|
1200 E. California Blvd.
Pasadena, CA
91125-9600
Mail Code: 114-96
Location: 130 Broad
(626) 395-6407
|
Research
Protein engineering is inefficient if each new protein functions identically
to the one before it or not at all. In order to determine which of seven library
design algorithms best introduces new protein function without destroying it
altogether, seven combinatorial libraries of green fluorescent protein (GFP)
variants were designed and synthesized. Each was evaluated by distributions of
emission intensity and color compiled from measurements made in vivo. Additional
comparisons were made with a library constructed by error-prone PCR. Among the
designed libraries, fluorescent function was preserved for the greatest fraction
of samples in a library designed using a novel structure-based computational
method. A trend was observed towards greater diversity of color in designed libraries
that better preserved fluorescence. Contrary to trends observed for libraries
constructed by error-prone PCR, preservation of function was observed to increase
with a library's average mutation level among the four libraries designed with
structure-based computational methods.
In order to investigate structure-function relationships with a reductive
approach that is more thorough and nuanced than site-directed mutagenesis,
one would ideally have many differently functional variants related
by the permutation of a small set of mutations. In this way, one would
isolate the effects of many individual mutations in many different
contexts. Not just any combinatorial library will do, however, since
random mutations at random positions are not likely to affect protein
function except to destroy it altogether. By targeting positions close
to green fluorescent protein's chromophore with mutations that have
been computationally pre-screened for their effects on stability,
we have generated considerable spectral diversity in a library of
only 29 variants. We have sequenced many of these functional variants
and show that 3 of 9 designed mutations are chiefly responsible for
much of the observed diversity of function. Using these mutations
and the T203Y mutation that defines the class of yellow fluorescent
proteins, we have constructed and characterized a quadruple mutant
cycle for which substantial emission is observed in vivo for 12 of
16 variants. We find that the effects of these four mutations on emission
and excitation spectra can be quite different depending on context.
Many observations suggest that vibronic mechanisms underlie much of
the observed spectral changes.
Computational protein design protocols use energy functions to score
interactions between pairs of sidechains in the context of a given
fold. These energy functions guide the search for a sequence or
library of sequences that will stabilize the folded structure. The
accuracy of the energy function is a limiting factor in designing
stable, well-folded proteins and can also hinder efforts in enzyme
design. We are interested in improving the energy terms that correspond
to electrostatic interactions and the solvation of polar groups
in the ORBIT protein design program. To this end, we have
incorporated into ORBIT the Finite Difference Poisson Boltzmann
(FDPB) model, which is generally considered a standard for accuracy
among continuum electrostatics models. In order to use the FDPB
model for protein design, we formulated it to be compatible with
'two-body' scoring functions, where each 'body' is a sidechain conformation.
We have found that, for the FDPB solver DelPhi, using approximate
representations of the protein surface and a two-body energy model,
it is possible to reproduce the results of traditional FDPB calculations
in which the entire dielectric environment around the protein is
defined. Using this model, we have designed a variant of the all
alpha-helical protein, Drosophila engrailed homeodomain, by only
varying the identities of amino acids on the protein surface. Experimental
characterization of this designed sequence has shown that it is
thermophilic and unfolds at a higher temperature than sequences
designed using electrostatic heuristics to bias sequence composition.
This result shows that the two-body requirement of most protein
design protocols does not preclude use of an FDPB model. This study
also underscores the power of electrostatic optimization of a protein's
surface as a means to stabilize the folded state.
home | research | lab
members | publications | contact | links | group
only
© Calfornia Institute of Technology
|
