Quantitative input-output mapping of cytoskeleton regulator localization demonstrates linearity in developing epithelia systems
Authors: Akanksha Sachan 1,2; Nilay Kumar 1; Alexander Dowling 1; Jeremiah J. Zartman 1
Affiliations: 1) University of Notre Dame; 2) IIT Bombay
Keywords: g. wing disc; a. cytoskeleton
The signalling pathways that operate in Drosophila are conserved with remarkable fidelity in all multicellular organisms and help us understand the emergence of birth defects and cancer in humans. For example, epithelial morphogenesis relies on coordination between multiple conserved cytoskeletal effectors; however, quantitative regulatory relationships governing cell mechanics and the tissue scale remain largely unknown. Here, we utilized immunohistochemistry, a semi-automated Python feature extraction workflow, and regression analysis to identify constitutive relationships between cytoskeletal regulators of cell mechanics in wildtype and mutant Drosophila wing imaginal discs. We evaluated possible correlation model functions between cytoskeletal regulators. In particular, we tested for linear regulation of actin accumulation by the strength of integrin concentration levels, which in turn regulates the basal curvature in the developing wing disc. RNAi-based knockdown of integrin with the Gal4/UAS system reduced basal actin levels, confirming integrin regulates actin. Both actin and integrins localize more in areas with increased local basal curvature and reduced cell height. Non-muscle Myosin II concentration also correlates with Integrin and actin concentration levels at the basal surface. Interestingly, basal actin concentration does not depend strongly on the spatial location along the Anterior-Posterior axis. Multiple polynomial relationships were analysed for predicting actin concentration, and a linear model with a single input of Integrin (or Myosin) as an independent predictor variable was selected based on the Akaike information criterion. Prediction error of the model was observed across different locations of the disc, to check how the model performs spatially. The standardized residual error was observed to be higher across the edges of the pouch region. In summary, concentrations of cytoskeletal effectors show a simple linear relationship, a signature feature of robustness in complex systems. Identification of highly collinear dependencies in biological systems also guides in constraining and calibrating biophysical models of epithelial growth and morphogenesis.