| Structure-based gene shuffling is proposed as a semi-rational method for directed evolution of proteins that enables construction of libraries of complex chimeras from non-homologous genes. Based on the concept of exon shuffling, coding segments of genes are designed to correspond to structural elements of proteins. Oligonucleotide primers are designed to code for segment boundaries and introduce variability between them as they direct the assembly of hybrid genes. Adapting the breeding scheme of Gregor Mendel into a strategy for DNA shuffling, hybrid genes are selectively "crossed" in an iterative process of molecular "inbreeding" until all possible combinations of coding segments are achieved.; This approach has been applied to rat DNA polymerase beta (Pol beta) and African swine fever virus DNA polymerase X (Pol X), two proteins with similar folds but low sequence identity, different size, and different activities. Libraries of chimeric genes have been constructed and screened via genetic complementation in E. coli. A number of chimeric Pol X-sized DNA polymerases with enhanced in vivo activity were produced and identified in this manner. Sequence analysis of complementing hybrid polymerases indicates that variability at the boundaries of combined segments is an important element of design.; The design principles developed here outline an approach for the construction of libraries with defined complexity and without the production of wild-type genes for the directed evolution of proteins. Structure-based shuffling allows the generation of libraries with greater "effective" complexity than existing techniques by accessing larger regions of sequence space through shuffling non-homologous genes. The "inbreeding" strategy enables their controlled synthesis resulting in partitioning of the library into simpler mixtures of discrete collections of unique members with the benefit of expanding the "numerical" complexity that can be screened. These approaches are applicable to the design of any chimeric library, and whereas there is never foreknowledge that a variant with a sought after property will be produced, the probability that all variants were evaluated is known. |
