Molecular architecture of full-length KcsA: role of cytoplasmic domains in ion permeation and activation gating

Abstract

The molecular architecture of the NH(2) and COOH termini of the prokaryotic potassium channel KcsA has been determined using site-directed spin-labeling methods and paramagnetic resonance EPR spectroscopy. Cysteine mutants were generated (residues 5-24 and 121-160) and spin labeled, and the X-band CW EPR spectra were obtained from liposome-reconstituted channels at room temperature. Data on probe mobility (DeltaHo(-1)), accessibility parameters (PiO(2) and PiNiEdda), and inter-subunit spin-spin interaction (Omega) were used as structural constraints to build a three-dimensional folding model of these cytoplasmic domains from a set of simulated annealing and restrained molecular dynamics runs. 32 backbone structures were generated and averaged using fourfold symmetry, and a final mean structure was obtained from the eight lowest energy runs. Based on the present data, together with information from the KcsA crystal structure, a model for the three-dimensional fold of full-length KcsA was constructed. In this model, the NH(2) terminus of KcsA forms an alpha-helix anchored at the membrane-water interface, while the COOH terminus forms a right-handed four-helix bundle that extend some 40-50 A towards the cytoplasm. Functional analysis of COOH-terminal deletion constructs suggest that, while the COOH terminus does not play a substantial role in determining ion permeation properties, it exerts a modulatory role in the pH-dependent gating mechanism.

Figure 1

Functional and structural consequences of the site-directed mutagenesis of NH₂ and COOH termini. (A) Linear representation of KcsA. The two transmembrane segments are depicted by gray blocks, while the selectivity filter is shown in white. The arrows indicate the location of individual Cysteine mutants. (B and D) Effects of spin labeling on KcsA-catalyzed ionic fluxes. Extent ⁸⁶Rb⁺ influx assays of single cysteine mutants at pH 4.0 for NH₂ and COOH termini. Values are shown as a percentage of wild-type fluxes. (C and E) Effects of spin labeling on the oligomeric stability of KcsA. A per-subunit destabilization energy was obtained from the following equation: ΔΔG = ΔT_m (ΔH_o/T_m°)/4, where ΔT_m is the difference in the midpoint of the melting curve between wild-type (T_m°) and mutant (T_m*) KcsA, with ΔH_o = 263 kcal mol⁻¹, as described by Perozo et al. 1998.

Figure 2

Site-directed spin labeling and structural analysis of the NH₂-terminal end. (A) X-band CW-EPR spectra of spin-labeled NH₂-terminal mutants reconstituted into asolectin liposomes. In all cases, spectra were obtained with a microwave power of 2 mW and field modulation of 1 G. Spectra are arranged sequentially from the NH₂ to the COOH terminus of the segment, in the direction represented by the arrow. (B) Environmental parameter profiles: mobility parameter ΔHo⁻¹ (▪), oxygen accessibility parameter ΠO₂ (○), and NiEdda accessibility parameter ΠNiEdda (▴). (C) Fourier analysis of environmental profile data. (Top) Power spectra of the ΠO₂ and ΠNiEdda profiles. Peak angular frequencies are 102° and 100°, corresponding to an α-helical structure contacting two distinct environments. Below, both ΠO₂ and ΠNiEdda profiles have been plotted in a polar coordinate representation. A resultant vector was calculated for both parameter (ΠNiEdda, blue arrow) and O₂ accessibility (ΠO₂, red arrow) pointing to the direction of highest accessibility relative to residue L5. (D) Membrane immersion depth of NH₂-terminus spin-labeled residues. Positive distances indicate immersion towards the center of the bilayers, relative to the membrane interface. These values have been calculated from a calibration curve of spin-labeled phospholipids following the equation Φ = ln[Π(O₂)/Π(NiEdda)] = αD + C. Where Π(O₂) and Π(NiEdda) are accessibility parameters for O2 and NiEdda, α is the slope D immersion depth, and C the intercept. Negative values have a very large uncertainty associated and are shown for reference purposes. The continuous curve was drawn according to a geometric model of an α-helix with a 1.5-Å pitch and an insertion angle of 14° relative to the membrane plane.

Figure 3

Site-directed spin-labeling and structural analysis of the COOH-terminal end. (A) CW-EPR spectra of reconstituted spin-labeled NH₂-terminal mutants. Details are similar to those in Fig. 2 A. *Spectra have been obtained at underlabeling conditions so that on average there is one spin-label per tetramer. (B) Environmental parameter profiles: mobility parameter ΔHo⁻¹ (▪), oxygen accessibility parameter ΠO₂ (○), and NiEdda accessibility parameter ΠNiEdda (▴). Arrows indicate residues that show spin–spin interaction. (C) Fourier analysis of environmental profile data. The power spectra of the ΠO₂ and ΠNiEdda profiles are shown (top), with peak angular frequencies of 65° and 103°. This corresponds, in the case of the ΠNiEdda profile, to a strong α-helical signal. A resultant vector was calculated for both parameters (ΠNiEdda, blue arrow) and O₂ accessibility (ΠO₂, red arrow) pointing to the direction of highest accessibility relative to residue R127.

Figure 4

Intersubunit proximities and the distribution of spin–spin dipolar couplings at the COOH terminus of KcsA. (A) CW-EPR spectra of residues showing strong spin–spin interactions. A comparison is made between fully labeled channels (a maximum of four spin/channel), in red, and channels that show no spin–spin interaction by virtue of being underlabeled at a 1:10 label/channel molar ratio (so that on average they have 1 spin/channel), in black. (B) Spin–spin interaction parameter (Ω) calculated for the all residues of the COOH terminus. As the Ω parameter serves as a gross estimate of intersubunit proximity, this particular pattern of per-residue Ω has been taken to indicate the formation of a four-helix bundle. The gray bars represent charged residues.

Figure 5

Secondary structure assignment and three-dimensional architecture from SA/RMD runs. (A) Windowed periodicity analysis of the O₂ accessibility parameter (ΠO₂) from residues at the NH₂ terminus of KcsA. A nine-residue window was used to calculate the αPI (Cornette et al. 1987), and values ≥2 were taken as an indication of a statistically significant α-helix within the window. (B) Windowed periodicity analysis of the NiEdda accessibility parameter (ΠNiEdda) from residues at the COOH terminus of KcsA. Periodicities for standard α-helix (•) and coiled-coil helix (○) were analyzed as in A. Also shown is a plot of the α-coiled-coil propensity-calculated based on amino acid sequence according to the program COILS (Lupas et al. 1991). (C) Secondary structure assignment for the NH₂- and COOH-terminal regions of KcsA. (D) Stereo representation of Cα diagrams of the eight lowest energy symmetric structures assembled from 32 independent SA/RMD runs of KcsA residues 23–160. Numbers signal the approximate location of specific residues. This picture was drawn using MOLMOL (Koradi et al. 1996).

Figure 6

The full-length structure of KcsA and its correlation with the EPR data. Ribbon diagram representation of the mean full-length KcsA structure in relation to the membrane bilayer. (B) EPR-derived structural data mapped onto KcsA solvent-accessible surface. (Top left) Probe mobility (ΔHo⁻¹), (top right) spin–spin interaction (Ω), (bottom left) oxygen accessibility parameter (ΠO₂), and (bottom right) NiEdda accessibility parameter (ΠNiEdda). For clarity, only two subunits are displayed.

Figure 7

Functional consequences of COOH-terminal deletions. (A) Representative single-channel traces from full-length KcsA and chymotrypsin-truncated KcsA (Δ125) from reconstituted proteoliposomes. Stationary recordings were obtained at 80 mV in symmetric 200 mM K⁺, pH 4.0, and were filtered at 2 kHz. Calibration scales are 10 pA and 250 ms. (B) Single-channel I-V curves calculated for control (•) and Δ125 (○) channels. Single-channel conductances calculated from the slope around 0 mV are 100 and 102 pS. (C) Effect of progressive COOH-terminal deletions on the pH dependence of ⁸⁶Rb⁺ influx. Error bars represent the SD and correspond to three to nine experiments. Individual curves were fitted to the Hill equation and the corresponding parameters for each truncation construct (apparent pKa and Hill coefficient, n) are plotted as a bar graph (right).