Supplemental Material 2

We simulated a 50-cell cable (one-dimensional tissue) comprised of individual myocytes coupled via gap junctions and ephaptic coupling, as in prior work by us and others.^{7-9, 29-31} We accounted for non-uniform sodium channel subcellular localization by spatially discretizing each cell into multiple (10) axial membrane patches along the length of the cell and two intercalated disc (ID) membrane patches at the ends of the cell. Unless otherwise stated, 90% of sodium channels were localized in the ID patches. A T-shaped network of cleft axial resistances and a radial resistance, proportional and inversely proportional, respectively, to the intercellular cleft width (w<i>, </i>proportional to experimentally measured W<i>_p</i>), governed extracellular potentials at the ID and cleft. ID sodium currents and diffusion with the bulk extracellular space governed the extracellular cleft sodium ion concentration dynamics, which also depended on the cleft volume. <br>We utilized an established ventricular guinea pig myocyte ionic model³² representing individual ion channel dynamics, and incorporated a Markov model of a LQT3-associated Na_v1.5 GOF mutant³³ which reproduces a pronounced I_NaL. Specifically, the mutant channel model exhibited two modes of gating - a “background mode” and a “burst mode,” the latter of which represented a population of channels that transiently fail to inactivate. Full model equations, parameters, and details of the numerical integration are reported in our prior work.⁷