Dynamic Cell-Matrix Interactions Dictate Thoracic Aortopathy

PROJECT SUMMARY – CORE B: SCIENTIFIC Altered cell signaling and gene expression secondary to predisposing genetic variants in thoracic aortopathy and related risk factors such as hypertension ultimately control the composition, structure, and function of the aorta that dictates its biomechanical properties and thus its potential vulnerability to aneurysmal dilatation, dissection, and rupture. There is, therefore, a pressing need to quantify the biomechanical phenotype of the diseased aorta and to relate it to its transcriptional profile. This biomechanical quantification must be multiaxial because of the complex loads experienced by the ascending aorta over a cardiac cycle. Core B will use unique, computer- controlled, multiaxial experimental devices for data collection and unique machine learning for data synthesis. Specifically, this Scientific Core will provide consistent biomechanical testing and computational modeling across all Projects that study mouse models of thoracic aortopathy, thus generating an essential data base from which general (rather than mouse-specific or lab-specific) conclusions can be inferred from an objective method for integrating data. In particular, we will use 2 validated multi-modality computer-controlled biomechanical testing devices to consistently quantify key geometric and mechanical metrics across all key mouse models used in Projects 1-4, testing equal numbers of male and female mice to study sex as a biological variable and testing young and older mice to study age as a biological variable. Importantly, computer-control of the testing protocols significantly increases reproducibility and rigor. For consistency, we will similarly test human aortic samples (from Project 2) using another device but the same cyclic biaxial testing protocols and the same nonlinear constitutive relation for quantification. Given the expected large, unique database in mice, we will also extend our new deep learning “constitutive artificial neural network” (CANN) to correlate the biomechanical phenotype and transcriptional profile in the mouse models, focusing initially on the top ~ 50 up- or down-regulated genes across Projects that relate to the mechano-biological axis: actomyosin, cell-matrix interactions, and matrix. Our recent work connecting biomechanics with 14 key measures of histology support this machine learning approach. This Scientific Core is significant for it will address a pressing unmet need for consistent multiaxial histological and biomechanical findings across diverse mouse models of thoracic aortopathy that can be correlated with biological findings as well as human data and it is innovative for it will enable a unique integrative analysis of a large data set (using a CANN) that promises to uncover true genotype – phenotype relationships for diseased aortas, based both on measured biaxial biomechanical changes and associated transcriptional changes.