Biology Reference
In-Depth Information
strategies to alter protein properties in a defined and predicted manner. Computational tools
that assist in predictive protein design strategy are continuously improving.
Owing to the great difficulties in understanding protein sequence
function
relationships, protein engineers must often accept some level of naivety and turn to a less
predictive design approach. In general, the alternative strategy is to clone and express many
different variants (create a combinatorial library), and then use a functional assay to screen
each library member to select variants with improvements in the property of interest.
Often this strategy of mutagenesis and selection is implemented iteratively, essentially
mimicking Darwinian evolution. Parameters such as the positioning of mutations, extent
of sequence diversity, and number or percentage of library members assayed can vary
considerably and depend primarily on the extent to which sequence
structure
function relationships
are understood and the nature of the function assay used to screen or select variants
(e.g. throughput and accuracy).
Rational design by site-directed mutagenesis has the potential to effect drastic
improvements in protein function, for example when coordinated mutations in a binding
pocket successfully improve binding or activity toward a nonnative substrate.
Unfortunately, this approach is not always as fruitful as one may hope, and predicted,
targeted mutations often do not furnish any improvements. Directed evolution by random
mutagenesis, on the other hand, has a high rate of success. However, since this approach
involves successive accumulation of low numbers of mutations, and since in any given
round of screening only a small number of possible mutations are tested, only moderate
improvements in function are typical in each round of screening (e.g.
two-fold).
Recombination of multiple parental sequences offers opportunities for more drastic
improvements, given that a much larger number of functional sequence changes are
accessible (i.e. a large number of
,
'
'
point mutations relative to a single
parent are presented in the recombinant library). As understanding of individual protein
sequence
fitness-conferring
25
structure relationships improve, strategies of site-directed and combinatorial
mutagenesis can be combined and often result in the most drastic improvements in
protein engineering.
Rational Design
Accurate prediction of mutagenesis strategies that improve protein properties of interest are
becoming increasingly possible due to: continued progress in protein structure
determination, increasing numbers of sequences and sequence
structure relationships
populating research databases, growing insights into structure
function relationships, and
advanced computational and molecular modeling tools to predict protein structure and
dynamics. Visual inspection of
crystal structures of proteins in complex with ligands
has been a common approach to rational design altered ligand binding. Structure analysis
might identify locations within a binding pocket to substitute hydrogen bond donors or
acceptors to improve ligand affinity, or replace a bulky residue with a smaller one to allow
access of a larger substrate. Once one or more desired mutations are determined, the
corresponding gene is synthesized and cloned. While sometimes successful, there are
obvious limitations with this technique alone, given the dynamic nature of proteins and
amino acid side-chains.
'
static
'
Improvements upon this
approach involve levels of increasing computational
complexity. This topic is covered in detail in Chapter 7, and is therefore only mentioned
briefly here. As an example, a ligand of interest may be computationally docked into the
protein binding pocket and its lowest energy position can be predicted. Side-chain
replacements can then be iteratively tested, each followed by energy minimization
calculations in which ligand and/or neighboring side-chain conformations are sampled to
reach a lowest energy complex. This may be performed with or without fixing the protein
'
inspection
'
'
s
