| Many organisms produce inorganic materials via protein-influenced crystal growth---a process known as biomineralization. Understanding this process would shed light on hard-tissue formation, guide efforts to develop biomaterials, and provide a deeper understanding of in vivo phase-boundary biophysics. Unfortunately, the structure of biomineral-associated proteins cannot be determined by X-ray crystallography or solution-state nuclear magnetic resonance (NMR); therefore, existing structural models of biomineral-associated proteins are inferential, incomplete, or low resolution. In this thesis I develop both computational and combined computational-experimental approaches to investigate protein structure and sequence determinants in biomineralization. The methodological centerpiece of this thesis is the development of a comprehensive suite of computational structural biology tools (RosettaSurface). This software package includes modules for rigid-body protein-surface docking, protein folding at the biomineral surface, solid-state NMR-biased structure determination of protein adsorbed states, and de novo design of biomineralization systems. These represent the first structure-prediction techniques developed for and applied to biomineralization problems. I use these algorithms to investigate a model system: human-salivary statherin and hydroxyapatite (hydroxyapatite is the primary mineral component of mammalian skeletal and dental tissue). Investigation of this model system culminates in the first-ever reasonably high-resolution structure of a protein adsorbed to a solid surface determined using a combined experimental-computational approach. Remarkably, I had predicted this structure a priori to high accuracy using RosettaSurface in the absence of experimental bias, including the accurate prediction of a molecular recognition motif. The last development of the RosettaSurface algorithm that I present in this thesis is the design module. This algorithm simultaneously optimizes protein fold, orientation, and sequence while adsorbed to a mineral surface. I used that program to design peptides to bind different crystal surfaces of the mineral calcite. We chemically synthesized a set of the designer peptides that were predicted to have a high affinity for calcite, introduced designer peptides to solutions of growing calcite crystals, and observed the resulting morphological changes using a scanning electron microscope. All designed peptides had significant biomineralization activity and we found a dependence on basic amino acid content for sequence-order specificity. |
