Connecting nanoscale images of proteins with their genetic sequences.

We present a technique for reconstructing biomolecular structures from scanning force microscope data. The technique works by iteratively refining model molecules by comparison of simulated and experimental images. It can remove instrument artifacts to yield accurate dimensional measurements from tip-broadened data. The result of the reconstruction is a model that can be chosen to include the physically significant parameters for the system at hand. We demonstrate this by reconstructing scanning force microscope images of the cartilage proteoglycan aggrecan. By explicitly including the protein backbone in the model, we are able to associate measured three-dimensional structures with sites in the protein primary structure. The distribution of aggrecan core protein lengths that we measure suggests that 48% of aggrecan molecules found in vivo have been partially catabolized at either the E(1480)-(1481)G or E(1667)-(1668)G aggrecanase cleavage site.