Markov state models of proton- and pore-dependent activation in a pentameric ligand-gated ion channel

  1. Cathrine Bergh
  2. Stephanie A Heusser
  3. Rebecca Howard
  4. Erik Lindahl  Is a corresponding author
  1. Science for Life Laboratory and Swedish e-Science Research Center, Department of Applied Physics, KTH Royal Institute of Technology, Sweden
  2. Science for Life Laboratory, Department of Biochemistry and Biophysics, Stockholm University, Sweden
7 figures and 1 additional file

Figures

Global architecture of GLIC, electrophysiology data and computational methodology.

GLIC in an open conformation shown from (A) the top and (B) the side in a POPC lipid bilayer. (C) Two opposing subunits highlighting the pore of the channel. Light colors represent the open …

Figure 2 with 4 supplements
Free-energy landscapes capture shifts upon protonation and mutation.

Free-energy landscapes projected onto the first two tICA coordinates for (A) deprotonated wild-type, (B) protonated wild-type, (C) deprotonated I233T mutant, (D) protonated I233T mutant, (E) …

Figure 2—figure supplement 1
Simulations sample the initial transition pathway broadly.

Simulations sample the initial transition pathway broadly. Simulation seeds, experimental structures and overall sampling projected onto principal component space for deprotonated wild-type (A), …

Figure 2—figure supplement 2
Assessment of convergence and sampling of the MSMs.

Assessment of convergence and sampling of the MSMs. The first 10 implied timescales representing the 10 slowest processes of the MSM for (A) deprotonated and (D) protonated system. The vertical …

Figure 2—figure supplement 3
Variational optimization for selection of hyperparameters.

Variational optimization of the slowest timescale enables selection of hyperparameters. Open probabilities for all two-state models as a function of the number of microstate clusters (A). The width …

Figure 2—figure supplement 4
Vector representation of the two tICA coordinates.

Vector representation of the two tICA coordinates. The 20 largest eigenvector components of the two tICA coordinates projected onto the GLIC transmembrane domain. TIC 1 is overall more focused on …

Figure 3 with 3 supplements
Two-state clustering distinguishes metastable open and closed states.

(A) Two metastable states separated by the highest free energy barrier, with red representing closed-like states and blue open-like states. White dots indicate experimental structures, with labels …

Figure 3—figure supplement 1
Pore hydration in wild-type GLIC simulations.

Hydration of all 100 wild-type simulations, where each row represents one simulation with frames on the x-axis.The simulations have been grouped based on location of the starting seed in the …

Figure 3—figure supplement 2
Pore hydration in I233T mutant GLIC simulations.

Hydration of all 100 I233T simulations, where each row represents one simulation with frames on the x-axis. The simulations have been grouped based on location of the starting seed in the …

Figure 3—figure supplement 3
Pore hydration in H235Q mutant GLIC simulations.

Hydration of all 100 H235Q simulations, where each row represents one simulation with frames on the x-axis. The simulations have been grouped based on location of the starting seed in the …

Higher order clustering of the GLIC free-energy landscapes.

(A) Each free-energy landscape can be further clustered into models with four or five macrostates that – despite not being metastable– allow for more fine-grained structural analysis of the energy …

Figure 5 with 11 supplements
Probability distributions of a few variables proposed to be important in GLIC gating.

The left-most cartoons illustrate the definition of each variable, while data is presented as probability distributions with means and standard deviations plotted as bars. Colors represent the …

Figure 5—figure supplement 1
Distributions of variables proposed to be important in GLIC gating.

Probability distributions of a few variables proposed to be important in GLIC gating. The left-most cartoons illustrate the definition of each variable, while data is presented as four probability …

Figure 5—figure supplement 2
ECD spread and twist values show rapid adaption in response to pH.

ECD spread and twist values show rapid adaption in response to pH. Time series of ECD spread and twist values for wild-type GLIC at protonated and deprotonated conditions. Initial simulation seeds …

Figure 5—figure supplement 3
M2 spread values projected onto the first two tICs.

M2 spread values projected onto the first two tICs. Line contours represent the macrostates in Figure 4.

Figure 5—figure supplement 4
9’ distances projected onto the first two tICs.

9’ distances projected onto the first two tICs. Line contours represent the macrostates in Figure 4.

Figure 5—figure supplement 5
M2-M1(-) values projected onto the first two tICs.

M2-M1(-) values projected onto the first two tICs. Line contours represent the macrostates in Figure 4.

Figure 5—figure supplement 6
Beta expansion values projected onto the first two tICs.

Beta expansion values projected onto the first two tICs. Line contours represent the macrostates in Figure 4.

Figure 5—figure supplement 7
ECD upper spread values projected onto the first two tICs.

Line contours represent the macrostates in Figure 4.

Figure 5—figure supplement 8
The distance between residues E35 and T158 projected onto the first two tICs.

Line contours represent the macrostates in Figure 4.

Figure 5—figure supplement 9
-2’ distances projected onto the first two tICs.

Line contours represent the macrostates in Figure 4.

Figure 5—figure supplement 10
M1 kink values projected onto the first two tICs.

Line contours represent the macrostates in Figure 4.

Figure 5—figure supplement 11
ECD twist values projected onto the first two tICs.

Line contours represent the macrostates in Figure 4.

State- and protonation-dependent differences in ion channel symmetry.

Heatmaps show pairwise RMSDs between all subunits of the channel, measuring the conformational symmetry of the pentamer. The two top bars show pairwise RMSDs of the transmembrane (TMD) and …

Proposed models for the free energy landscape and symmetrization in GLIC gating.

Sketches of the free-energy landscapes for protonated and deprotonated (A) wild-type and I233T variants, and (B) wild-type and H235Q variants. An open-state free energy well is formed when the …

Additional files

Download links