Abstract

Genetic variation of alpha-1 antitrypsin (AAT) is responsible for alpha-1-antitrypsin deficiency (AATD) leading to gain-of-toxic aggregation in the liver and loss-of-function on neutrophil elastase (NE) inhibitory activity in the lung contributing to chronic obstructive pulmonary disease (COPD) during aging. To probe the molecular basis for how biology designs the protein fold to achieve balance between sequence, function and structure contributing to AATD in the population, we measured the intracellular monomer and polymer, secreted monomer and polymer and NE inhibitory activity of 75 alpha-1-antitrypsin (AAT) variants. To address the complex folding dynamics affecting the form and function of the protein fold that is differentially impacted by variants in the population, we applied a Gaussian process regression (GPR) based machine learning approach termed variation spatial profiling (VSP). By using a sparse collection of extant variants to link genotype to phenotype, VSP maps spatial covariance (SCV) relationships that quantitate the functional value of every residue in the wild-type (WT) AAT sequence with defined uncertainty in the context of its protein fold design. The SCV-based uncertainty allows us to pinpoint critical short- and long-range residue interactions involving 3 regions-the N-terminal (N1), middle (M2) and carboxyl-terminal (C3) of AAT polypeptide sequence that differentially contribute to the balance between function and misfolding of AAT, thus providing an unanticipated platform for precision therapeutic development for liver and lung disease. By understanding mechanistically the complex fold design of the metastable WT AAT fold, we posit that GPR-based SCV provides a foundation for understanding the evolutionary design of the fold from the ensemble of structures found in the population driving biology for precision management of AATD in the individual.