(A) Effects of miR-26 and its antisense on AF in mice. Upper panels: representative atrial electrogram recordings. WT, control mice receiving vehicle injections; adv–miR-free, adenovirus vector without miR-26; MM LNA–anti–miR-26a, mismatched LNA–anti–miR-26a as negative control constructs. Burst pacing is highlighted by solid underlines, whereas dashed underlines indicate AF. Lower panels: percentage of animals with successful AF induction (left: results are shown as n/N, where n = number inducible into AF/total of N mice) and AF duration in animals with successful AF induction (right). *P < 0.05, ***P < 0.001 vs. WT; ^ϕϕϕP < 0.001 vs. LNA–anti–miR-26a. Note: here and in Figure 4, related data sets for AF incidence and AF duration are shown separately for clarity in display. However, statistical comparisons were performed between all animals with interventions simultaneously (see Supplemental Figure 6 for all comparisons), with the statistical comparisons here reflecting the results of simultaneous comparisons of all data in Supplemental Figure 6. (B) I_K1 in atrial myocytes isolated from mice treated with various constructs. Left panels: I_K1 recordings. Right panel: I_K1 density-voltage relationships. Results for adv–miR-26a and LNA–anti–miR-26a are shown with solid lines and the symbols defined on the figure; results for their controls (adv-miR free and MM LNA–anti–miR-26a) are shown with dashed lines. *P < 0.05 vs. WT; n = 12 cells/group. (C) qPCR verification of atrial miR-26 expression changes resulting from various constructs (note: anti–miR-26a is complementary to both miR-26a and miR-26b). ***P < 0.001 vs. WT; n = 8 mice/group. Values are mean ± SEM.