Visualizing the Inter-Domain Motions of a Pathogenic Protein using Sparse RDC Data

Abstract

SpA-N is the N-terminal half of Staphylococcal protein A and it is composed of five protein binding domains. The five domains could bind to antibody, TNFR1 and von Willebrand factor and facilitate the evasion of Staphylococcus aureus into the human immune system. The functional plasticity is suggested to be the result of structural flexibility. Heteronuclear spin relaxation experiments demonstrated that the five domains of SpA-N are connected by four flexible linkers. In order to obtain additional dynamic information, we constructed a di-domain mimic of SpA-N with a lanthanide binding tag and measured residual dipolar couplings (RDCs). The di-domain construct can bind lanthanide ions and consequently be aligned in the presence of a strong magnetic field. By using different combinations of lanthanide ions and protein constructs, we obtained two orthogonal alignments, which have enough information content to determine a model with maximally 10 parameters. In addition, we designed a de novo method to extract dynamic information from RDCs. Instead of determining conventional structure ensembles, our method determines an inter-domain orientation distribution to describe the structure of a flexible protein using continuous probability distributions. By using continuous models, orientation distributions can be parameterized by a small number of variables and still remain general enough to describe a broad spectrum of inter-domain motions. As a result, the method can avoid the over-fitting problem even with sparse data. Because no force field is applied, the generated distribution is purely determined by RDC data and least biased. Using the method, we determined the inter-domain orientation distribution of the di-domain construct with only two orthogonal alignments. A strong correlation was observed in the distribution and conformations were well populated in a limited region.