Atrial tissue engineering using conditionally immortalized human myocytes: establishing matured and geometrically defined 3D models of atrial arrhythmias

Atrial impulse generation and propagation is largely determined by electrophysiological and geometrical parameters of the atria, which thereby also affect arrhythmia dynamics and therapeutic outcome. However, current human three-dimensional in vitro models of atrial arrhythmias poorly reflect relevant geometry and show immature electrophysiology. In particular, they exhibit low conduction velocities (CV), which are far from those measured in diseased human atria (~20-40 cm/s). This limits the translational relevance of studied interventions, including drugs targeting anatomical re-entry, such as atrial flutter. Recently, we developed conditionally immortalized human atrial myocytes (hiAMs), which combine electrophysiological maturity with rapid, scalable expansion.

To further increase translatability of human models of atrial arrhythmias, we developed scalable, geometrically defined engineered atrial tissues (EATs) with matured electrophysiological properties using hiAMs.

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 in the PDMS mold. Electrophysiological properties were evaluated via optical voltage mapping. The optimized EATs were further characterized by sharp electrode measurements and RNA-sequencing. To model atrial flutter-like re-entry, we engineered circular EATs using 2 cm-diameter PDMS molds.

Using only collagen as extracellular matrix, we identified an optimal cell density. Next, the addition of Geltrex dose-dependently increased CV, with a mean of 32.5 ± 1.3 (SD) cm/s for the optimal hiAM-EAT condition vs. 20.5 ± 1.1 cm/s in hiAM-monolayers upon electrical 1 Hz pacing. Importantly, EATs from multiple consecutive passages consistently showed reproducible electrophysiological outcomes over two independent batches. Sharp electrode recordings confirmed a resting membrane potential of −75 ± 1.8 mV, maximum upstroke velocity of 138 ± 32.6 mV/ms, and action potential durations (APD20 = 4.1 ± 1.5 ms; APD50 = 27.6 ± 8.1 ms; APD90 = 140 ± 26 ms) consistent with human atrial action potential morphology. Next, we assessed the use-dependent sodium channel blocker Vernakalant in EATs and monolayer, showing an increase in frequency-dependent conduction slowing for EATs. In circular EATs, a single anatomical re-entrant circuit at 4 Hz (i.e., typical atrial flutter frequency) was induced by S1S2 pacing. Vernakalant failed to terminate anatomical re-entry in EATs—consistent with clinical findings—whereas electrical cardioversion did result in termination.

We successfully developed geometrically defined hiAM-EATs with matured electrophysiological properties that recapitulate clinical findings, providing a scalable platform for disease investigation and therapeutic testing.Figure 1

Engineered Atrial Tissues