A Study On Protein Backbone Design Based On A Statistical Energy Function

The ultimate goal of protein design is to create proteins with specific functions.Although there have been rapid progresses in this area,many problems remain about protein design,especially backbone design.In most protein design processes,one needs to first specify the desired topology and construct structural models of designable backbones,and then select the amino acid sequences for the backbones,probably with consideration of function-related restraints.In protein design literature,the two approaches to designing new backbones are the fragment assembling approach and the parametrization approach.Fragment assembling relies on available high quality protein structures,which are decomposed into fragments.These fragments are used as building blocks for assembling into new overall structures.The paramertrization approach only suits for the design of certain types of structures,such as helix bundles.In this approach,existing structure units are repetitively used to compose new structures,with the relative positioning between different units controlled by a parametric model.Neither approach allow unrestricted exploration of the designable backbone space to achieve on-demand design.This constitutes significant barrier for the development and applications of protein design.Our group have developed a backbone-centred statistical energy model named SCUBA(sidechain unknown backbone arrangement)to achieve the goal of designing new backbones without relying on the use of existing fragments.SCUBA aims at producing plausible,new backbones through continuous sampling and optimization without considering specific amino acid sequences.This thesis explored a series of protocols for carrying out backbone designs using SCUBA.First,this research explored a protocol in which an initial backbone is generated first,according to a chosen topology,and then the backbone structure is continuously sampled and optimized.Here,this research tried to design an extra new domain for an existing protein.this research then designed four helix bundles of various topologies.Both explorations produced proteins of experimental X-ray structures closely matching the designed models.After that,this research applied this protocol to change the fluorescence protein smURFP from dimer to monomer while retained its function.This research also tested the ability of SCUBA to explore plausible protein structure space without pre-defined topologies.This research generated initial structures of 6 randomly-placed helices without any predefined topology,optimized the structures with SCUBA and then used loops to link the helices into continuous chains.This research chose 13 finally designed sequences for experiment verification,and obtained 3 high-resolution structures which match closely the designed models.Finally,this research tried to design proteins each with an internal cavity.This would provide a basis for the design of a ligand-binding protein which requires a binding pocket.This research added pocket-attracting and repulsing potentials to SCUBA,and generated backbone conformations with properly-sized cavities.In final experimental examinations,eight of ten designed proteins are soluble,and one led to a high-resolution structure which showed an obvious cavity,in agreement with our designed model.In conclusion,this research developed and verified protocols for de novo protein backbone design with SCUBA,and showed proteins of various topologies,including novel ones not observed in nature,could be deigned to high resolutions.It is also a major progress that a de novo protein with a novel topology and a pre-defined cavity has been successfully designed.