Comprehensive optical and data management framework for sympathetic network investigation in large-scale infarct border-zone of pig hearts.

Myocardial infarction (MI) affects millions of people every year and those who survive have an increased risk of developing arrhythmias and to experience sudden cardiac death. Indeed, prolonged ischemia leads to cell death and can result in the adverse remodeling of the left ventricular tissue, mainly consisting in the formation of large, non- conductive fibrotic patches that provide a convenient substrate for the development of arrhythmias. However, recently the cardiac sympathetic nervous system (SNS) has also been observed to remodel after an episode of MI, thus potentially taking part in triggering arrhythmias. Among the arrhythmogenic events associated with MI, premature ventricular complexes (PVCs) have shown to be the most common ones and, moreover, to correlate with sympathetic stimulation in pigs. While a lot of evidence has been provided in regards to the fibrotic response that follows a MI, the nature and the extent of the SNS remodeling are still poorly comprehended, mainly due to methodological limitations that, over the years, have restrained previous research to bidimensional investigations or to the study of small animal models. Since the sympathetic network exerts significant effects on the electrophysiological properties of the heart, understanding the structural outcomes of this remodeling could offer important insights to comprehend the pivotal factors involved in determining whether or not arrhythmias would arise after an infarct. The aim of this PhD project was to develop a comprehensive methodological framework to three-dimensionally reconstruct the infarct border-zone (BZ) in large-scale pig heart slices at high resolution. In this proof of concept conducted on a small population, we wanted to assess whether the differences between arrhythmia-initiating (PVC) and not initiating sites (NO-PVC) detected in-vivo could rely on the structural features that the tissue develops during the remodeling process. To this goal, at first we had to solve the problem of light scattering in large-size biological samples, which has been one of the major limitations that have restrained previous studies. We exploited mouse hearts to test and optimize three of the most successful optical clearing methodologies (uDISCO, CLARITY and SHIELD) aimed at reducing the refractive index mismatch across the sample. We investigated their performances in terms of preservation of the tissue features and compatibility with a homogenous staining of the whole organ. Among the three tested protocols, the CLARITY technique showed to be the most effective one. Thus, we applied the developed clearing and staining approach to the selected pig heart slices, with a further implementation to avoid out-of-plain distortions of the samples, and we obtained 3D reconstructions of the entire volume of the samples by two-photon fluorescence microscopy. The high resolution of the images allowed us to exploit a machine-learning based software tool developed to map and analyze the detected structural features. We analyzed cellular disarray along with density and alignment coherency of the sympathetic innervation, in both PVC and NO-PVC sites, comparing them to remote sites isolated from the healthy distal myocardium. We found a solid consistency between the remote sites in terms of both cellular organization and neuronal density and coherency, thus validating our high-fidelity analysis approach. Moreover, we found a highly significant reduction (68 ± 15%) of innervation in the scar regions with respect to the myocardium in all the BZ samples, suggesting a strong denervation process that occurs in the fibrotic patches during the remodeling. An overall reduction of neurofilament alignment was also observed in the BZ samples with respect to the distal healthy myocardium which could benefit from investigating a larger cohort of samples. However, in this preliminary study no differences were found between arrhythmia-initiating and not- initiating sites. We believe that, due to the high relevance of 3D information achieved on a human-mimicking animal model, we have provided a clearer and more reliable view of the cardiac SNS remodeling that occurs after a MI. Moreover, the methodological pipeline that we developed has shown the capability of providing accurate three– dimensional reconstructions of massive samples isolated from large, human-resembling animal models, thus paving the way to the possibility of expanding the analysis to investigate different structural features that may be involved in the rise of arrhythmias.