Page 38 - YORAM RUDY BOOK FINAL

P. 38
Figure 2.21E shows traces of KCNQ1 single channel current during a depolarizing pulse,

computed with the model of Figure 2.21D; the traces reflect the random nature of channel
opening. By adding 1000 single channel currents, macroscopic KCNQ1 current is obtained
(Figure 2.21F). In Figure 2.22, simulated macroscopic currents (KCNQ1 in panel B and I in
Ks
panel C) are compared to experimental traces for wild-type channels and for channels with
various mutations at E160 (E1), a negatively charged residue on S2. As was shown by the
simulations above (Figure 2.20F), E1 interacts with positive charges on S4, affecting its motion
during gating. Three mutations were simulated computationally, and introduced experimentally
as well: E160Q, E160A and E160K (computationally and experimentally, E1 on S2 was replaced by Q,

A or K). These mutations progressively reduce the amount of negative charge at E1 (Q is polar, A
neutral, and K is positively charged) and correspondingly slow channel activation (in both
simulations and experiments). As revealed by the energy landscape, activation slowing of E160Q

is caused by a higher energy barrier for reaching the permissive state. In addition, the open
configuration is destabilized because R4 no longer interacts with the full E1 negative charge on
S2 at this configuration (R4 stabilization by E160Q is reduced by 30% relative to wild-type). Similar
destabilization occurs in the deep closed state, where R2 interacts with Q instead of E1; this
destabilization reduces the sigmoidicity of the simulated macroscopic current.

The energy landscape for E160A is very similar to that of E160Q, except for the absence of a
minimum near the intermediate closed state that is present in the E16Q landscape (Figure 2.22A,

arrow). As a consequence, E160A activation is slower than E160Q. The experiments also show a
significant reduction of E160A current (due to reduced channel expression, conductance, or both).
This reduction is accounted for in the simulation. Finally, for E160K there is electrostatic repulsion
between R4 and (positive) K that substitutes E1. This is reflected in the energy landscape of Figure
2.22A (right panel). Experimentally, E160K mutant channels do not generate current. This

mutation occurs naturally and has been linked to the long QT syndrome (LQT1).

As described earlier, I is a heteromeric channel that consists of KCNQ1 and its ß-subunit
Ks
KCNE1, which slows channel activation. In the model, the diffusion constant on the energy
landscape was decreased to account for slowing of the voltage sensor movement (experiments
indicate that KCNE1 is proximal to S4, suggesting direct effect of KCNE1 on S4 motion). As can be
seen in Figure 2.22, the simulated macroscopic currents (blue traces) resemble closely the
experimentally recorded currents (black traces) and reproduce the voltage dependent activation

of both homomeric KCNQ1 (panel B) and I (panel C).
Ks

Further methodology development 114-117 permitted us to construct a large conformational

space of the ion-channel protein based on 20 degrees of freedom that represent large-
scale conformational changes during gating. Analysis of these conformations determined that
117
during activation, outward translation (toward the extracellular domain) of the voltage sensor is
accompanied by translation away from the pore and a counterclockwise twist (viewed from the