Supplementary Figure 4: Network analysis of the Kv1.2 simulations.

(a-d) Shortest pathways along the covariance network of the activated-state Kv1.2 simulation. Pathways begin at Arg 365 (R2; Shaker numbering) in each subunit and end at Val 474. The choice of sink residue (which subunit containing Val 474) was chosen based on the shortest path between each of the four possible Val (Supplementary Table 3). For consistency, colors for each subunit are the same as those in Fig. 2. For two of the subunits, the shortest path remains entirely within one subunit as it travels down S4, along the S4-S5 linker to Val 474 in S6 whereas for the other two, the shortest path goes from S4 to the neighboring S5 and moves down S5 to S6 rather than S6. (e-f) For subunits shown in panels C and D, the intersubunit pathways are dominant even when the sink residue is a V474 on the same subunit. (g-h) Betweenness is calculated in the activated state, using the Valine on the same subunit as the sink for subunits C and D. The intersubunit pathway is consistent when going to either the same or adjacent subunit valine. (i-l) Shortest pathways along the covariance network for the resting-state simulations. The source residue is Arg 365 in each subunit and V474. The shortest path remains entirely within one subunit as it travels down S4, along the S4-S5 linker to Val 474 in S6. In the resting state, unlike the activated state, there is no intersubunit pathway. (m-n) Experimentally determined long range interactions are shown along the optimal pathway in subunit A in the activated state and subunit D in the resting state. It is possible that these residues show interactions while being distant from one another because they are on pathways allosterically linking the VSD to the pore domain.