Understanding the specificity of intramembrane proteases using biophysical principles and artificial intelligence

Intramembrane proteases regulate a large number of important biological and pathophysiological processes, but the principles underlying their substrate recognition are still only rudimentarily understood. The collaborative groups of Prof. Steiner and Prof. Frishman (LMU and TU Munich) have developed a methodology based on physicochemical profiling and explainable artificial intelligence (eAI) that can detect and distinguish biophysical features of substrates and non-substrates of the intramembrane protease gamma secretase, the enzyme responsible for the pathogenesis of Alzheimer's disease. In collaboration with these groups, we will analyze a set of substrates and non-substrates of the human intramembrane rhomboid protease RHBDL2. This enzyme regulates EGFR signaling in keratinocytes, potentiates proliferation, migration and invasion of pancreatic cancer cells, and has been implicated in lung epithelial homeostasis. The supervisor's laboratory studies the biological and pathophysiological functions of RHBDL2 in cellular and mouse models, which includes substrate identification by established proteomic methods. As part of this project, we will analyze the biophysical characteristics of RHBDL2 substrates, and will expand the set, which will include proteomic analyses, characterization of cleavage of RHBDL2 substrates in vitro and in cells, and preparation of recombinant proteins. Iterative refinement of the eAI-based substrate specificity model will result in a predictive model that we will then apply to the proteome of selected tissues/cell types with biological relevance. We may next apply this approach also to the human mitochondrial rhomboid PARL, which has a key function in mitochondrial biology and is relevant for Parkinson's disease.

Field of study: Developmental and Cell Biology; Molecular and Cell Biology, Genetics and Virology; Biochemistry