Protein Engineering Protocols - Mycobacteriology research center

Combinatorial Protein Design Strategies 7design of specific sequences and can also highlight sites likely to tolerate mutationwith minimal impact on structure; such sites can be targets of variationafter multiple rounds of protein design.Probabilistic methods may be used in several ways to guide protein design.Sequences should be generated in a manner consistent with the calculated probabilities.First, the most straightforward choice is a consensus sequence or thesequence comprising the most probable amino acid at each position. If necessary,repeated calculations may be performed with successive iterations determiningan increasing fraction of the residues in the protein. Such an approachhas been used to arrive at a 114-residue, dinuclear metalloprotein (32) and a solubilizedvariant of an integral membrane protein (33). Second, the calculatedprobabilities may be used to guide a search for sequences. A Monte Carlo-basedmethod has been presented, wherein the calculated amino acid probabilities areused to bias the selection of trial sequences that are either accepted or rejectedat each point in the Monte Carlo Markov trajectory (34). Such methods addresscorrelated amino acid identities, but at the cost of the computational overheadassociated with the search, although this overhead is diminished if informationis used to guide the search. Last, probabilistic methods may be used to quantitativelyguide the design of combinatorial libraries of proteins (35).1.4. Combinatorial ExperimentsCombinatorial protein experiments may be used to investigate sequence structurecompatibility and to discover novel sequences folding to a specific structure.In protein combinatorial design experiments, large numbers of sequences(libraries) are screened for evidence of folding to a predetermined structure.Depending on how the sequence diversity is generated and assayed, experimentsof this type can explore a large number of sequences, up to 10 12 sequences (36).A library can then be screened using a selection assay, such as ligand binding orenzymatic activity. Such experiments can go “beyond the protein sequence database,”in a manner in which the diversity of the sequences is at the control of theresearcher. Features important to folding (and other biological properties) maybe explored in a manner decoupled from the evolutionary pressures of nature’sproteins. Combinatorial methods have been used to identify helical proteins(37–39); ubiquitin variants (40); self-assembled protein monolayers (41); proteinswith amyloid-like properties (41); metal-binding peptides (42); and stableinterhelical oligomers (43). Several excellent reviews of combinatorial experimentsand methodology have appeared recently (44–47).2. MethodsProbabilistic methods for protein design provide estimates of the site-specificprobabilities of the amino acids within a particular protein structure. Here, we