12.07.2015 Views

Protein Engineering Protocols - Mycobacteriology research center

Protein Engineering Protocols - Mycobacteriology research center

Protein Engineering Protocols - Mycobacteriology research center

SHOW MORE
SHOW LESS
  • No tags were found...

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Combinatorial <strong>Protein</strong> Design Strategies 5such sequence can then be experimentally realized using peptide synthesis orgene expression, to confirm its folded structure and other molecular properties.Early efforts in design were guided by trends observed among naturally occurringstructures and identified protein sequences that were compact, with substantialsecondary structure but not necessarily well-defined tertiary structures(5). With their abilities to quantify and tabulate interresidue interactions, computationalmethods have dramatically accelerated successful protein design.Typically, such methods cast the sequence search as an optimization process, inwhich amino acid identity and side-chain conformation are varied to optimizea scoring function that quantifies sequence structure compatibility. Exhaustivesearching of all m N possible sequences is feasible only if a small number ofresidues N are allowed to vary or the number of allowed amino acids m is substantiallyreduced, e.g., from m = 20 to m = 2. To arrive at sequences with wellpackedinteriors having favorable interatomic interactions on average, thesearch must also include variation in the different side-chain conformations(rotamer states) of each amino acid (see ref. 6). As a result, the complexity ofthe search increases, because m, the number of possible states for a residue,increases by a factor of 10 or more, depending on the number of rotamersassociated with each amino acid. If only a few residues are allowed to varyand the conformations of the remaining residues are constrained, complete enumerationof all possible combinations can be performed to identify low-energysequence–rotamer combinations. Such complete enumeration is typically notfeasible, because of the exponential dependence on chain length and numbersof rotamers. For such cases, the sequence space can be sampled in a directedmanner to move progressively toward optimal (or nearly optimal) sequences.Stochastic methods, such as genetic algorithms and simulated annealing,involve searching sequence space in a partially random fashion, in which, onaverage, the search progressively moves toward better scoring (lower energy)sequences (7–10). Such searches have sufficient “noise” or recombination topermit escape from local minima in the sequence–rotamer landscape. Whenapplied to atomically detailed representations, the stochastic methods focus primarilyon repacking the interior of a structure with hydrophobic residues (9)and have been applied to the wild-type structures of 434 Cro (10), ubiquitin(11), the B1 domain of protein G (12), the WW domain (4), and helical bundles(13,14). Although, in many cases, these methods have aided in identifyingexperimentally viable sequences (4,15), stochastic search methods need notidentify global optima (16). For potentials comprising only site and pair interactions,elimination methods, such as “dead-end elimination” can find theglobal optimum (16–20). Such methods successively remove individual aminoacid–rotamer states that can not be part of the global optimum until no furtherstates can be eliminated. The Mayo group has applied such methods to automatethe full sequence design of a 28-residue zinc finger mimic (21) and, after

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!