Biology Reference
In-Depth Information
TABLE 6.1
Computationally Designed Proteins
Synthetic Biology
Application
Protein Design Task
Inputs Required
Representative
Examples
Redesign of cell
signaling
Protein
protein interface redesign
Crystal structure of complex
19
Interfering with
target proteins
Proteinprotein interface redesign,
de novo protein interface design
Crystal structure of target for de novo
design, structure of complex for
redesign
2, 16, 17, 24
New catalytic
activity/
metabolic
pathways
Enzyme redesign or de novo
enzyme design
Crystal structure of enzyme with
substrate for redesign, theozyme for
de novo design
3, 4, 37, 52
Small molecule
binders and
sensors
Same as for enzyme design, without
catalytic machinery
Crystal structure of wild-type receptor
with ligand
None yet
Material design
Protein
protein interface redesign
or de novo protein
Crystal structure of scaffold protein
None yet
protein
interface design
Stability increase
Monomer design
Crystal structure of protein of interest
5, 59
on an underlying structural model of the protein,
6
these approaches will not be described in
this chapter.
The computational methods and algorithms typically used in CPD can broadly be divided
into three categories:
103
1.
Side-chain placement algorithms that, given a model of the protein backbone, select a
set of amino-acid side-chain conformations compatible with that backbone. Since the
amino acid identities of the selected side-chain set can be any of the 20 natural amino
acids (or even unnatural ones), the sequence design happens at this stage.
2.
Backbone conformation-generating algorithms, whose purpose is to generate models
of backbone conformations according to the requirements of the specific design task. The
backbone models generated by these algorithms are usually passed to side-chain place-
ment algorithms for sequence design.
3.
Rigid-body placement algorithms, which are used to place two protein models or a pro-
tein and a small molecule model in a relative spatial orientation to each other. These
algorithms are often the first step when designing binding or catalytic proteins, where
one has to design a functional site on one of the proteins, and thus needs to first design
the spatial relation of the two interacting partners.
CPD algorithms from all three categories generally employ classical molecular mechanics
7
representations of the designed system. All atom models of the protein are used, where
bond lengths and angles are usually held constant to increase computational speed.
Similar to molecular dynamics force fields, CPD energy functions usually contain terms
for van der Waals interactions, bond-dihedral potentials, hydrogen-bonding, simplified
electrostatics, and implicit solvation.
8
A peculiarity of CPD energy functions is the
additional inclusion of sequence composition terms, which are parameterized to make the
distributions of amino acid identities in designed sequences similar to those in natural
proteins. In this section, we will focus on describing the most often used side-chain
placement algorithms (category 1), since these are broadly utilized in virtually all CPD
calculations. Examples of category 2 and 3 algorithms will then be presented in
subsequent sections, along with the specific design tasks these algorithms are meant to
address.
