Recent advances in computational protein design
Highlights
► Modeling of protein function hinges upon accurate structure prediction. ► Proteins must be stable at the desired operating conditions. ► Inter-molecule interactions dictate the specifics of protein function. ► Successful de novo protein design requires meeting multiple criteria.
Introduction
Computationally designing proteins is a crosscutting challenge that directly impacts many scientific and engineering endeavors, ranging from improved catalytic activity, genetic circuits, biosensors, chiral separations, creation of gene switches, and signal transduction pathways. Although purely experimental design efforts relying on combinatorial library construction and screening have been widely successful, the lessons learned do not easily generalize to inform the redesign of other systems. Proteins have been previously computationally designed to bind new ligands [1], proteins [2], and nucleic acids [3], to improve protein stability [4, 5], as well as to introduce novel enzymatic activity [6, 7], demonstrating that the fundamental rudiments of molecular recognition and interactions can be adequately captured via computational design. Despite these successes, predictably changing or even improving a protein's function in response to a performance target remains a formidable challenge. Successful de novo computational protein design requires accurate structure prediction, protein stability at the desired operating conditions, and correct modeling of the protein's interactions with other molecules (e.g. substrates, ligands, and cofactors). As illustrated in Figure 1, this review will discuss advances reached over the past couple of years in addressing each of these design challenges as well as examples where all three have been brought to bear in de novo design efforts.
Section snippets
Modeling and predicting protein structure
Reliable protein structure prediction is paramount in protein design, as protein geometry and flexibility along with proper presentation of charges and molecular groups on the surface determine function (or lack-thereof). The central dogma behind protein structure prediction is that the native structure reaches a conformation that achieves (near) global minimum energy. The bi-annual Critical Assessment for protein Structure Prediction (CASP) benchmarks the current state of the art in protein
Designing stabilized proteins
After an appropriate structure for a protein has been modeled, care must be taken to ensure that it will be stable at the desired pH and temperature. Although literature attention to this topic waned recently, it remains a critical factor in protein engineering. Belien et al. [25] used the pKD software to improve the low-pH stability of the B. subtilis endo-β-1,4-xylanase by making mutations that affected the local pKa of key residues. Heinzelman et al. [26] used SCHEMA to recombine several
Engineering proteins for molecular interactions
Computational protein design for a given function relies on optimizing a complex choreography of interactions with other molecules. A significant number of recent studies have focused on engineering these inter-molecule contacts. An important class of protein interaction partners is in fact other proteins. Tuncbag et al. [31] developed a computational method to identify “hot-spot” residues that are most important in mediating protein–protein interactions. In a study aimed at redesigning the
Designing new proteins
By bringing to bear structure elucidation, stability safeguards, and molecular interaction descriptions, a number of efforts achieved de novo design of novel proteins. One particularly intriguing target is antibodies, because there are well-established rules governing their structures and their functions are limited to binding, not catalysis. RosettaAntibody [50] was recently developed for the homology modeling of antibody variable domains and SnugDock [51] can be used in conjunction to predict
Conclusions
Successful computational protein design depends on accurate structure modeling, ensuring protein stability, and optimizing inter-molecule interactions. Each of these major hurdles has received significant attention in the past two years and many de novo protein designs have been put forth as a result. However, the dream of efficiently, predictably and reliably computationally designing improved proteins remains beyond reach. Baker [57•] eloquently reviewed in detail many of the unresolved
References and recommended reading
Papers of particular interest, published within the period of review, have been highlighted as:
• of special interest
•• of outstanding interest
Acknowledgement
This work was funded by the National Science Foundation [CBET-0639962].
References (57)
- et al.
Computational design of protein–protein interactions
Curr Opin Chem Biol
(2004) - et al.
Sampling bottlenecks in de novo protein structure prediction
J Mol Biol
(2009) - et al.
Conformation dependence of backbone geometry in proteins
Structure
(2009) - et al.
Experimental library screening demonstrates the successful application of computational protein design to large structural ensembles
Proc Natl Acad Sci U S A
(2010) - et al.
Computational design of epitope-scaffolds allows induction of antibodies specific for a poorly immunogenic HIV vaccine epitope
Structure
(2010) - et al.
Rationally designed integrin beta3 mutants stabilized in the high affinity conformation
J Biol Chem
(2009) - et al.
Simultaneous prediction of protein folding and docking at high resolution
Proc Natl Acad Sci U S A
(2009) - et al.
A mixed QM/MM scoring function to predict protein–ligand binding affinity
J Chem Theory Comput
(2010) - et al.
Computational predictions of the mutant behavior of AraC
J Mol Biol
(2010) - et al.
Identification of structural mechanisms of HIV-1 protease specificity using computational peptide docking: implications for drug resistance
Structure
(2009)
Design of protein-interaction specificity gives selective bZIP-binding peptides
Nature
Computational design of receptor and sensor proteins with novel functions
Nature
Computational redesign of endonuclease DNA binding and cleavage specificity
Nature
De novo protein design: fully automated sequence selection
Science
Design of a novel globular protein fold with atomic-level accuracy
Science
De novo computational design of retro-aldol enzymes
Science
Kemp elimination catalysts by computational enzyme design
Nature
Enhanced bounding techniques to reduce the protein conformational search space
Optim Methods Softw
Identifying and reducing error in cluster-expansion approximations of protein energies
J Comput Chem
Motif-directed flexible backbone design of functional interactions
Protein Sci
Improved prediction of protein side-chain conformations with SCWRL4
Proteins
Development of a rotamer library for use in beta-peptide foldamer computational design
J Am Chem Soc
Feature space resampling for protein conformational search
Proteins
PONDR-FIT: a meta-predictor of intrinsically disordered amino acids
Biochim Biophys Acta
An improved hybrid global optimization method for protein tertiary structure prediction
Comput Optim Appl
An efficient algorithm for multistate protein design based on FASTER
J Comput Chem
Selecting high quality protein structures from diverse conformational ensembles
Biophys J
Computation of conformational coupling in allosteric proteins
PLoS Comput Biol
Cited by (76)
Improving the Accuracy of Protein Thermostability Predictions for Single Point Mutations
2020, Biophysical JournalProtein sequence design and its applications
2016, Current Opinion in Structural BiologyCitation Excerpt :The design of proteins that can fold stably and function as anticipated, tests our understanding of the sequence–structure–function paradigm. It provides ground to address problems such as synthesis of natural mimics, re-purposing of an enzyme active site to accommodate alternate substrates, to filling-in as missing links to relate highly diverged proteins [2–5]. The tie-up between protein sequence and structure, assembly and ultimately function, suggests that tampering with the sequence could affect function and/or structure [6,7], although, contradictorily, structure is far more conserved than sequence [8,9].