Biology Reference
In-Depth Information
concave features and cavities that could harbor functional sites is arguably harder, as the
fold needs to be stable enough to not collapse around and bury the cavity. Computational
algorithms to generate backbone templates that feature cavities and are at the same time
designable have yet to be developed.
COMPLEMENTARITY WITH DIRECTED EVOLUTION
Before the onset of the computational methods described in this chapter, the design and
engineering of protein function was mainly carried out by directed evolution (DE). Many
powerful high-throughput screening techniques (HTS) have been developed to address the
various engineering challenges, such as a number of display strategies for evolving
protein
protein interactions (e.g. phage-, yeast-, E. coli -, and ribosome-, or mRNA-display)
and growth selections, microfluidic/in vitro compartmentalization systems, or plate screens
for the evolution of new catalysts. The details and relative advantages of these different
approaches have been comprehensively reviewed elsewhere. 64 Computational approaches
and DE are by no means mutually exclusive. Since each of these two methods offers its own
unique advantages but also has shortcomings, they are perfectly complementary, and we
anticipate that these methods will usually be applied hand-in-hand in future design efforts.
This prognosis is supported by the fact that many of the designed proteins described in this
chapter have been optimized by DE.
Perhaps currently the biggest shortcoming of CPD methods is the inaccuracy of the energy
function, both in terms of estimating the precise energetics of a candidate-designed
sequence and in terms of translating this estimated value into a precise estimate for the
functional parameter of interest, i.e. a reaction rate or a dissociation constant. However,
energy functions are certainly accurate enough to identify the usually extremely small subset
of sequence space that is considered compatible with a function of interest, and, as
described earlier in this chapter, current algorithms are fast enough to search through
sequence spaces of size 10 130 and above.
The sequence throughput of DE methods can approach 10 12
119
for some of the in vitro
10 8 for in vivo display approaches and growth selections, and 10 4
for plate screening methods, and is thus far below that of CPD. Considering that most
active or binding sites are comprised of one or two dozen residues (i.e. a sequence space
of
display techniques,
B
10 26 for a 20-residue site), no currently available HTS method can cover more than a
tiny subset of possible sequences in a functional design problem. In addition,
experimental work is usually much more laborious and resource-intensive than
computation. However, DE allows for the selection of the best variant from the sequence
pool based on the actual activity of interest instead of based on an (inaccurate) energy
function value.
B
Thus, an ideal way to combine CPD and DE (assuming that an HTS assay is available) is to
use computation to come up with initial designs, and then use those initial designs that
show measurable activity as a foothold in sequence space and starting point for successive
rounds of DE. This strategy was used for several of the cases described here, and the
reported successes could not have been achieved with either CPD or DE alone. In Fleishman
et al.
s 24 influenza binder, the affinity of the initial computational designs was increased
10-fold. In Khare et al.
'
s 4 designed organophosphate hydrolase, the best variant after DE
was four orders of magnitude more active than the original design, but to obtain activity by
DE alone, a very large library containing all possible quadruple mutations of the scaffold
'
s
'
s two-loop graft, 32 the authors directly
compared a computation-informed library to a naïve library, and were able to isolate much
tighter binders from the former. Further, in the future, with both advanced DNA synthesis
and deep sequencing technologies becoming more readily available, we expect that
active site would have to be screened. In Azoitei et al.
'
Search WWH ::




Custom Search