3D transmural arrhythmia modelling using matured engineered human atrial tissue

Introduction Atrial arrhythmias are critically shaped by atrial electrophysiology and three-dimensional (3D) geometry, which together modulate impulse propagation. Yet, mechanistic and translational studies are hampered by the lack of human in vitro models that support 3D arrhythmia dynamics. Existing models based on engineered atrial tissues (EATs) poorly reflect relevant geometry and display immature electrophysiology, including low conduction velocities (~5-15 cm/s), far below those measured in diseased human atria. Recently, we developed conditionally immortalized human atrial myocytes (hiAMs), which combine more mature electrophysiology with scalable expansion. Purpose To develop matured and geometrically defined in vitro models of 3D atrial arrhythmias using hiAMs for mechanistic and translational studies. Methods First optimization efforts focused on establishing classical EATs using polydimethylsiloxane (PDMS) molds featuring two pillars. Cell suspensions containing hiAMs were combined with collagen (type I) +/- Geltrex to support tissue formation. For arrhythmia studies, we engineered Large-EAT using molds (32x22x3 mm) featuring either eight small pillars or four pulmonary vein-like pillars. Electrophysiological parameters were evaluated by optical voltage mapping and sharp electrode measurements. Results Using classical EAT, optimal cell density and concentrations of collagen I and Geltrex were determined, showing an increased conduction velocity (CV) for EATs (33±1 (SD) cm/s) vs. monolayers (21±1 cm/s) at 1 Hz pacing. Sharp electrode measurements confirmed action potential durations (APD20 = 4.1±1.5 ms; APD50 = 28±8 ms; APD90 = 140±26 ms) consistent with human atrial action potential morphology. Compared to monolayers, EATs showed lower resting membrane potentials (-72±2 mV vs. -69±3 mV, p=0.04) and higher maximum upstroke velocities (138±33 mV/ms vs. 87±17 mV/ms, p=0.001). Pharmacological testing using the sodium channel blocker Vernakalant (30 µM) demonstrated increased, use-dependent, conduction slowing in EATs vs. monolayers. Next, Large-EATs (~0.5 mm thickness) displayed uniform conduction, similar CV vs. classical EATs, and longitudinal conduction anisotropy (ratio ~1.5-2). After inducing re-entrant circuits by burst pacing, these arrhythmias showed 3D activation patterns, including transmural breakthrough and “epicard/endocard” (i.e., transmural) dissociation. Dominant frequency analysis revealed spatial heterogeneity in activation frequencies. However, the highest stable frequencies did not reliably identify arrhythmia drivers, instead often reflecting 3D wavefront interactions or transmural dissociation. Contrary, monolayers only displayed spiral waves without frequency heterogeneity. Conclusion We established electrophysiologically matured, scalable engineered atrial tissues using hiAMs that support 3D transmural arrhythmias for mechanistic and translational studies.Figure 1Figure 2