|
Extended Data Figure 2. Cryo-EM analysis of TMEM16A reconstituted in nanodiscsa, A representative cryo-EM micrograph of nanodisc-reconstituted TMEM16A. Green circles indicate individual particles. b, Representative 2D class averages from boxed particles with 256-pixel box size (261.12 Å). c, Eular angle distribution of all particles included in the final 3D reconstruction. The size of the spheres is proportional to the number of particles seen from that specific orientation. d, FSC curves of two independently refined maps before (blue) and after (red) post-processing in RELION. Curves with resolution corresponding to FSC = 0.143 are shown. e, Planar slices through the unsharpened EM density map at different levels along the channel symmetry axis. f, Local resolution of TMEM16A as estimated by RELION and shown with pseudo-color representation of resolution. g, Cross-validation using FSC curves of the density map calculated from the refined model versus half map 1 (work), versus half map 2 (free), and versus summed map. h, Directional FSC from different Fourier cones. Each curve indicates a different direction. i, Calculated resolution from different views. The directions are indicated as x, y, and z in the 3D resolution map. The highest and lowest resolutions are labeled with red and blue circles respectively. The green circle shows global average resolution.
|
|
Extended Data Figure 3. Cryo-EM analysis of TMEM16A solubilized in LMNGa, A representative cryo-EM micrograph of LMNG-solubilized TMEM16A. Green circles indicate individual particles. b, Representative 2D class averages from boxed particles with a 256 pixels box size (261.12 Å) and TMEM16A in complex with Fabs (bottom row, with the two right panels showing particles after subtraction of densities for Fabs). c, Eular angle distribution of all particles included in the final 3D reconstruction. The size of the spheres is proportional to the number of particles visualized from that specific orientation. d, FSC curves of two independently refined maps before (blue) and after (red) post-processing in RELION. Curves with resolution corresponding to FSC = 0.143 are shown. e, Planar slices through the unsharpened EM density map at different levels along the channel symmetry axis. f, Local resolution of TMEM16A as estimated by RELION and shown with pseudo-color representation of resolution. g, Cross-validation using FSC curves of the density map calculated from the refined model versus half map 1 (work), versus half map 2 (free), and versus summed map. h, Directional FSC (dFSC) from different Fourier cones. Each curve indicates a different direction. dFSC for TMEM16A alone in LMNG in grey (average in yellow); dFSC for combination of TMEM16A alone and with Fabs bound in LMNG in purple (average in red). i, Calculated resolution from different views (grey for combination of TMEM16A alone and with Fabs bound, yellow for TMEM16A alone). The directions are indicated as x, y, and z in the 3D resolution map. The highest and lowest resolutions are labeled with red and blue circles respectively. The green circle shows global average resolution.
|
|
Extended Data Figure 4. Cryo-EM densities of the ten transmembrane helices of TMEM16A, summary of cryo-EM data collection and processing, and summaries of sidechain assignmentsa, Representative cryo-EM densities of the ten transmembrane helices (TM1-TM10) of nanodisc-reconstituted TMEM16A (right) or LMNG-solubilized TMEM16A (left) are superimposed on the corresponding atomic model. The EM densities are shown in blue meshes for nanodisc-reconstituted TMEM16A, or green meshes for LMNG-solubilized TMEM16A, and the model is show as sticks and colored according to atom type (C: light grey; N: blue; O: red; S: yellow). b, Summary of cryo-EM data collection and model refinement. c, Summary of sidechain assignment of TMEM16A in nanodiscs. d, Summary of sidechain assignment of TMEM16A in LMNG.
|
|
Extended Data Figure 5. Atomic models of TMEM16A in two conformationsa–c, Ribbon diagrams of TMEM16A reconstituted in nanodiscs (in green and yellow) with lipids (in red), overlayed with EM density map (sharpened, in light grey). Two Ca2+ ions (orange spheres) are present in each monomer. d–f, Ribbon diagrams of TMEM16A solubilized in LMNG (in blue) with lipids (in red), overlayed with EM density map (sharpened, in light grey). One Ca2+ ion (orange sphere) is present in each monomer g–i, EM densities of nanodisc-reconstituted TMEM16A (unsharpened, in green and yellow) overlayed with digitonin-solubilized TMEM16A12 (in grey).
|
|
Extended Data Figure 6. Anion selectivity depends on residues lining the pore surrounded by TM3–8 but not TM10 residues at the dimer interfacea, Bi-ionic conditions for assessing the effect of the V595L mutation on permeability ratios. b, Effects of different substitutions of V595 on the permeability ratio PI-/PCl- (2.70 ± 0.09, N = 28 for WT; 4.00 ± 0.16, N = 9 for V595A; 3.49 ± 0.21, N = 6 for V595K; 3.81 ± 0.07, N = 7 for V595L; 3.48 ± 0.14, N = 7 for V595R). c, Effects of different substitutions of V595 on the permeability ratio PSCN-/PCl- (5.47 ± 0.21, N = 28 for WT; 9.96 ± 0.30, N = 9 for V595A; 6.64 ± 0.58, N = 6 for V595K; 7.86 ± 0.37, N = 7 for V595L; 6.30 ± 0.37, N = 7 for V595R). d, Permeability ratios determined in bi-ionic conditions for TMEM16A mutants. The exact n values (independent experimental samples from individually recorded HEK293 cells) are given for every experiment. The P-values are generated after a Dunnett’s posthoc test following one-way ANOVA. For these multiplicity adjusted P-values, values smaller than 0.0001 cannot be estimated precisely; Prism’s documentation suggests this approach is the most rigorous and conservative way to generate a P-value from a multiple comparison test53.
|
|
Extended Data Figure 7. Comparisons of extracellular loops and lipids in nanodisc-resonstituted and LMNG-solubilized TMEM16A. a–c, Lipids (in red) in the nanodisc-reconstituted TMEM16A (in green and yellow, overlayed with EM density map in light grey) (b, c), with helical distortions of TM6 near G640 (a). d–f, Lipids (in red) in LMNG-solubilized TMEM16A (in blue, overlayed with EM density map in light grey) (e, f), with the lower half of TM6 beyond G640 disordered and hence absent from the reconstruction (d). g–i, Extracellular domains of nanodisc-reconstituted TMEM16A (unsharpened, in green and yellow) overlayed with those of LMNG-solubilized TMEM16A (unsharpened, in blue). j–l, Extracellular TM5-TM6 and TM9-TM10 loops in ribbon diagrams for nanodisc-reconstituted TMEM16A (in green and yellow) overlayed with those of LMNG-solubilized TMEM16A (in blue).
|
|
Extended Data Figure 8. Data processing of TMEM16A in nanodisc (a) or LMNG (b). a, Data processing of nanodisc-reconstituted TMEM16A. Particle picking was performed with Gautomatch with templates from 2D classes from the LMNG dataset and generated 927,414 particles in total. All particles were extracted and binned 4 (pixel size is 4.08 Å) and then 2D classified. 341,875 particles from good 2D classes were used in 3D refinement with an initial model from the LMNG structure low pass filtered to 60 Å, giving rise to a 5.5 Å map. The 5.5 Å map was then low pass filtered to 10 Å as the initial model for 3D classification without applied symmetry for all particles with a 1.02 Å pixel size. Of the seven classes, two classes (15.92% and 11.15% of the 927,414 particles) gave maps with improved resolution (4.7 Å and 5.1 Å respectively) after 3D auto-refinement with C2 symmetry. These two classes were combined together yielding a total of 251,851 particles, and another 3D auto-refinement was run to generate the unmasked map at a resolution of 4.6 Å. The map was then masked to get the final map at resolution of 3.8 Å. b, Data processing of LMNG-solubilized TMEM16A. Approximately 4000 particles were manually picked and classified by 2D classification in SAMUEL to generate the templates for automatic particle picking with samautopick.py. 533,545 particles were identified after manually inspection. The crystal structure of nhTMEM16A10 (PDB: 4WIS) was converted to mrc with e2pdb2mrc.py and low pass filtered to 60 Å as the initial model. 44 of 200 2D classes were used for 3D auto-refinement with C2 symmetry. Since 3D classification failed for further separation, the reported resolution of the final map was 3.8 Å. To reduce anisotropy due to underrepresentation of side views, this dataset was merged with another dataset for Fabs bound to TMEM16A in LMNG. Starting with 338,705 particles from automatic particle picking, 4 of 40 2D classes (132,444 particles) were used for 3D auto-refinement. Then the Fab density for each particle was subtracted. The 132,444 subtracted particles without Fab density were combined with the 342,875 particles from all 5 classes of TMEM16A in LMNG dataset with resolution of 3.8 Å and processed for 3D auto-refinement to generate the unmasked map with resolution of 3.9 Å. This map was then masked to get the final map at resolution of 3.4 Å. Pixel sizes are shown in parenthesis for each class.
|
|
Extended Data Figure 9. Multiple open and closed states of TMEM16A calcium-activated chloride channel (CaCC) and involvement of pore-lining residues in channel gating. a, Reduction from 150 mM NaCl to 15 mM NaCl in the intracellular solution containing 1 μM or 1 mM Ca2+ caused identical shift of reversal potential of wildtype TMEM16A but not K584Q mutant channels in excised inside-out patch held at +80 mV and subjected to a ramp to −80 mV. Repeated independently 8 times for WT with similar results, and 5 times for K584Q with similar results. b, The K584Q mutation altered the permeability ratio PNa+/PCl- at 1 μM but not 1 mM Ca2+. N = 8 for WT, N = 5 for K584Q. c, Calcium sensitivity of channel activation of wildtype and mutant TMEM16A channels (number of independent experiments and P values are given in this table). Permeability ratios for mutants were compared to those of wild-type (WT) using one-way ANOVA followed by the Bonferroni post-hoc test for significance; **** designates p < 0.0001; *** designates p < 0.001; ** designates p < 0.005; data are presented as mean ± SEM.
|
|
Extended Data Figure 10. Sequence alignment of TMEM16A homologues. a, Sequences of TMEM16A homologues were analyzed by Clustal omega. Conserved residues are highlighted. Transmembrane helices are indicated above the sequences. The residues that were shown in this study (D550, N587, S635, Q705 and F712) and previous studies (R511, K584 and K599) to be crucial for selectivity are marked in orange and blue respectively. The residues (I546, Y589, I592, L639 and F708) that were shown in this study to be critical for gating are marked in green. The residues (N542 and V595) that contributed to both selectivity and gating property is marked in purple. The residues (E650, E698, E701, E730 and D734) important for Ca2+ binding are marked in red. b, Sequence alignment of mouse TMEM16A and nhTMEM16. Conserved residues are highlighted. Transmembrane helices of TMEM16A are indicated above the sequence.
|
|
Figure 1. Structure of TMEM16A in nanodiscs. a, EM density map of TMEM16A (sharpened, green and yellow) over a density map with a lower threshold (unsharpened) in light grey for nanodiscs. b, TM1–10 overlayed with EM density map of TMEM16A (sharpened, light grey). c–e, The channel pore (marked in orange) lined with N542 and V595 (in purple) that affect both anion selectivity and channel gating, five other residues that affect channel gating (in magenta) (see Fig. 4) and eight other residues that affect ion selectivity (in green) (see Fig. 3).
|
|
Figure 2. Differences in pore-lining helices and Ca2+ ion binding in the two structures. a–b, Density map of nanodisc-reconstituted TMEM16A (unsharpened, green and yellow) overlayed with that of LMNG-solubilized TMEM16A (unsharpened, blue). c, Conformational differences of pore-lining helices. d–f, Comparisons of TM3–8 around the pore (orange asterisk) in two structures. g–i, Two Ca2+ ions (orange spheres) coordinated by five acidic residues and one asparagine (N646) in nanodisc-reconstituted TMEM16A (g, h) versus one Ca2+ ion surrounding by four acidic residues in LMNG-solubilized TMEM16A (i, h) (Methods).
|
|
|
|
|
|
Figure 5. Different conformations of TMEM16A. a, Schematic showing the dimeric CaCC in a calcium-free closed conformation (top left panel), CaCC with single calcium occupancy that remains closed (top right panel) until depolarization causes it to open (bottom left panel), CaCC with double calcium occupancy that opens in a voltage-independent manner (bottom middle panel), and channel rundown following prolonged activation (bottom right panel). b–e, TMEM16A with C-terminal truncation yielded voltage-dependent current in 30 nM Ca2+ but voltage-independent current in 1 mM Ca2+ (b), while prolonged activation causes channel rundown (c) (N = 5; p = 0.03) (d) without significant change in Ca2+ sensitivity (EC50 = 0.11 ± 0.02 μM before exposure to 1 mM Ca2+, EC50 = 0.15 ± 0.04 μM after exposure to EC50 = 0.15 ± 0.04 μM, p = 0.13, N = 4, two-tailed Wilcoxon test) (e); mean ± SEM.
|