Allosteric mechanism for KCNE1 modulation of KCNQ1 potassium channel activation

Figures

• 

  Figure 1.. KCNQ1 channel architecture and sequence of KCNE proteins. (A) Topology diagram of the KCNQ1-KCNE1 channel complex. The KCNQ1 voltage-sensing (VSD, helix S1-S4), pore-forming (PD, S5–P–S6), and cytosolic domains (helix HA-HD) are colored green, blue, and gray, respectively. KCNE1 exhibits a single-span transmembrane domain (TMD) that is flanked by intra- and extracellular domains containing helical segments. (B) Amino acid sequence alignment of KCNE1 and KCNE3. Similar and identical amino acid residues are colored light and dark gray, respectively. The TMD region is indicated by a black box. The activation motif regions in KCNE1 and KCNE3 are highlighted in red.

• 

  Figure 2.. Oxidation state-dependent electrophysiology measurements indicate that KCNQ1 V141 and I274 are close to L48 and L45 in KCNE1, respectively. (A) Whole-cell currents (left) and average current-voltage (I–V) relationships (right) of CHO-K1 cells transiently expressing KCNQ1 I274C and KCNE1 L45C. Cells were exposed to control bath solution containing DTT or Cu-phenanthroline (Cu-phen.). (mean ± SEM, Control n = 6, DTT n = 6, Cu-phen n = 5). (B) Whole-cell currents (left) and average I-V relationships (right) of CHO-K1 cells expressing KCNQ1 V141C and KCNE1 L48C, which were exposed to control bath solution, DTT, or Cu-phenanthroline, respectively. (Control n = 7, DTT n = 5, Cu-phen n = 5). Solid lines represent fits with a Boltzmann function (Itail/Itailmax = (1-IBottom) / (1+exp[(V1/2app-V)/k]) + IBottom) and the parameters of the fit are summarized in Supplementary file 1 – Table 1. Control measurements of KCNQ1 WT, V141C, and I274C with and without KCNE1 under reducing (+DTT) and oxidizing (+Cu-phen.) conditions are displayed in Figure 2—figure supplement 1 and Figure 2—figure supplement 2. Figure 2—source data 1.Excel file with numerical electrophysiology data used for Figure 2.

• 

  Figure 2—figure supplement 1.. Electrophysiology measurements of KCNQ1 WT, V141C, and I274C under reducing and oxidizing conditions. (A) Whole-cell currents (left) and average I-V relationships (right) of CHO-K1 cells expressing KCNQ1 WT, which were exposed to control bath solution, DTT, or Cu-phenanthroline (Cu-phen.), respectively. (mean ± SEM, Control n = 9, DTT n = 5, Cu-phen. n = 5) (B) Whole-cell currents and average I-V relationships of CHO-K1 cells expressing KCNQ1 V141C, which were exposed to control bath solution containing DTT or Cu-phenanthroline. (Control n = 13, DTT n = 7, Cu-phen. n = 5) (C) Whole-cell currents and average I-V relationships of CHO-K1 cells expressing KCNQ1 I274C measured with or without DTT or Cu-phenanthroline in the bath solution. (Control n = 12, DTT n = 7, Cu-phen. n = 7) Solid lines represent fits with a Boltzmann function (Itail/Itailmax = (1 - IBottom) / (1+exp[(V1/2app - V)/k]) + IBottom) and the parameters of the fit are summarized in Supplementary file 1 – Table 1. Figure 2—figure supplement 1—source data 1.Excel file with numerical electrophysiology data used for Figure 2—figure supplement 1. Figure 2—figure supplement 1—source data 2.Excel file with numerical…

• 

  Figure 2—figure supplement 2.. Electrophysiology measurements of KCNQ1 WT, V141C, and I274C with KCNE1 WT under reducing and oxidizing conditions. (A) Whole-cell currents (left) and average I-V relationships (right) of CHO-K1 cells expressing KCNQ1 and KCNE1 WT, which were exposed to control bath solution, DTT, or Cu-phenanthroline (Cu-phen.), respectively. (mean ± SEM, Control n = 10, DTT n = 4, Cu-phen. n = 6) (B) Whole-cell currents and average I-V relationships of CHO-K1 cells expressing KCNQ1 V141C and KCNE1 WT, which were exposed to control bath solution containing DTT or Cu-phenanthroline. (Control n = 18, DTT n = 12, Cu-phen. n = 9) (C) Whole-cell currents and average I-V relationships of CHO-K1 cells expressing KCNQ1 I274C and KCNE1 WT measured with or without DTT or Cu-phenanthroline in the bath solution. (Control n = 20, DTT n = 9, Cu-phen. n = 11) Solid lines represent fits with a Boltzmann function (Itail/Itailmax = (1 - IBottom) / (1+exp[(V1/2app - V)/k]) + IBottom) and the parameters of the fit are summarized in Supplementary file 1 – Table 1.

• 

  Figure 2—figure supplement 3.. Electrophysiology measurements of KCNQ1 V141C or I274C with KCNE1 V47C or L48C under reducing and oxidizing conditions. (A) Average I-V relationships and activation curves of currents recorded from CHO-K1 cells transiently expressing KCNQ1 I274C and KCNE1 L48C. Cells were exposed to control bath solution, DTT, or Cu-phenanthroline (Cu-phen.), respectively. (mean ± SEM, Control n = 12, DTT n = 7, Cu-phen. n = 7) (B) Average I-V relationships and activation curves of currents measured from CHO-K1 cells transiently expressing KCNQ1 V141C and KCNE1 V47C, which were exposed to control bath solution, DTT, or Cu-phenanthroline (Cu-phen.), respectively. (mean ± SEM, Control n = 13, DTT n = 10, Cu-phen. n = 7) (C) Average I-V relationships and activation curves of CHO-K1 cells expressing KCNQ1 I274C and KCNE1 V47C with and without DTT or Cu-phenanthroline in the bath solution. (Control n = 8, DTT n = 6, Cu-phen n = 6) Solid lines represent fits with a Boltzmann function (Itail/Itailmax = (1 - IBottom) / (1+exp[(V1/2app - V)/k]) + IBottom) and the parameters of the fit are summarized in Supplementary file 1 – Table 1.

• 

  Figure 3.. Molecular models of the KCNQ1-KCNE1 channel in RC and AO conformations. (A) Schematic cartoon depicting the functional states of the VSD (green box) and PD (blue box) in the KCNQ1-KCNE1 models. The S4 helix (with positive gating charges “+”) and S4-S5L, which connects S4 to the PD, are shown as green and blue cylinders, respectively. KCNE1 was docked to KCNQ1 with the VSD/PD in the resting/closed (RC) or activated/open (AO) conformation. (B) Side view of the KCNQ1-KCNE1 docking models. KCNQ1 is represented with cylindrical helices and KCNE1 is depicted as yellow ribbon. Residues F57, T58, and L59 are drawn as spheres and colored light blue, red, and dark blue, respectively. The approximate position of the membrane bilayer is indicated by horizontal lines and the extracellular and intracellular side are labeled EXT and INT, respectively. (C) View of the KCNQ1-KCNE1 models from the extracellular side. KCNE1 is bound in a cleft between the VSD and PD and makes contacts to three KCNQ1 subunits. The position of the other three equivalent KCNE1-binding sites in the tetrameric KCNQ1 channel is indicated.

• 

  Figure 3—figure supplement 1.. Structural comparison of Rosetta-generated computational models of human KCNQ1, which were used for docking, with cryo-EM-determined models of human KCNQ1 (Sun and MacKinnon, 2020). (A) Comparison of the PD from the Rosetta model of the resting/closed (RC) state with the PD from the cryo-EM model of the activated/closed (AC) conformation (PDB: 6V00). (B) Comparison of the PD from the Rosetta model of the activated/open (AO) state with the PD from the cryo-EM model of the AO conformation (PDB: 6V01). (C) Comparison of the VSD in an activated conformation from the Rosetta AO model with the VSD from the cryo-EM AC model (PDB: 6V00). (D) Comparison of one KCNQ1 subunit from the Rosetta RC model with the corresponding subunit from the cryo-EM AC model (PDB: 6V00). The RMSD was calculated excluding the S4 and S4-S5L helix because those two helices undergo conformational changes during transition from the resting (RC) to the activated (AC) state. (E) Comparison of one KCNQ1 subunit from the Rosetta AO model with the corresponding subunit from the cryo-EM AO model (PDB: 6V01). (F) Summary of the Cα-RMSD values (in Å) calculated for the structural comparisons in (A) to (E). (G) Comparison…

• 

  Figure 3—figure supplement 2.. Flowchart of the Rosetta protein-protein docking protocol for building KCNQ1-KCNE1 models. Modeling was conducted in three stages: preparation of KCNE1 input structures and KCNQ1 homology models, iterative protein-protein docking and distance restraint score-based filtering for model generation, and visual model inspection and MD simulation for model analysis. Individual modeling steps are written within rounded rectangular boxes whereas the names of used algorithms and programs are within skewed rectangular boxes. Compare also with Materials and method details.

• 

  Figure 3—figure supplement 3.. Rosetta binding energy (ΔGBinding) versus interface root-mean-square deviation (RMSD) plots of KCNQ1-KCNE1 docking models. For every model of KCNE1 docked to KCNQ1 in (A) the resting/closed or (B) activated/open conformation, the Rosetta binding energy (ΔGBinding) and protein-protein interface RMSD relative to the best scoring model is plotted. Colors represent the docking round (left) or a model’s atom pair restraint score (right), respectively. Figure 3—figure supplement 3—source data 1.Excel file with numerical data used for the energy-vs-RMSD plots in panel A of Figure 3—figure supplement 3. Figure 3—figure supplement 3—source data 2.Excel file with numerical data used for the energy-vs-RMSD plots in panel B of Figure 3—figure supplement 3.

• 

  Figure 3—figure supplement 4.. Control docking calculations for KCNQ1-KCNE1 and KCNQ1-KCNE3 complexes using the Rosetta and cryo-EM models of the activated/open state structure. (A) Docking of the NMR-determined model of KCNE1 (S37–L71) (PDB: 2K21) (Kang et al., 2008) to the cryo-EM-determined open state model of KCNQ1 (PDB: 6V01) (Sun and MacKinnon, 2020). Left: Rosetta-binding energy (ΔGBinding) for KCNE1 versus the docking models’ interface RMSD (I-RMSD) and fraction of recovered contacts compared to the model developed by docking KCNE1 to the Rosetta homology model of the KCNQ1 open state structure. Right: Comparison of the best-scoring model obtained by docking KCNE1 to the cryo-EM model of KCNQ1 versus the Rosetta KCNQ1-KCNE1 model developed in this work. The I-RMSD between the two models is indicated. (B) Docking of the NMR-determined model of KCNE3 (P51–V85) (PDB: 2NDJ) (Kroncke et al., 2016) to the cryo-EM-determined open state model of human KCNQ1 (PDB: 6V01) (Sun and MacKinnon, 2020). Left: Rosetta binding energy for KCNE3 versus the docking models’ I-RMSD and fraction of recovered contacts compared to the cryo-EM model of the KCNQ1-KCNE3 complex. Right: Comparison of the best-scoring KCNQ1-KCNE3…

• 

  Figure 4.. Mapping of experimental distance restraint sites onto the KCNQ1-KCNE1 models. (A) Interaction of KCNE1 with the KCNQ1 RC and AO model in the transmembrane region. Residues whose distance was restrained in docking are indicated as spheres. (B) Restrained residue pairs and their Cα-Cα distances (in Å) in the KCNQ1-KCNE1 RC and AO models.

• 

  Figure 5.. Computational detection and experimental validation of the KCNE1-binding site of KCNQ1. (A) Left: Surface representation of the KCNE1 binding cleft in the KCNQ1 RC model. Residues are colored by their average MD contact number with KCNE1 (indicated in parentheses). Right: Matrix of KCNQ1-KCNE1 contacts (bottom) and histogram of the number of intermolecular contacts for KCNE1 (top) (mean ± SD). (B) Left: Surface representation of the KCNE1 binding site in the KCNQ1 AO model with residues colored by their average MD contact number. Right: Matrix and histogram of the number of intermolecular contacts for KCNE1 (mean ± SD). (C) Interaction of KCNQ1 with the upper, middle, and lower part of the KCNE1 TMD. Three selected sites in KCNQ1 and their neighboring residues in KCNQ1 and KCNE1 are displayed: left – W323, middle – Y267, right – K362+N365. Residue sidechains are drawn as sticks and potential H-bond contacts are indicated by dashed lines. Histograms of the average MD contact number with KCNE1 for the selected residues in the KCNQ1 RC and AO model are shown next to the structural models. (D) Whole-cell currents of CHO-K1 cells stably expressing KCNE1 and transfected with KCNQ1 WT…