Molecular phenomenology of membrane-bound protein machines

The basic functions of living things are predominantly carried out by uniquely adapted protein nanomachines. Molecular phenomenology is an approach to biology driven by the most basic questions about these species: what are their atomic structures, and how do they move? The past two centuries have seen great strides in biology, yet we are only beginning to understand its fundamental units of function. In the past decade, revolutionary advances in electron microscopy have enabled direct imaging of biological molecules in solution, opening what may be the final frontier of molecular phenomenology. My doctoral work has focused especially on applying single-particle electron microscopy to the dynamics of those protein machines that lie at the plasma membrane that divides cells from their environment.In the first chapter of this dissertation I introduce a previously unpublished method for modeling protein motions through multi-body refinement and principal component analysis with geodesic kernels. While multi-body analysis is not new, other methods have not considered the intrinsic nature of rigid-body motions as a non-linear manifold. As such, they have been forced to rely on ad-hoc parameter scaling, and are valid only over very small motions where manifold curvature can be neglected. The kernel approach, in contrast, eliminates the parameter scaling problem and is applicable to arbitrary rigid-body motions. The basic method may be of use outside cryo-EM of biological molecules, for example in 3D animation of articulated figures or controlling swarms of drones or satellites. In the second chapter, I and others apply molecular phenomenology to uncover the mechanisms of action underlying antibody modulation of the fusogenic activity of the SARS-CoV-2 Spike protein. This work was the first study demonstrating that Spike-reactive antibodies can either inhibit or enhance Spike-mediated membrane fusion leading to viral entry or pathological cell-cell fusion and formation of syncytia (giant, multinucleated cells). The results epitomize the deep connection between basic biophysical understanding and biological function, by tying molecular structure and conformation to the virulence of an emergent pathogen responsible for millions of deaths and hundreds of millions of injuries. In the third chapter, I and others report high resolution structures of the murine Dispatched and its complex with Sonic hedgehog. I then use 3D variability analysis, a state-of-the-art phenomenological method, to uncover coupling between transmembrane ion flux and release of Sonic hedgehog from organizer cells during animal development. While multi-body refinement and 3D variability analysis both visualize interdomain motions, the latter also captures intradomain conformational changes by direct 3D reconstruction of orthogonal axes of particle covariance. This approach enables the production of true molecular movies that go beyond animations based on multi-body refinement. Dispatched bioactivity is characterized by subtle conformational rearrangements in its transmembrane domain that direct the energy of the plasma membrane Na+ gradient towards extraction of lipid-modified Sonic hedgehog from the plasma membrane and activation of Sonic hedgehog for efficient recruitment of its critical adapter protein SCUBE2. High-resolution molecular movies of active Dispatched, validation through traditional 3D classification, and detailed mutational analysis using a direct read-out of protein function yield one of the most complete studies in molecular phenomenology to-date, foreshadowing a new era of biophysics in which modeling dynamic structural ensembles may become de rigueur