Structural basis of fast N-type inactivation in Kv channels

Abstract

Action potentials are generated by opening of voltage-activated sodium (Na_v) and potassium (K_v) channels¹, which can rapidly inactivate to shape the nerve impulse and contribute to synaptic facilitation and short-term memory^1-4. The mechanism of fast inactivation was proposed to involve an intracellular domain that blocks the internal pore in both Na_v^5,6 and K_v^7-9 channels; however, recent studies in Na_v^10,11 and K_v^12,13 channels support a mechanism in which the internal pore closes during inactivation. Here we investigate the mechanism of fast inactivation in the Shaker K_v channel using cryo-electron microscopy, mass spectrometry and electrophysiology. We resolved structures of a fully inactivated state in which the non-polar end of the N terminus plugs the internal pore in an extended conformation. The N-terminal methionine is deleted, leaving an alanine that is acetylated and interacts with a pore-lining isoleucine residue where RNA editing regulates fast inactivation¹⁴. Opening of the internal activation gate is required for fast inactivation because it enables the plug domain to block the pore and repositions gate residues to interact with and stabilize that domain. We also show that external K⁺ destabilizes the inactivated state by altering the conformation of the ion selectivity filter rather than by electrostatic repulsion. These findings establish the mechanism of fast inactivation in K_v channels, revealing how it is regulated by RNA editing and N-terminal acetylation, and providing a framework for understanding related mechanisms in other voltage-activated channels.

Fig. 1. Structures of modified Shaker K_v channels.

a, Current traces recorded from an RNA-injected oocyte using 2 mM external K⁺. The holding voltage was −100 mV, with 10-mV steps up to +80 mV. The red dashed line is zero current. b, Side and external views of the cryo-EM C4 map and model for proteolysed samples of C-terminally tagged WT Shaker. PD, poredomain; TM, transmembrane; VSD, voltage-sensing domain. c,d, Side views of the class A (c) and class B (d) models for proteolysed samples of C-terminally tagged WT Shaker with C1 symmetry. The light blue density in the chamber between the transmembrane and T1 domains and within the pore is not seen with C4 symmetry. e, Current traces for Shaker expressed in oocytes by injecting proteoliposomes. The same protocol as panel a with a tail voltage of −50 mV. P/−4 subtraction was used. f, Time constants of fast inactivation (τ) versus test voltage for WT Shaker using RNA injection compared with GT Shaker using proteoliposome injection. τ were obtained from single exponential fits to current traces (n = 7 cells in 4 independent experiments for WT Shaker and n = 8 cells in 5 independent experiments for GT Shaker). g, Current traces measuring recovery from inactivation for GT Shaker using proteoliposome injection. The holding voltage was −100 mV, with initial 60-ms steps to +30 mV followed by varying time for recovery at −100 mV (initially 2 ms and increased in 5-ms intervals) before eliciting a second step to +30 mV. External K⁺ was 100 mM. P/−4 subtraction was used, and the red dashed line denotes zero current. h, Fraction of current recovered (fR) versus time for GT Shaker using proteoliposome injection (n = 3 cells in 2 independent experiments for 2 mM external K⁺ and n = 3 cells in 2 independent experiments for 100 mM external K⁺) compared with WT Shaker using RNA injection (n = 7 cells in 4 independent experiments for 2 mM external K⁺ and n = 5 cells in 4 independent experiments for 100 mM external K⁺) using the protocol in panel g. See Extended Data Fig. 4a,b for traces for WT Shaker. i–k, Side and internal views of the cryo-EM C1 maps and models for classes A (i), B (j) and C (k) of GT Shaker. Extra N-terminal density in the internal pore is shown in light blue. Error bars are s.e.m. (f,h).

Fig. 2. Modifications of the N terminus in the full-length Shaker K_v channel.

a, Sequence of the N terminus of Shaker, with the dominant species detected shown in red. b, MS/MS spectrum of the (Ac)AAVAGLYGLGEDR peptide. The left y axis shows relative intensity and the right y axis shows absolute intensity. Fragments matched to the theoretical masses are marked in red. c, Extracted ion chromatogram of (Ac)AAVAGLYGLGEDR (Ac-Ala; blue) and AAVAGLYGLGEDR (Ala; red). d, Current traces for AcA Shaker expressed in oocytes using proteoliposome injection. External K⁺ was 2 mM. The holding voltage was −100 mV, with 10-mV steps up to +80 mV. P/−4 subtraction was used, and the red dashed line is zero current. e, Time constants of fast inactivation (τ) versus test voltage for WT Shaker using RNA injection (n = 7 cells in 4 independent experiments for 2 mM external K⁺ and n = 5 cells in 4 independent experiments for 100 mM external K⁺) compared with AcA Shaker using proteoliposome injection (n = 7 cells in 3 independent experiments for 2 mM external K⁺ and n = 5 cells in 2 independent experiments for 100 mM external K⁺). τ was obtained from single exponential fits to current traces. f, Family of current traces to measure recovery from inactivation for AcA Shaker expressed using proteoliposome injection. The holding voltage was −100 mV, with initial 60-ms steps to +30 mV, followed by varying time for recovery at −100 mV (initially 2 ms and increased in 5-ms intervals) before eliciting a second step to +30 mV. External K⁺ was 100 mM. P/−4 subtraction was used, and the red dashed line is zero current. g, Fraction of current recovered (fR) versus recovery time for AcA Shaker using proteoliposome injection (n = 6 cells in 2 independent experiments for 2 mM external K⁺ and n = 5 cells in 2 independent experiments for 100 mM external K⁺) compared with WT Shaker using RNA injection (data from Fig. 1h) using the protocol shown in panel f. Error bars are s.e.m.

Fig. 3. Structures of the N-type-inactivated state of the Shaker K_v channel.

a, Side view of the cryo-EM C1 map and model of the pore domain S6 helices from two opposing subunits for GT Shaker class A. b, Side views of the cryo-EM C1 map and model for GT Shaker class A, highlighting interactions between L7 in the N terminus and P475 and V478 in the S6 helix. Red asterisk highlights Cα of G6. c, Bottom views of the cryo-EM C1 map and model for GT Shaker class A, highlighting interactions between Y8 in the N terminus and N482 (arrow) and H486 (dashed line) in the S6 helix. d, Side view of the cryo-EM C1 map and model for AcA-EI Shaker class C. e, Side views of the cryo-EM C1 map and model for AcA-EI Shaker class C, highlighting interactions between L7 in the N terminus and P475 and V478 in the S6 helix. f, Bottom views of the cryo-EM C1 map and model for AcA-EI Shaker class C, highlighting interactions between Y8 in the N terminus and Ac-A2 from a second N-terminal peptide. g, Side view of the cryo-EM C1 map and model for the AcA-EI Shaker with added free N-terminal peptide. h, Side views of the cryo-EM C1 map and model for the AcA-EI Shaker with added free N-terminal peptide, highlighting interactions between L7 in the N terminus and P475 and V478 in the S6 helix. Red asterisk highlights Cα of G6. i, Bottom views of the cryo-EM C1 map and model for the AcA-EI Shaker with added free N-terminal peptide, highlighting interactions between Y8 in the N terminus and N482 (arrow) in the S6 helix.

Fig. 4. Slow C-type inactivation regulates recovery from N-type inactivation.

a, Current traces for Shaker(T449V) expressed in oocytes by injecting RNA recorded in 2 mM external K⁺ from a holding voltage of −100 mV and step depolarizations from −100 mV to +80 mV (10-mV increments). b, Time constants of fast inactivation (τ) plotted against test voltage in 2 mM external K⁺ for Shaker(T449V) compared with WT Shaker expressed by RNA injection. A single exponential fit to current traces like those in panel a was used to obtain values of τ. Data for WT Shaker are from Fig. 1f and n = 5 cells in 2 independent experiments for Shaker(T449V). c, Family of current traces to measure recovery from inactivation for WT Shaker and Shaker(T449V) expressed by RNA injection. The holding voltage was −100 mV and initial step depolarizations were to +80 mV for 500 ms followed by varying amounts of time for recovery at −100 mV (initially 2 ms and increased in 5-ms intervals) before eliciting a second step depolarization to +80 mV. External K⁺ was either 2 mM or 100 mM as indicated. All traces shown are P/−4 subtracted, and the red dotted line denotes zero current. d, Fraction of current recovered at −100 mV (f_R) as a function of recovery time for WT Shaker expressed by RNA injection. For WT Shaker, n = 3 cells in 2 independent experiments for 2 mM external K⁺ and n = 3 in 2 independent experiments with 100 mM external K⁺. e, Fraction of current recovered at −100 mV (f_R) as a function of recovery time for T449V expressed by RNA injection. For T449V, n = 5 cells in 2 independent experiments for 2 mM external K⁺ and n = 4 in 2 independent experiments with 100 mM external K⁺. Error bars are s.e.m.

Fig. 5. Exploring key interactions of the N-terminal plug domain with the Shaker K_v channel.

a, Currents recorded in 2 mM external K⁺ for N482A. The holding voltage was −100 mV, with 10-mV steps up to +80 mV. b, Time constants of fast inactivation (τ) versus test voltage for Shaker(N482A) (n = 3 cells in 2 independent experiments) compared with WT Shaker (from Fig. 1f). c, Current traces to measure recovery from inactivation for Shaker(N482A). The holding voltage was −100 mV, with initial 60-ms steps to +30 mV, followed by varying time for recovery at −100 mV (initially 2 ms and increased in 5-ms intervals) before eliciting a second step to +30 mV. External K⁺ was 100 mM. d, Fraction of current recovered at −100 mV (f_R) versus recovery time for N482A (n = 4 cells in 3 independent experiments for 2 mM external K⁺ and n = 3 in 2 independent experiments with 100 mM external K⁺). Data for WT Shaker are from Fig. 1h. e, Current traces for N482W using the same protocol as in panel a. f, Time constants of fast inactivation obtained as in panel b for N482W (n = 6 cells in 3 independent experiments). g, Family of currents to measure recovery from inactivation for Shaker(N482W). The holding voltage was −120 mV, but otherwise the protocol and conditions are as in panel c. h, Fraction of current recovered at −100 mV versus recovery time for Shaker(N482W) (n = 4 cells in 2 independent experiments for 2 mM external K⁺ and n = 3 in 2 independent experiments with 100 mM external K⁺). i, Current traces for Shaker(V478A) using the same protocol as in panel a. j, Time constants of fast inactivation versus test voltage for Shaker(V478A) (n = 3 cells in 2 independent experiments). k, Currents to measure recovery from inactivation for Shaker(V478A). The same protocol and conditions are as in panel c except initial steps were 150 ms. l, Fraction of current recovered at −100 mV versus recovery time for Shaker(V478A) (n = 3 cells in 2 independent experiments for 2 mM external K⁺ and n = 3 in 2 independent experiments with 100 mM external K⁺). P/−4 subtraction was used throughout, and the red dashed lines denote zero current. Error bars are s.e.m.

Extended Data Fig. 1. Cryo-EM imaging for the C-terminally mVenus tagged Shaker Kv channel and GT-Shaker constructs.

a) Local resolution maps for each structure. b) Fourier Shell Correlation (FSC) curves: FSC calculated without mask (blue); FSC calculated with a soft solvent mask (green); FSC calculated with the tight mask (red); FSC calculated using the tight mask with correction by noise substitution (purple). c) Direction distribution plots of the 3D reconstruction illustrating the distribution of particles in different orientations.

Extended Data Fig. 2. Mass spectrometry of the full-length Shaker Kv channel.

a) N-terminal sequences of WT-Shaker and EI-Shaker. b) MS/MS spectrum of an N-terminal peptide containing an additional Gly and Thr resulting from cleavage of the N-terminally mVenus-tagged Shaker Kv channel with TEV protease. c) MS/MS spectrum of an N-terminal peptide containing deletion of the N-terminal Met and acylation of Ala2 for the C-terminally mVenus-tagged Shaker Kv channel. d) MS/MS spectrum of an N-terminal peptide containing deletion of the N-terminal Met for the C-terminally mVenus-tagged Shaker Kv channel. e) MS/MS spectrum of an N-terminal peptide containing deletion of the N-terminal Met and acylation of Ala2 for the C-terminally mVenus-tagged EI-Shaker Kv channel. f) MS/MS spectrum of an N-terminal peptide containing deletion of the N-terminal Met for the C-terminally mVenus-tagged EI-Shaker Kv channel. g) Extracted ion chromatogram (XIC) of (Ac)AAVAGLYGLGK (Ac-Ala, labeled in blue) and AAVAGLYGLGK (Ala, labeled in red).

Extended Data Fig. 3. Data processing workflow for the cryo-EM structure of the GT-Shaker Kv channel.

a) Cyro-EM data processing pipeline for the GT-Shaker Kv channel. b) Representative micrograph from the 14,173 movies collected. About 2.4% of micrographs were discarded and at least 50% were of comparable quality to the representative one shown. c) 2D class averages of particles showing different orientations.

Extended Data Fig. 4. Recovery from inactivation for WT Shaker and inactivation of purified AcA-EI-Shaker.

a,b) Family of current traces to measure recovery from inactivation for WT-Shaker expressed by RNA injection with 2 mM (a) or 100 mM (b) external K⁺. Holding voltage was −100 mV and initial step depolarizations were to +30 mV for 60 ms followed by varying amounts of time for recovery at −100 mV (initially 2 ms and increased in 5 ms intervals) before eliciting a second step depolarization to +30 mV. Traces shown are P/−4 subtracted and red dotted line denotes zero current. Population data for recovery is shown in Fig. 1h. c) Current traces for AcA-EI-Shaker expressed in oocytes by injecting liposomes containing reconstituted purified proteins expressed in HEK cells. Oocytes were recorded in 2 mM external K⁺ from a holding voltage of −100 mV and step depolarizations from −100 mV to +80 mV (10 mV increments). Traces shown are P/−4 subtracted and red dotted line denotes zero current. d) Time constants of fast inactivation (τ) plotted against test voltage in WT-Shaker expressed by RNA injection compared to AcA-EI-Shaker expressed by proteoliposome injection. A single exponential fit to current traces like those in panel a was used to obtain values of τ. Data for WT-Shaker are from Fig. 2e and for AcA-EI-Shaker n = 3 cells in 2 independent experiments for 2 mM external K⁺ and n = 3 cells in 2 independent experiments for 100 mM external K⁺. Error bars are S.E.M.

Extended Data Fig. 5. Cryo-EM imaging for full-length Shaker Kv channel with enhanced inactivation mutations.

a) Local resolution maps for each structure. b) Fourier Shell Correlation (FSC) curves: FSC calculated without mask (blue); FSC calculated with a soft solvent mask (green); FSC calculated with the tight mask (red); FSC calculated using the tight mask with correction by noise substitution (purple). c) Direction distribution plots of the 3D reconstruction illustrating the distribution of particles in different orientations.

Extended Data Fig. 6. Cryo-EM structures for the full-length Shaker Kv channel with enhanced inactivation mutations.

a) Structural model for class A of the AcA-EI-Shaker Kv channel with cryo-EM density for the N-terminus shown in light blue from three rotational perspectives. b) Structural model for class B of the AcA-EI-Shaker Kv channel with cryo-EM density for the N-terminus shown in light blue from two rotational perspectives. c) Magnified N-terminal cryo-EM density for class B AcA-EI-Shaker, likely representing the average density for three conformations of the N-terminal peptide. d) Structural model for class C of the AcA-EI-Shaker Kv channel with cryo-EM density for the N-terminus shown in light blue from two rotational perspectives. In the second perspective, A391 in the S4-S5 linker is within 4.5–6 Å of the N-terminal cryo-EM density. e) Additional low-resolution cryo-EM density for a second N-terminal peptide in AcA-EI-Shaker class C (black box) connecting density within the chamber to the T1 domain. A391 in the S4-S5 linker is within 4.5–6 Å of the N-terminal cryo-EM density.

Extended Data Fig. 7. Cryo-EM structures for the full-length Shaker Kv channel with enhanced inactivation mutations in the presence of free N-terminal peptide.

a) Structural model of the pore domain S6 helices for AcA-EI-Shaker Kv channel in the presence of additional N-terminal peptide with cryo-EM density for the N-terminus shown in light blue. b) Model of conformation A where two N-terminal peptides (blue and green) were modeled into the cryo-EM density. c) Model of conformation B where two N-terminal peptides (blue and green) were modeled into the cryo-EM density.

Extended Data Fig. 8. Inactivation and recovery from inactivation for mutants of Shaker.

a) Fraction of current recovered at −100 mV (f_R) as a function of recovery time for WT-Shaker with a 60 ms inactivating depolarization to +30 mV compared to T449V with a 500 ms inactivating depolarization to +80 mV, both in 2 mM external K⁺. Data for WT-Shaker are from Fig. 1h and that for T449V are from Fig. 4e. b) Fraction of current recovered at −100 mV (f_R) as a function of recovery time for WT-Shaker with a 60 ms inactivating depolarization to +30 mV compared to T449V with a 500 ms inactivating depolarization to +80 mV, both in 100 mM external K⁺. Data for WT-Shaker are from Fig. 1h and that for T449V are from Fig. 4e. c) Fraction of current recovered at −100 mV (f_R) as a function of recovery time for WT-Shaker with a 60 ms inactivating depolarization to +30 mV compared to a 500 ms inactivating depolarization to +80 mV, both in 2 mM external K⁺. Data for the 60 ms inactivating pulse are from Fig. 1h and that for the 500 ms inactivating pulse are from Fig. 4d. d) Fraction of current recovered at −100 mV (f_R) as a function of recovery time for WT-Shaker with a 60 ms inactivating depolarization to +30 mV compared to a 500 ms inactivating depolarization to +80 mV, both in 100 mM external K⁺. Data for the 60 ms inactivating pulse are from Fig. 1h and that for the 500 ms inactivating pulse are from Fig. 4d. e) Current traces for Shaker N482L expressed in oocytes by injecting RNA recorded in 2 mM external K⁺ from a holding voltage of −100 mV and step depolarizations from −100 mV to +80 mV (10 mV increments). f) Time constants of fast inactivation (τ) plotted against test voltage in 2 mM external K⁺ for Shaker N482L compared to WT-Shaker expressed by RNA injection. A single exponential fit to current traces like those in panel a was used to obtain values of τ. Data for WT-Shaker are from Fig. 1f and for N482L n = 5 cells in 4 independent experiments. g) Family of current traces to measure recovery from inactivation for Shaker N482L expressed by RNA injection. Holding voltage was −100 mV and initial step depolarizations were to +30 mV for 60 ms followed by varying amounts of time for recovery at −100 mV (initially 2 ms and increased in 5 ms intervals) before eliciting a second step depolarization to +30 mV. External K⁺ was 100 mM. h) Fraction of current recovered at −100 mV (f_R) as a function of recovery time for Shaker N482L and WT-Shaker expressed by RNA injection. Data for WT-Shaker are from Fig. 1h and for N482L n = 4 cells in 4 independent experiments for 2 mM external K⁺ and n = 5 in 3 independent experiments for 100 mM external K⁺. i) Current traces for Shaker N480W. Same protocol and conditions as panel e. j) Time constants of fast inactivation (τ) plotted against test voltage in 2 mM external K⁺ for Shaker N480W compared to WT-Shaker expressed by RNA injection. Data for WT-Shaker are from Fig. 1f and for N480W n = 4 cells in 2 independent experiments. k) Family of current traces to measure recovery from inactivation for Shaker N480W expressed by RNA injection. Same protocol and conditions as panel g. l) Fraction of current recovered at −100 mV (f_R) as a function of recovery time for Shaker N480W and WT-Shaker expressed by RNA injection. Data for WT-Shaker are from Fig. 1h and for N480W n = 3 cells in 2 independent experiments for 2 mM external K⁺ and n = 3 in 2 independent experiments for 100 mM external K⁺. P/−4 subtraction was used for all traces shown and red dotted line denotes zero current. Error bars are S.E.M.

Extended Data Fig. 9. Functional properties of Shaker constructs.

a) Normalized tail current voltage-activation (G-V) relations for mVenus-GT-Shaker expressed in oocytes by injecting liposomes containing reconstituted purified proteins expressed in HEK cells compared to Shaker-IR expressed by injection of RNA. In both cases, external K⁺ was 2 mM, holding voltage was −100 mV and tail voltage was −50 mV. For Shaker-IR, n = 3 cells in 2 independent experiments and single Boltzmann fits yielded a V_1/2 of −29 ± 1.8 mV. For mVenus-GT-Shaker, n = 5 cells in 2 independent experiments and single Boltzmann fits yielded a V_1/2 of −25.6 ± 1.7 mV. b) Normalized tail current voltage-activation (G-V) relations for V478A Shaker-IR compared to Shaker-IR expressed by injection of RNA. In both cases, external K⁺ was 100 mM, holding voltage was −100 mV and tail voltage was −50 mV. For Shaker-IR, n = 3 cells in 2 independent experiments and single Boltzmann fits yielded a V_1/2 of −30.1 ± 3.8 mV. For V478A Shaker-IR, n = 3 cells in 2 independent experiments and single Boltzmann fits yielded a V_1/2 of −37.0 ± 2.5 mV. c) Normalized tail current voltage-activation (G-V) relations for N482W Shaker-IR compared to Shaker-IR expressed by injection of RNA. For N482W, external K⁺ was 100 mM, holding voltage was −120 mV and tail voltage was −120 mV. Data for Shaker-IR are from panel b and for N482W Shaker-IR, n = 3 cells in 2 independent experiments and single Boltzmann fits yielded a V_1/2 of −59.0 ± 3.3 mV. d) Current traces for N482W Shaker-IR expressed in oocytes by injecting RNA recorded in 100 mM external K⁺ using a holding voltage of −120 mV, step depolarizations from −100 mV to +30 mV (10 mV increments) and a tail voltage of −120 mV. e) Family of current traces to measure recovery from inactivation for Shaker N482W expressed by RNA injection. Holding voltage was −120 mV and initial step depolarizations were to +30 mV for 60 ms followed by varying amounts of time for recovery at −160 mV (initially 2 ms and increased in 5 ms intervals) before eliciting a second step depolarization to +30 mV. External K⁺ was 100 mM. f) Fraction of current recovered at −100 or −160 mV (f_R) as a function of recovery time for Shaker N482W and WT-Shaker expressed by RNA injection. Data for WT-Shaker are from Fig. 1h and for N482W n = 3 cells in 2 independent experiments for recovery at −100 mV and n = 3 in 2 independent experiments for recovery at −160 mV, in both cases using 100 mM external K⁺. P/−4 subtraction was used for all traces shown and red dotted line denotes zero current. Error bars are S.E.M.

Extended Data Fig. 10. Mechanism of fast inactivation.

Illustrations of the mechanism of fast N-type inactivation depicting key structural features and interactions with other states of the Shaker Kv channel. Under physiological conditions, the concentration of K⁺ ions are low outside and high inside the cell. The internal pore of the channel is closed when the peripheral voltage-sensing domains are in resting conformations at negative membrane voltages and opens with membrane depolarization and activation of the voltage-sensing domains^,,. The uncharged N-terminus enters the open internal pore to diminish ion permeation^–, with an acetylated Ala on the immediate N-terminus interacting directly with I470, a key residue that is modified by RNA editing to regulate N-type inactivation. V478 is a key residue forming the activation gate when the internal pore is closed^,,, while in the N-type inactivated state it interacts intimately with L7, one of the most critical residues within the N-terminus that stabilizes the N-type inactivated state. N-type inactivation promotes C-type inactivation of the ion selectivity filter by lowering the concentration of K⁺ on the internal side of the selectivity filter and holding the activation gate open. C-type inactivation also stabilizes the N-type inactivated state. Raising external K⁺ concentrations speeds recovery from inactivation by shifting the selectivity filter from the C-type inactivated state into a conducting state. Strong hyperpolarization when channels are N-type inactivated can trap the inactivation particle inside the pore as the internal pore closes off around an extended conformation of the N-terminus bound within the pore.