consisting of 5 mM MgCl₂, 5 mM NaCl and 5 mM KCl buffer, and then flash-frozen at −80 °C. Proteoliposomes were diluted 1:1 (v/v) with non-activating buffer made up of 10 mM MES-Tris at the desired pH, 200 mM choline chloride, 5 mM MgCl₂ and sonicated using a bath sonicator before sensor preparation. All electrophysiology measurements were recorded using the SURFE²R N1 Nanion instrument. Sensor preparation for SSM-based electrophysiology on the SURFE²R N1 system (Nanion Technologies), was performed according to the manufacturer’s instructions³⁰. For symmetrical pH measurements and different LPR comparisons, 3 mm sensors were used, otherwise 1 mm sensors were used for all other experiments. In brief, sensors were incubated with 50 μl of a 0.5 mM octadecanethiol solution (Sigma-Aldrich) and kept at room temperature for 30 min. After the incubation, the sensors were washed with 100% isopropanol (Sigma-Aldrich) and deionized water. A 1 μl droplet of lipid solution (1,2-diphytanoyl-sn-glycero-3-phosphatidylcholine (Avanti) in n-decane (Sigma-Aldrich) at a final concentration of 7.5 mg ml ⁻¹) was added to the gold surface, followed by 50 μl of non-activating buffer. Then, 5 μl of the diluted proteoliposomes was added to the sensors and the sample was then centrifuged at 3,000g for 30 min to ensure the complete adhesion of the proteliposomes to the surface. SLC9C1 was activated by changing from non-activating buffer to an activating buffer containing the NaCl. To measure the binding kinetics, 200 mM choline chloride in non-activating buffer was replaced by x mM NaCl and (200–x mM) choline chloride in the activating buffer. The peak with the largest absolute value was recorded as the peak current. Three sensors were prepared for each sample and at least two replicate measurements were made for each sensor. Current traces were corrected for small offset differences (<50 pA). The current showed pre-steady-state ion translocation rather than steady-state. The decay time constant τ was determined using a monoexponential fit (one-phase decay) in GraphPad Prism using the slope calculated between the highest to the lowest peak plateau at around 1.3 s for each sensor.

Lipid–protein interactions assessed by FSEC

SLC9C1 was repurified in the detergent n-dodecyl-β-d-maltopyranoside (DDM)/cholesteryl hemisuccinate Tris salt (CHS) without GFP cleavage. In brief, the SLC9C1–GFP-containing membranes were solubilized in solubilization buffer containing 1% (w/v) DDM (Glycon technologies), 0.2% CHS (Sigma-Aldrich), 50 mM Tris-HCl pH 7.5, 150 mM NaCl and 0.003% brain fraction 7 (w/v) for 1 h at 4 °C with mild agitation. The solution was subsequently cleared by ultracentrifugation 195,000g, 4 °C for 1 h. The supernatant was applied to Strep-Tactin Sepharose resin (IBA) and incubated for 1 h at 4 °C and subsequently washed with 10 CV of wash buffer consisting of 0.05% DDM (w/v), 0.01% (w/v) CHS, 20 mM Tris-HCl pH 7.5 and 150 mM NaCl. The resin-bound SLC9C1–GFP was eluted in 2 CV of elution buffer containing 0.03% DDM (w/v), 0.006% (w/v) CHS, 20 mM Tris-HCl pH 7.5, 150 mM NaCl and 1× BXT (IBA). The elution was applied to a pre-equilibrated PD-10 desalting column (Cytiva) using reaction buffer consisting of 20 mM Tris pH 6.5, 150 mM NaCl and 0.03% DDM. The samples were concentrated to a final concentration of 0.5–0.75 mg ml⁻¹.

To characterize the thermostability and lipid thermal stabilization of SLC9C1–GFP, ~0.05 mg of samples were incubated for 10 min at 4 °C without lipids as a control and with the individual lipids at a final concentration of 1 mg ml⁻¹ in reaction buffer. Stock solutions of the respective lipids (1,2-dioleoyl-sn-glycero-3-phosphate, sodium salt (DOPA, Avanti, 840875P), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE, Avanti, 850725P), 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC, Avanti, 850375P), 1,2-dioleoyl-sn-glycero-3-phospho-(1′-rac-glycerol), sodium salt (DOPG, Avanti, 840475P) and 1,2-dioleoyl-sn-glycero-3-phospho-l-serine, sodium salt (DOPS, Avanti, 840035P) were prepared by solubilization in 10% (w/v) DDM to a final concentration of 30 mg ml⁻¹. Subsequently, β-d-octyl-glucoside (Anatrac) was added to a final concentration of 0.1% (w/v) and the sample aliquots of 120 μl were heated for 10 min either at 4 °C or 45 °C after the addition of a final concentration of 0.33% (w/v) DDM or individual lipids in a PCR thermocycler (Veriti, Applied Biosystems). The heat-denatured material was subsequently pelleted at 18,000g for 30 min at 4 °C. The supernatant was injected into the Bio-Rad ENrich 650 column, pre-equilibrated with 20 mM Tris, 150 mM NaCl and 0.03% (w/v) DDM, analysed using fluorescence-detection size-exclusion chromatography (FSEC; Shimadzu HPLC LC-20AD/RF-20A)^52,53. The FSEC traces of purified SLC9C1–GFP were recorded without lipid addition at 4 °C and with and without lipid addition at 45 °C. Data used for bar chart comparisons were the fluorescence intensity of the FSEC dimerization peak in response to different lipids or additional DDM from three technical repeats.

Cryo-EM sample preparation and data acquisition

GDN sample

For structural studies, 3.5 μl of 3 mg ml ⁻¹ purified SLC9C1 in GDN was applied to a freshly glow-discharged Quantifoil R1.2/1.3 Cu300 grid with a blot force of 0, a wait time of 15 s wait and a blot time of 3 s using the Vitrobot Mark IV (Thermo Fisher Scientific) at 4 °C and 100% humidity, and plunge-frozen into liquid ethane. Cryo-EM datasets were collected on the Titan Krios G2 electron microscope operated at 300 kV equipped with a GIF (Gatan) and a K3 BioQuantum direct electron detector (Gatan) in counting mode. The video stacks were collected at ×130,000, corresponding to a pixel size of 0.6645 Å at a dose rate of 13–14 e⁻ per physical pixel per second. The total exposure time for each video was 1.8 s, leading to a total accumulated dose of 58.45 e⁻ Å⁻² and was fractionated into 40 frames. All videos were recorded with a defocus range of −0.6 to −2.0 µm

Nanodisc sample

For structural studies, 3.5 μl of 1 mg ml⁻¹ SLC9C1 in nanodiscs was blotted on quantifoil 2/1 grids with a blot force of 0, a wait time of 5 s and a blot time of 3 s using the Vitrobot Mark IV maintained at 4 °C and 100% humidity. Cryo-EM datasets were collected on the Titan Krios G2 electron microscope operated at 300 kV equipped with a GIF (Gatan) and a K3 BioQuantum direct electron detector (Gatan) in counting mode. The video stacks were collected at ×130,000 corresponding to a pixel size of 0.6645 Å at a dose rate of 14.37 e⁻ per physical pixel per second. The total exposure time for each video was 1.8 s, leading to a total accumulated dose of 58.61 e⁻ Å⁻² and was fractionated into 40 frames. All videos were recorded with a defocus range of −0.6 to −2.0 µm.

GDN sample with cAMP

Prior to blotting, a final concentration of 0.1 mM cAMP and 0.1 mM MgCl₂ was added to 3 mg ml⁻¹ of SLC9C1 purified in GDN.  A sample volume of 3.5 μl was blotted onto the Quantifoil R 2/1 grid with a blot force of 0, a wait time of 5 s and a blot time of 3 s using the Vitrobot Mark IV maintained at 4 °C and 100% humidity. Cryo-EM datasets were collected on the Titan Krios G2 electron microscope operated at 300 kV equipped with a GIF (Gatan) and a K3 BioQuantum direct electron detector (Gatan) in counting mode. The video stacks were collected at ×130,000 magnification corresponding to a pixel size of 0.6645 Å at a dose rate of 13.30 e⁻ per physical pixel per second. The total exposure time for each video was 1.7 s, leading to a total accumulated dose of 51.21 e⁻ Å⁻² and was fractionated into 40 frames. All videos were recorded with a defocus range of −0.6 to −2.0 µm. All statistics of cryo-EM data acquisition are summarized in Extended Data Table 1.

Image processing

GDN sample

The dataset for the GDN sample was processed by CryoSparc⁵⁴. Dose-fractionated video frames were aligned using patch motion corrections, and the contrast transfer function (CTF) was estimated using ‘patch CTF estimation. Exposures with an estimated CTF fit resolution of lower than 5.0 Å were discarded. Automated particle picking was performed using a blob picker from 3,000 random images to generate 2D templates for template-based particle picking. A total of 2,039,463 particles picked with a box size of 512 pixels were extracted and Fourier-resampled to 128 pixels. After several rounds of 2D classification, 312,610 particles were used for ab initio model building followed by several rounds of non-uniform refinement. 3D variability cluster analysis was performed to generate five clusters, which were used as the input for further rounds of hetero refinement applying C₂ symmetry. A final map was produced by performing local refinement at 3.23 Å. To improve the map density of the VSD, the final particle stack was expanded using symmetry expansion. Local refinement was performed using a monomer mask which improved the overall density and generated gold standard FSC resolution estimation at 3.2 Å. To further improve the density of VSD, local refinement was performed using a mask encompassing only the VSD and cytosolic domains.

Nanodisc sample

The dataset for the nanodisc sample was processed by cryoSparc⁵⁴. Does-fractionated video frames were aligned using patch motion corrections, and the CTF was estimated using patch CTF estimation. Exposures with an estimated CTF fit resolution worse than 4.0 were discarded. Automated particle picking was performed using a blob picker from 500 random images to generate 2D templates for template-based particle picking. A total of 1,737,146 particles were picked and extracted with a box size of 550 pixels and Fourier-resampled to 400 pixels. After several rounds of 2D classification, 276,031 remaining particles were processed for multimodel ab initio model building followed by several rounds of hetero refinements using C₂ symmetry. 3D variability cluster analysis was performed to generate three clusters, which were used as the input for further rounds of hetero refinement applying C₂ symmetry. A final map was produced by performing non-uniform refinement at 3.08 Å containing 97,279 particles. To improve the density for the VSD domain, an intermediate map with C₂ symmetry produced after 3DVA in cryoSPARC⁵⁴ and hetero refinement was used to produce five additional ab initio classes. From this, a class with the best features for the VSD domain was processed for a round of hetero refinement along with a C₂map to segregate particles with clearer density for the VSD domain. After further rounds of hetero refinement in C₁ symmetry, the best volume was processed for sequential masked local refinement, first by masking out the density for the nanodisc, followed by masking out the density for one monomer corresponding to a poor VSD to finally yield a map with enhanced features for a single monomer with its corresponding VSD at an overall resolution of 3.30 Å containing 100,697 particles.

GDN and cAMP sample

The dataset for the GDN sample in the presence of cAMP was processed in CryoSPARC⁵⁴. Dose-fractionated video frames were aligned using patch motion corrections and CTF was estimated using patch CTF estimation. Exposures with an estimated CTF fit resolution worse than 5.0 were discarded. Particles were picked using an automated blob picker with a picking diameter of 180 Å. Picked particles were extracted with a box size of 448 pixels and Fourier-cropped to 128 pixels and subsequently used for several rounds of 2D classification. Particles were then rescaled to a box size of 240 pixels and used for ab initio maps. The best ab initio model along with ‘junk’ models were processed for several rounds of heterogeneous refinement. Particles corresponding to the best resulting map were rescaled to a box size of 320 pixels and processed for a round of non-uniform refinement in C₁ resulting in a map with an overall resolution of 3.68 Å containing 94,091 particles. 3DVA of the final reconstruction was performed in cryoSPARC⁵⁴.

Model building and refinement of SLC9C1 in GDN, nanodiscs and GDN with cAMP

The SLC9C1 homology model was taken from AlphaFold⁵⁵ and each domain was extensively refitted into the C₂ GDN map using the fit in map utility of Chimera⁵⁶ and rebuilt extensively in Coot⁵⁷. The structure was refined using real-space refinement in Phenix⁵⁸. The side chains for the VSD were built in Coot⁵⁷ using the focused refinement maps obtained after symmetry expansion and masked local refinement, shown in Supplementary Video 1. The built SLC9C1 VSD monomer was refitted into the C₂ maps of the SLC9C1 homodimer and processed for rigid-body refinement using Phenix⁵⁸. The final refinement statistics are shown in Extended Data Table 1. For model building of the SLC9C1 structure in nanodiscs, the SLC9C1 structure in GDN was first fitted into the C₂ nanodisc cryo-EM map using the fit in map utility of Chimera⁵⁶ and manually adjusted in Coot⁵⁷. As the nanodisc cryo-EM map had weak density for the VSDs, the side chains were trimmed at the Cβ position in the final model. The SLC9C1 nanodisc structure was refined with real-space refinement in Phenix⁵⁸; the final refinement statistics are shown in Extended Data Table 1. For model building of the SLC9C1 structure in GDN with cAMP, the apo SLC9C1 structure in GDN was fitted into the cryo-EM map using the fit in map utility of Chimera⁵⁶. The structure was manually inspected and trimmed to remove regions that were not supported by cryo-EM map density. Iterative model building into the map was performed using Coot⁵⁷, and the structures were refined using real-space refinement by Phenix⁵⁸; final refinement statistics are shown in Extended Data Table 1. Density for VSDs was poor in the GDN maps with cAMP and, as such, only one VSD domain was built with residues trimmed at the Cβ position. To illustrate the structure and cryo-EM maps, PyMol⁵⁹ and Chimera⁵⁶ were used.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.