CryoEM structure of PIP2-bound, open SUR1/Kir6.2Q52R KATP channel.

(a) The 2.9 Å cryoEM map (the micelle not shown), from C4 non-uniform refinement of 14,115 particles, shown from the side and bottom (cytoplasmic view). (b) Structural model of the SUR1/Kir6.2Q52R KATP channel in complex with PIP2. The potassium ion selectivity filter is shown as sticks with oxygens colored red, the K+ conduction pathway is shown as purple mesh. Only two Kir6.2 and two SUR1 subunits are shown in the side view for clarity. A view rotated 90° (cytoplasmic view) displays all four Kir6.2 subunits and all four SUR1 subunits. (c) Structural model of a closed WT SUR1/Kir6.2 KATP channel in complex with ATP and gliblenclamide (Glib) (PDB ID 6BAA), shown for comparison and rendered in PyMOL similar to (b), except for the Kir6.2 subunits which are colored grey.

Two PIP2 binding sites at the interface of Kir6.2 and SUR1.

(a) CryoEM map features of two PIP2 molecules colored in cyan and magenta (0.08V, 4.0σ contour), respectively. (b) Structural model of the PIP2 binding pocket (red boxed region in (a)) viewed from the side, with cryoEM density of PIP2 shown as grey mesh. Residues from both Kir6.2 (adjacent subunit denoted “B”) and SUR1 (grey outline) surrounding bound PIP2 molecules are labeled. (c) Structural model of the PIP2 binding pocket view from the top (extracellular side).

Functional testing of the two PIP2 binding sites.

(a) Representative inside-out patch-clamp recordings (−50mV, inward currents shown as upward deflections; the red dashed line indicates 0 currents) of various channels: WT, SUR1K134A (the second PIP2 binding site mutation)/Kir6.2, SUR1/Kir6.2R176A (the first PIP2 binding site mutation), and SUR1K134A/Kir6.2R176A double mutant, exposed alternately in 5 μM PIP2, 0.1 mM ATP, and 1 mM ATP as indicated by the bars above the recordings. The brief exposures to 0.1 mM and 1 mM ATP between PIP2 exposures were used to monitor the gradual decrease in ATP sensitivity as channel opening became increasingly stabilized by PIP2. (b) Group data of recordings shown in (a) comparing initial currents in K-INT at the time of patch excision. For each group, individual data points of 12 (WT), 6 (SUR1K134A), 7 (Kir6.2R176A), and 6 (SUR1K134A/Kir6.2R176A) patches and the mean ± SEM of the data points are shown. (c) Comparison of the maximum fold-increase in currents after PIP2 exposure in different channels from the same recordings analyzed in (b). (d) Comparison of the time of exposure in PIP2 for currents to reach maximum in different channels using the same recordings analyzed in (b) and (c). Note for (b) and (c), the y-axis is in log scale for better visualization. Statistical significance is marked by *p<0.05, **p<0.005, ***p<0.001 between WT and the mutants, #p<0.05, ##p<0.005 between SUR1K134A and the double mutant, and $p<0.05 between Kir6.2R176A and the double mutant, using one-way ANOVA and Tukey’s post-hoc test.

Comparison of the inner helix gate in the open and closed KATP channel structures.

(a) Structure of Kir6.2Q52R in a PIP2-bound open state. Bound PIP2 are shown as cyan and magenta sticks and spheres, and the side chains of inner helix gating residues L164 and F168 at the helix bundle crossing and G295 at the G-loop are shown as yellow sticks and labeled in red. The pore for the ion pathway (purple mesh) is constricted at the selectivity filter (SF) but open at the inner helix gate (3.3 Å). The transmembrane α-helices (M1 and M2), the interfacial α-helices (IFH) and the pore α-helices (PH) are labeled. (b) Kir6.2 in an ATP-bound closed state (PDB ID 6BAA) with similar labels as panel A for comparison. (c) Pore radii for ion conduction pathway plotted against the distance from the extracellular opening, with the location of the selectivity filter (residues 130-133), the inner helix gate (residues L164 and F168), and the G-loop gate (residues 294-297) shown. The pore radii were calculated using the program HOLE implemented in coot and viewed using a MOLE rendering in PyMOL.

SUR1 and Kir6.2 cytoplasm-plasma membrane interface in open and closed KATP channel conformations.

(a) SUR1-Kir6.2 cytoplasm-plasma membrane interface in open PIP2 (cyan and magenta sticks)-bound SUR1/Kir6.2Q52R (SUR1 in teal and Kir6.2 in yellow) structure, and (b) closed ATP-(orange carbon sticks) and repaglinide (out of view)-bound WT (SUR1 in pink and Kir6.2 in grey) KATP channel (PDB ID 7TYS) viewed from the extracellular side. Note reorientation of side chains of the Kir6.2 inner helix gate residue F168 and M1 residue F60, as well as W51 at the bottom of TM1 of SUR1-TMD0 in the two conformations. In (a), a 6.4° clockwise rotation of the Kir6.2-CTD comparing the open to the closed conformation is indicated by the red curved arrow (with D323 Cα in each structure marked as spheres), and SUR1-L0, marked by the ATP-binding residue K205 and an adjacent residue E203, movement away from the ATP binding pocket in the open conformation relative to the closed conformation is marked by a red arrow. (c) A side view of the interface in the open conformation. SUR1-intracellular loop 1 (ICL1) forms a β-sheet with Kir6.2-βA just before the interfacial helix (IFH). Remodeling of residues involved in ATP binding and gating compared to the closed conformation in (d), including Kir6.2-R50, R54, E51 and K39(B) (“B” denotes the adjacent Kir6.2 subunit) as well as SUR1-E203 and K205 in L0 are shown. Most notably, the Kir6.2 mutant residue Q52R has its side chain reoriented to interact with the side chain of SUR1-W51 which also has its side chain reoriented from the closed conformation. Color scheme same as (a). (d) The same side view as in (b) but of the closed conformation. In all panels, SUR1 residue labels are outlined in grey. The red dashed line in (c) and (d) is the distance between the Cα of Kir6.2 E51 and SUR1 K205. In all panels, SUR1 helices are indicated with blue numbers.

Structural changes between open- and closed-forms of the KATP channel at the SUR1 and Kir6.2 TM interface.

(a) An overlay of the SUR1-Kir6.2 TM interface between open SUR1/Kir6.2Q52R KATP channel and the closed KATP channel (PDB ID 6BAA) showing the outward movement of the cytoplasmic half of the SUR1-TMD0. In the open structure, SUR1 is shown in teal and Kir6.2 in yellow. The closed structure is shown in grey. (b) Close-up view of the open structure in the boxed region in (a). CryoEM map (blue mesh, 0.17V/8.5σ map contour) is superposed with the structural model. For clarity, PIP2 is omitted. (c) An overlay of the open with the closed KATP channel structures in the boxed region in (A) (same color scheme) showing significant changes in residues of SUR1-TM1 at the inner leaflet half of the plasma membrane.

Kir6.2-Q52R interacts with SUR1-W51 to reduce channel sensitivity to ATP inhibition and enhance channel activity.

(a) A close-up view of the interaction between Kir6.2-Q52R and SUR1-W51 with cryoEM map (blue mesh, 0.17V/8.5σ contour) superposed with the structural model in two different angles. (b) Representative inside-out patch-clamp recordings (−50mV, inward currents shown as upward deflections; red dashed line indicates baseline 0 currents) of the SUR1/Kir6.2Q52R channel and SUR1W51C/Kir6.2Q52R channel exposed to differing concentrations of ATP as indicated by the bars above the recordings. (c) ATP dose response of WT (SUR1/Kir6.2) channels, channels containing the SUR1W51Cmutation (SUR1W51C), Kir6.2Q52R mutation (Kir6.2Q52R), or both SUR1W51C and Kir6.2Q52R mutations (SUR1W51C/Kir6.2Q52R). Curves were obtained by fitting the data points (mean ± SEM of 3-8 patches) to the Hill equation (see Methods). *p < 0.01, **p < 0.0001 compared to WT; #p < 0.0001 compared to Kir6.2Q52R, one-way ANOVA with Tukey’s post hoc test (d) Rb+ efflux of various channels (same labeling as in (c)) expressed in COSm6 cells. Untransfected cells were included to show background efflux. Only SUR1+Kir6.2Q52R channels showed significant activity above WT (p < 0.0001, n=5; one-way ANOVA with Tukey’s post hoc test).

KATP channel disease mutations near the PIP2 binding pocket.

Residues with variants that have been reported to cause neonatal diabetes/DEND syndrome are shown as green spheres and congenital hyperinsulinism shown as orange spheres (with the exception of L135, which is colored half green and half orange as mutation at this position has been linked to both diseases depending on the substituting amino acid). Neonatal diabetes/DEND syndrome mutations include SUR1-P45L, I49F, F132L/V, L135P, and Kir6.2-K39R, E51A/G, Q52L/R, G53D/N/R/S/R/V, V59A/G/M, W68C/G/L/R, K170N/R/T, E179A/K. Congenital hyperinsulinism mutations include SUR1-I46T, P133R and L135V, and Kir6.2-R54C, L56G, K67D, R176H, and R177W. SUR1 and Kir6.2 are colored grey and yellow, respectively.

CryoEM data processing.

(a) Workflow showing the total number of particles yielding the 2.9 Å C4 Non-uniform refinement reconstruction is 14,115 (FSC auto mask for GSFSC calculation). The three raw images on the top left corner serve to illustrate the particles on GO layer (GO edges are seen in the left micrograph) selected (red circles) for processing (b) Fourier Shell Correlation (FSC) curves between two independent half-maps calculated within masks shown in (c), and for model:map (unmasked) for the full KATP channel model and map (PDB ID 8TI2; EMD-41278) (left), and for the KATP channel without NBD2 (PDB ID 8TI1) (right). (c) Masks used for FSC calculations: full particle including micelle (green), a mask for full KATP channel without the micelle (pink), and a mask created by cryoSPARC auto-mask tightening that excludes NBD2 (blue), shown in cytoplasmic view. (d) The local resolution estimate for the reconstructed map at 4 σ (0.08 V) contour, with micelle density not shown, in side view (left), top view (middle) and bottom view (right). No local filtering or local sharpening was used for visualization. Local resolution estimates were calculated in cryoSPARC and visualized in ChimeraX. FSC curves were calculated in PHENIX and cryoSPARC.

Distinct conformation classes revealed by symmetry expansion and focused 3D classification.

(a) Workflow to obtain distinct conformation classes (details described in Methods). (b) Overlay of full channel map generated by combining the two yellow classes from the second round of 3D classification (shown in yellow) and the blue map from the second round of 3D classification(shown in blue) viewed from the side and the bottom. The yellow map resembles the PIP2-bound open structure derived from C4 non-uniform refinement (Figs. 1a, S1a), while the blue map resembles previously published closed apo WT structure (PDB ID 7UQR), and will be referred to as apo SUR1/Kir6.2Q52R structure. (c) Cytoplasmic view of the Kir6.2Q52R tetramer plus one SUR1 in the two maps superimposed (top) or separately. In the blue map, Kir6.2Q52R-CTD is rotated clockwise relative to the yellow map. Also in the blue map, the NBDs of SUR1 are closer to the Kir6.2Q52R tetramer and SUR1 is rotated opposite of Kir6.2Q52R-CTD, compare to the yellow map. (d) Side view of the maps in (c) showing Kir6.2Q52R-CTD is docked up to the membrane in the yellow map, and extended down from the plasma membrane in the blue map. Also, SUR1 is tilted away from the Kir6.2Q52R tetramer in the yellow map, but tilted towards the Kir6.2Q52R tetramer in the blue map. The curved and straight open arrows mark the relative rotation and translation from the membrane of the Kir6.2Q52R-CTD. The solid arrows indicate relative movements of the SUR1. (e) Side view of the SUR1 ABC core showing the presence of cryoEM density corresponding to the Kir6.2Q52R N-terminal peptide (KNtp, red density) in the apo closed SUR1/Kir6.2Q52R map (blue) but absence of KNtp density in open SUR1/Kir6.2Q52R map (yellow).

CryoEM lipid density fitting.

(a) CryoEM map reconstructed from symmetry expansion of the ∼14k particles followed by local refinement with a focused mask of Kir6.2 tetramer plus one SUR1, with density modification in Phenix and map segmentation in Chimera, shows the cryoEM densities of SUR1 and the Kir6.2 tetramer colored blue and yellow, respectively. Two PIP2 molecules are shown in magenta and cyan, and other density including lipid density shown in grey (0.08V/4σ contour). (b) Isolated lipid density located in the inner leaflet between SUR1-TMD0 and SUR1-TMD1 fit with a model of phosphatidylserine (PS, grey carbons, 0.08V/4σ contour) and isolated lipid density at each of the two PIP2 binding sites located between SUR1-TMD0 and Kir6.2 shown fit with model of the first-(cyan carbons) and second-(magenta carbons) (C18:0/C20:4) PI(4,5)P2 in two different rotational views. (c) Rotation of the map 180° relative to panel A reveals other densities (grey, 0.08V/4.0σ contour) corresponding to two adjacent lipids primarily associated with the TMD0 domain. (d) Adjacent lipid densities can be well fit as a phosphatidylserine (PS) and phosphatidylethanolamine (PE), with the amine group of PE hydrogen bonding to the phosphate group of PS. On the right, density for each of these adjacent lipid molecules is isolated to show the fit of the lipid model into the lipid’s corresponding density. These five lipid densities (20 lipids for full-channel) that are captured in the micelle are closely associated withthe KATP channel and are sufficiently resolved to allow modeling. Note lipid or detergent densities are also observed in the outer leaflet space; however, they are not sufficiently resolved to allow modeling.

Comparison of PIP2-bound open KATP channel structure with other open KATP structures.

(a, b) Comparison of the lipid density (grey mesh) in the PIP 2 binding pocket in the PIP2-bound SUR1/Kir6.2Q52R structure, SUR1-Kir6.2H175K fusion channel pre-open structure (in the presence of diC8-PIP2; EMD-32310, PDB ID 7W4O), and SUR1/Kir6.2C166S,G334D open structure (no PIP2 added; EMD-24842, PDB ID 7S5X), viewed from the top (a) and the side (b). In (b), the cryoEM density for Kir6.2 R54 is shown as a reference. (c, d) Top view of PIP2-bound SUR1/Kir6.2Q52R open channel (c) and SUR1-Kir6.2H175K fusion pre-open channel (PDB ID 7W4O) (d), showing similarities in side chain orientations for Kir6.2 F60 and F168 (the gate residue at the helix bundle crossing) and SUR1-W51, and rotation position of the Kir6.2 cytoplasmic domain (CTD) marked by residue D323. (e, f) Comparison of the two structures in (c) and (d) viewed on the side, showing close proximity of SUR1-W51 to Kir6.2-Q52R (e) or Q52 (f), as well as the widening of the ATP binding pocket marked by the distance (dashed red line) between the Cα atoms of Kir6.2-E51 and SUR1-K205.

Comparison of KNtp density in the SUR1-ABC core in the PIP2-bound open versus apo closed SUR1/Kir6.2Q52R structures.

(a) In the PIP2-bound SUR1/Kir6.2Q52R open channel, no KNtp density is observed, consistent with the absence of the Kir6.2Q52R-Ntp or a highly flexible Kir6.2Q52R-Ntp. Note the density protruding from SUR1-TMD2 TM helix corresponds to W1297 with some R1145 contribution. (b) In the apo SUR1/Kir6.2Q52R closed channel, a clear peptide density corresponding to Kir6.2Q52R-Ntp is clearly present. For density shown in the KNtp binding cleft in SUR1, both maps were filtered to 7 Å and contoured to 6.5 σ.

Conformational dynamicsof the SUR1 subunit revealed by CryoSPARC 3D classification.

(a,b) Dynamic range of SUR1 within the final open particle class (Fig. S1) was determined by symmetry expansion followed by 3D classification. The 14,115 particles within the final open class gave 56,460 particles, which were subjected to 3D classification with a focused mask of Kir6.2 tetramer plus one SUR1. Three unique NBD-separated SUR1 positions were resolved into three classes. Reconstructed maps of class 1 (3.0 Å resolution, 23,454 particles, red map), class 2 (3.3 Å resolution, 16,863 particles, yellow map) and class 3 (3.4 Å resolution, 13,604 particles, blue map) reconstructed from a local non-uniform refinement with density modification in Phenix. (a) Transmembrane view shows the variability in SUR1 including the transmembrane regions of TMD1 and TMD2, in an open Kir6.2 tetramer position. (b) View from the cytoplasmic side highlights the difference in the SUR1-NBD positions within the open particle class, with mobility seen in both SUR1-NBD1 and SUR1-NBD2. The grey arrow indicates the direction of the NBD position differences. (c,d) The reconstructed maps for the PIP2-bound open class (yellow, 7.0 Å filtered resolution) and apo closed class (blue, 7.0 Å filtered resolution) within the SUR1/Kir6.2Q52R dataset (workflow in Fig. S2), show the relative SUR1 and Kir6.2 domain positions as the channel transitions between open and closed conformations. (c) Transmembrane view shows the difference in the SUR1-NBD position and the rotation of the Kir6.2-CTD. (d) View from the cytoplasmic side highlights the difference in the SUR1-NBD positions relative to the Kir6.2 subunit, with the open conformation having the SUR1-NBD further away from the Kir6.2 cytoplasmic domain which is rotated counterclockwise relative to the closed position. The grey curved arrows indicate the rotational position difference in Kir6.2Q52R-CTD and SUR1-NBDs

Perturbation of PIP2 binding residues diminished the gain-of-function effect of Kir6.2Q52R.

(a) ATP dose response of WT (SUR1/Kir6.2) channels, and channels containing the PIP2 binding site mutations SUR1K134A (SUR1K134A) or Kir6.2R176A, as well as the gain-of-function disease mutation Kir6.2Q52Ralone or in combination with single PIP2 binding site mutations (SUR1K134A/Kir6.2Q52R, or Kir6.2Q52R/R176A) or double PIP2 binding site mutations (SUR1K134A/Kir6.2Q52R/R176A). Curves were obtained by fitting the data points (mean ± SEM of 3-6 patches) to the Hill equation (see Methods). #p < 0.001. compared to WT channel; *p < 0.001, compared to SUR1/Kir6.2Q52R channel (Kir6.2Q52R), by one-way ANOVA with Tukey’s post hoc test. (b) Rb+ efflux of various channels expressed in COSm6 cells measured in Ringer’s supplemented with 5 mM glucose. Untransfected cells were included to show background efflux. *p < 0.05, **p < 0.0001 compared to WT; #p < 0.05, ##p < 0.0001 compared to Kir6.2Q52R, n=3-6; one-way ANOVA with Tukey post hoc test).

The KATP channel closed conformations clash with the novel second PIP2 molecule, but both the open and closed conformations could accommodate the conserved first PIP 2.

(a) The open SUR1/Kir6.2Q52R KATP channel with the conserved first PIP2 (cyan carbons) and novel second PIP2 (magenta carbons) shown interacting with Kir6.2 subunits (pink, green and blue surface) and the SUR1 subunit (orange surface), with the two panels rotated ∼15° to show the surface of each PIP2 site. (b) The closed KATP channel bound to repaglinide and ATP (PDB ID 7TYS, with Kir6.2-CTD in the up position) aligned with the PIP2 molecules of the open channel shows that the closed KATP channel accommodates the conserved first PIP2 but clashes with the novel second PIP2. (c) The apo closed SUR1/Kir6.2Q52R KATP channel (with Kir6.2-CTD in the down position, similar to previously reported apo WT structure PDB ID 7UQR; see Fig. S2) also shows clash with PIP2 at the second site but not the first site.

Model statistics of the PIP2-bound SUR1/Kir6.2Q52R structure.