Expression and purification of NBCe1
Human N-terminally Strep(II)-tagged wild-type (wt) NBCe1-A in the pTT vector (Addgene) was transfected into HEK-293 cell (ATCC) monolayers using the calcium phosphate method. The cells were grown in Dulbecco’s Modified Eagle’s medium (Thermo Fisher Scientific) with 5% fetal bovine serum (Thermo Fisher Scientific) on 10-cm plates for ~24 h. The transfected cells from ~200 plates were pelleted by centrifugation at 2000 × g for 10 min. The cell pellets were solubilized with 1% Triton X-100 in buffer A (50 mM Tris-HCl, pH 7.5, 500 mM NaCl) supplemented with complete protease inhibitor cocktail (Thermo Scientific) for 30 min. Detergent insoluble material was removed by centrifugation (20,000 × g x 30 min) and the supernatant was loaded onto the 5-ml StrepTrap HP column (GE Healthcare). The column was washed with buffer A containing 0.01% n-dodecyl α-d-maltopyranoside (DDM, Anatrace), and bound protein was eluted with the same buffer containing 0.03% DDM supplemented with 2.5 mM D-desthiobiotin.
For preparation of samples for cryoEM, NBCe1-A was mixed with amphipol PMAL-C8 (Anatrace) at 1:3 (w/w) dilution with gentle agitation overnight at 4 °C. Detergent was removed with Bio-Beads SM-2 (Bio-Rad) incubated with samples for 1 h at 4 °C, and the beads were subsequently removed by centrifugation (2000 x g x 5 min). Amphipol containing protein was further purified on a Superose 6 column (GE Healthcare) in 20 mM Tris-HCl, pH 7.5, 150 mM NaCl. The peak corresponding to dimeric NBCe1-A was collected for cryoEM analysis.
Electron microscopy sample preparation and imaging
For electron microscopy of negatively stained protein, 2 µl of NBCe1-A sample (~0.1 mg ml−1) was applied to a glow-discharged EM grid covered with a thin layer of carbon film. After 10 s incubation, the grid was stained with 0.8% uranyl formate. EM micrographs were recorded on a TIETZ F415MP 16-megapixel CCD camera at 50,000 × nominal magnification in an FEI Tecnai F20 electron microscope operated at 200 kV. Micrographs were saved by 2 × binning to yield a calibrated pixel size of 4.41 Å.
For cryoEM, 3 μl of NBCe1-A (~0.4 mg ml−1) was applied to a glow-discharged Quantifoil 300-mesh R1.2/1.3 grid. The grid was blotted with filter paper to remove excess sample and flash-frozen in liquid ethane with FEI Vitrobot Mark IV. The frozen-hydrated grids were loaded into an FEI Titan Krios electron microscope operated at 300 kV for automated image acquisition with Leginon56. Micrographs (dose-fractionated movies) were acquired with a Gatan K2 Summit direct electron detection camera operated in the super-resolution mode (7676 × 7420 pixels) at a calibrated magnification of 36,764 × and defocus values ranging from −1.4 to −3.2 μm. A GIF Quantum LS Imaging Filter (Gatan) was installed between the electron microscope and the K2 camera with the energy filter (slit) set to 20 eV. The dose rate on the camera was set to ~8 e‒ pixel−1 s−1 and the total exposure time was 12 s fractionated into 48 frames of images with 0.25 s exposure time for each frame. A total of 3378 micrographs were collected.
The frame images of each micrograph were aligned and averaged for correction of beam-induced drift using MotionCor217. The local motion within a micrograph was corrected using 5×5 patches. Two average images, with and without dose-weighting, from all except the first frame were generated with 2×binning (final pixel size of 1.36 Å on the sample level) for further data processing. A total number of 3136 good micrographs were picked for image processing by visual inspection of the average images and power spectra after the drift correction.
The defocus values of the micrographs were measured on the dose-unweighted average images by CTFFIND457. The dose-weighted average images were used for particle picking and subsequent image processing. A total of 1,729,419 particles were automatically picked using Gautomatch [K. Zhang, MRC LMB (www.mrc-lmb.cam.ac.uk/kzhang/)] and boxed out in 192 × 192 pixels.
The particles were first subjected to 3D classifications by GPU-accelerated RELION-258,59 using an oval-shaped disk low-pass filtered to 60 Å as the initial model. The particles were separated into six classes for 69 iterations and the best class contained 687,322 particles, which were then sent to a second run of 3D classification to be sorted into five classes. A number of 214,196 particles were found in the best class after 60 iterations of the second run of 3D classification. A twofold symmetry was enforced in the above 3D classifications. 3D classifications without symmetry (i.e., C1) were also tested but no further improvement on the separation of heterogeneity was found. The 214,196 particles were then sent to 2D classification using RELION-2 and sorted into 200 classes for 72 iterations.
A total number of 184,561 particles in 134 good classes that show discernible high-resolution features such as α-helices were combined for 3D auto-refinement. The 3D auto-refinement was performed using a spherical mask (180 Å in diameter) by RELION-2. The resolution was estimated to be 3.9 Å by relion_postprocess in RELION-2 with a soft auto-mask using the ‘‘gold-standard’’ FSC at 0.143 criterion. The final cryoEM map was sharpened with B-factor and low-pass filtered to the stated resolution using RELION-2. The local resolution was calculated by ResMap60 using two cryoEM maps to independently refine from halves of the data.
Model building and refinement and structure visualization
The overall 3.9 Å cryoEM structure was at a sufficient resolution to build a de novo atomic model for TMs 1–3, TMs 5–14, amphipathic helices H1–4 and short (single turn) cytoplasmic helix H5 in Coot61.
Aromatic residues in many of these TMs were clearly visible in our cryoEM structure and were used as landmarks for our model building, especially in areas where multiple aromatic residues are in close proximity. TMs 1–3 along with the loop between each helix were built de novo starting at residues Phe431-Tyr433 and continued with side chain placements into the protrusions visible in the helical densities. Aromatic residues including Phe443, Phe461, Phe471, Phe474, Phe791, Phe496, and Phe498 provided checkpoints for building TMs 1–3. The side chain positions were temporarily refined to the cryoEM helical densities using the real-space refine zone feature in Coot with planar peptide, Ramachandran and α-helix restraints. The loops between the TMs were similarly refined in Coot without the α-helix restraints. This procedure was repeated for helices TMs 5–14. TM5 were built in reverse from C- to N-terminus starting at aromatic residues Tyr666–Tyr667 and guided by residues Phe539, Phe544, Phe550, Phe552, Tyr554, and Phe557. TMs 6 and 7 were built starting with residues Tyr674-Phe675, located at the small loop region between the two helices. CryoEM densities for Phe653, Phe656, Tyr660, Phe669, Phe686, and Phe695 supported our model building for TMs 6 and 7. Model building for TMs 8–10 and the loops originated from residues Tyr775 and His776 was done with guidance from residues Trp781, Trp796, Tyr797, and His807. TMs 11–14 were built starting with residues Tyr862, Phe865, and Tyr867 from TM12. Aromatic residues located in both the helices and loop regions between TMs 11–14 aided the model building of these regions. TM4 was slightly challenging to model as some of the aromatic residue densities were missing. Using Phyre262 to predict the secondary structure based on the sequence of TM4, we were able to model this helix with guidance from residues Trp512, Trp516, and Phe519. Helix H1 contains two sites of consecutive aromatic residues (Phe112-Phe113 and Phe117-Tyr118) that were used as starting points for de novo model building. Helices H2–H4 and cytoplasmic helix H5 were simultaneously built during the model building of the other TMs.
In addition to the TMs, we were able to build a model for a portion of EL3 located between TMs 5 and 6. A polyalanine chain was built between residues Tyr566-Ser582 and Val638-Asp647 of EL3 since we were able to accurately trace the backbone for this dynamic loop.
The monomeric model of NBCe1-A was duplicated across the twofold symmetry in Chimera63 and saved as one dimeric model. The dimeric model of NBCe1-A was subjected to further global refinement with simulated annealing using the real-space refinement (RSR) feature in the PHENIX software package64. After five iterations in RSR with simulated annealing, the model was further analyzed in MolProbity65. Residues with poor rotamer or considered a Ramachandran outlier were fixed in Coot. Another round of RSR and MolProbity analysis were performed. The Molprobity score of the final atomic model was 2.11 with a 9.23 clashscore, and without Ramachandran outliers, Cβ deviations, bad bonds, poor rotamers, and cis-peptides. The cross-correlation coefficient between the final atomic model and the cryoEM density calculated from RSR is 0.801. The structure is visualized in UCSF Chimera63. Protein sequences were aligned using Clustal Omega66.
The final NBCe1-A atomic model was superposed onto AE122 (PDB: 4YZF) model using the superpose command from the CCP4 package67. The superimposed figures between NBCe1-A and AE1 were displayed in UCSF Chimera. Permeation pathway analysis was calculated using the HOLE2 software23 and displayed in Pymol68. The electrostatic analysis was calculated using APBS69 once the model was modified in PDB2PQR70. The electrostatic map was visualized in UCSF Chimera.
Mutations were introduced in wt-NBCe1-A using the QuikChange Lighting Site-Directed Mutagenesis Kit (Agilent Technologies; Supplementary Table 1). Twenty-seven individual constructs were studied. All sequences were verified using the University of California Los Angeles Genotyping and Sequencing Core using the BigDye terminator kit version 3.1 (Invitrogen Life Technologies) and resolved with a 3730 XL ABI sequencer (Applied Biosystems, Life Technologies).
Cell surface biotinylation and immunocytochemistry
The constructs were initially assessed for plasma membrane expression using sulfo-NHS-SS-biotin plasma membrane labeling and immunocytochemistry using a previously characterized rabbit polyclonal NBCe1-A antibody25. Twenty-four hours following transfection of HEK-293 cells with wt-NBCe1-A and various NBCe1-A mutants using Lipofectamine 2000 (Thermo Fisher Scientific), plasma membrane proteins were labeled using sulfo-NHS-SS-biotin (Pierce) and pulled down using streptavidin-agarose resin according to the manufacturer’s protocol. Immunoblots of plasma membrane pulled down proteins were probed with the NBCe1-A antibody (1:10,000 dilution) and cell lysates were probed with a GAPDH (A-3) antibody (sc-137179, Santa Cruz Biotechnology; 1:5000 dilution). Cell surface expression was normalized to GAPDH. The experiments were done in triplicate. All constructs except one (483SST-GFS/754D-E/798V-S/800A-T/803I-R) had normal membrane expression (Fig. 4l) For immunocytochemistry (Fig. 4m), 24 h post-transfection cells grown on coated coverslips were permeabilized with methanol (room temperature), washed with phosphate-buffered saline (PBS) and incubated with the NBCe1-A antibody (1:100 dilution in PBS; room temperature for 30 min). Following another PBS wash, the coverslips were incubated with Alexa Fluor 594 goat anti-rabbit secondary antibody (30 min at room temperature; 1:500 dilution in PBS; Thermo Fisher Scientific). The antibody was washed off with PBS and fluorescence images of the cells were captured with a PXL charge-coupled device camera (model CH1; Photometrics) coupled to a Nikon Microphot-FXA epifluorescence microscope (Melville).
Base transport activity assays
HEK-293 cells grown on coated coverslips were transfected with wt-NBCe1-A, various NBCe1-A mutants, and wt-AE1 and studied 24 h later. Intracellular pH (pHi) was monitored with a microscope-fluorometer in cells loaded at room temperature for ~20 min with the fluorescent intracellular pH (pHi) probe BCECF (using esterified BCECF-AM; Life Technologies). The intracellular fluorescence (excitation ratio 500 nm/440 nm; emission 530 nm) was acquired every 0.5 s and calibrated at the end of each experiment with valinomycin (Sigma-Aldrich) and nigericin (Sigma-Aldrich). To determine the Na+-driven base flux, the initial rate of change of [Hin+] (d[Hin+]/dt) was measured using a linear curve fit in the first 5–10 s following the increase in bath Na+ from zero to 140 mM. A similar analysis was done in the Cl−-driven base flux protocol following the increase in bath Cl− from zero to 119 mM. Base flux was calculated as d[Hin+]/dt x βT, where the total cell buffer capacity (βT) is equal to the intrinsic buffer capacity (βi
) plus the HCO3−-buffer capacity. The data were background (mock-transfected, i.e., empty plasmid) subtracted and depicted as percentage of the wild-type flux.
Na+-driven base flux: The cells were initially bathed in Na+-free buffer: 115 mM TMACl, 2.5 mM K2HPO4, 1 mM CaCl2, 1 mM MgCl2, 5 mM glucose, 25 mM TMA-HCO3, pH 7.4, and 30 µM EIPA (to block endogenous Na+/H+ exchange) until a steady-state was achieved. Following the addition of Na+ (115 mM NaCl, 2.5 mM K2HPO4, 1 mM CaCl2, 1 mM MgCl2, 5 mM glucose, 25 mM NaHCO3, pH 7.4, with 30 µM EIPA), the Na+-driven base flux was measured. The data are the mean of 3−8 experiments.
Cl−-driven base flux: The cells were initially bathed in a Cl−-free buffer containing 115 mM Na gluconate, 2.5 mM K2HPO4, 7 mM Ca gluconate, 1 mM Mg gluconate, 5 mM glucose, 25 mM NaHCO3, pH 7.4, and 30 μM EIPA. After a steady-state was achieved, a Cl−-containing buffer (115 mM NaCl, 2.5 mM K2HPO4, 1 mM CaCl2, 1 mM MgCl2, 5 mM glucose, 25 mM NaHCO3, pH 7.4, and 30 μM EIPA) was added and the Cl−-driven base flux was measured. The data are the mean of 4–8 experiments.
Whole-cell patch clamping was used to compare the electrogenicity of wt-NBCe1-A and an 483SST-GFS/754D-E/798V-S/800A-T/803I-R mutant expressed in HEK-293 cells as described28. Steady-state currents were measured using a holding potential of −55 mV, with a series of 400-ms voltage pulses (increment of 10 mV from –100 to + 30 mV). The data are the mean of 3–10 experiments.
−-dependent current: The patch solution contained 125 mM CsOH, 10 mM NaOH, 100 mM gluconic acid, 1 mM CaCl2, 10 mM TEACl, 10 mM EGTA, 10 mM HEPES, pH 7.4, and the bath solutions contained 120 mM NaCl, 1.5 mM CaCl2, 10 mM CsCl, 10 mM HEPES, pH 7.4, with 25 mM Na gluconate or 25 mM NaHCO3 (replacing 25 mM Na gluconate).
Na+-driven current: The patch solution contained 125 mM CsOH, 10 mM NaOH, 100 mM gluconic acid, 1 mM CaCl2, 10 mM TEACl, 10 mM EGTA, 10 mM HEPES, pH 7.4, and the bath solutions contained 1.5 mM CaCl2, 10 mM CsCl, 10 mM HEPES, pH 7.4, with 120 mM TMACl and 25 mM choline HCO3, or 120 mM NaCl (replacing 120 mM TMACl) and 25 mM NaHCO3 (replacing 25 mM choline HCO3).
Cl−-driven current: The patch solution contained 105 mM CsOH, 70 mM gluconic acid, 1 mM CaCl2, 10 mM TEACl, 10 mM EGTA, 10 mM HEPES, 10 mM NaHCO3, 15 mM choline HCO3, pH 7.4, and the bath solutions contained 1.5 mM CaCl2, 9 mM CsCl, 10 mM HEPES, 10 mM NaHCO3, 15 mM choline HCO3, pH 7.4, with 120 mM Na gluconate or 120 mM NaCl (replacing 120 mM Na gluconate).
The final cryoEM density map of human NBCe1-A has been deposited to the Electron Microscopy DataBank (EMDB) under the accession code EMD-7441. The final atomic model was deposited into the Protein Data Bank (PDB) under the accession code 6CAA. All other relevant data are available in this article and its Supplementary Information files, or from the corresponding authors upon reasonable request.