Utilizing Optoacoustic Tomography to Measure Transmembrane Voltage

Catheter-based electro-anatomical mapping in clinical settings and optical mapping in basic sciences are gold standards for measuring cardiac electrophysiology (EP) non-destructively at high resolution. However, both methods primarily provide information about EP near a cardiac surface. Studies using plunge electrodes (both in animals and patients) and plunge optrodes (animals only) have been used to gain insight into transmural EP. These have shown that most re-entrant arrhythmias in the ventricles involve transmural re-entry – which is difficult to assess using surface mapping approaches. There is, thus, a need for a tissue-non-destructive, transmurally-resolved approach for measuring EP in native hearts. Optoacoustic (OA) tomography shows great promise for addressing this challenge by combining optical stimulation and acoustic imaging modalities. [1] The primary aim of this study is to measure transmembrane voltage ex vivo across different model systems, and to enhance data quality by developing a customized denoising pipeline. As a first step, we identified reporter agents such as the voltage-sensitive dye BeRST1 (and the genetically-encoded voltage indicator Archon1; not used in the results below) that allow OA assessment of transmembrane voltage, opening the possibility for time-resolved, intramural imaging of cardiac EP. Resulting voltage signals exhibit a low signal-to-noise-ratio (SNR), which restricts detailed functional analysis of data. To address this, we applied machine-learning based denoising algorithms to extract EP information by increasing the SNR.

OA voltage images were obtained from genetically modified HEK cells and Langendorff-perfused hearts using a multispectral OA tomography system (iThera Medical). The transmembrane voltage of cells was controlled optogenetically, using expression of a light-activated ion channel (CheRiff, for depolarisation) and a voltage-sensitive potassium channel (Kir2.1, for repolarisation) to generate controlled 1.0 Hz-cycles of membrane voltage changes, confirmed utilizing BeRST1. Langendorff-perfused rabbit hearts were loaded with BeRST1, electromechanically uncoupled (using 2,3-butadione monoxomine) and OA-imaged during electrical pacing at 2.5 Hz. The developed denoising pipeline is based on the DeepCAD-RT algorithm, [2] which – together with appropriate pre-processing – allowed an up to 4-fold improvement of SNR in cells, and a 3.5-fold improvement of SNR in hearts (see Figure 1). 

In future work, acquisition and denoising of transmembrane voltage data will be extended to more complex (e.g. arrhythmic) signal patterns in cardiac tissue to assess whether it is possible to not only improve temporal, but also spatial resolution. Once validated, our approach will be applied to assessing transmural EP properties in whole hearts during normal and disturbed heart rhythms.

References:

[1] Ntziachristos V. et al. Chem. Rev 2010/110:2783-2794

[2] Li X. et al. Nat Biotechnol 2023/41,282-292