Experimental library screening demonstrates the successful application of computational protein design to large structural ensembles

Robust genome and cell engineering via in vitro and in situ circularized RNAs

Circularization can improve RNA persistence, yet simple and scalable approaches to achieve this are lacking. Here we report two methods that facilitate the pursuit of circular RNAs (cRNAs): cRNAs developed via in vitro circularization using group II introns, and cRNAs developed via in-cell circularization by the ubiquitously expressed RtcB protein. We also report simple purification protocols that enable high cRNA yields (40-75%) while maintaining low immune responses. These methods and protocols facilitate a broad range of applications in stem cell engineering as well as robust genome and epigenome targeting via zinc finger proteins and CRISPR-Cas9. Notably, cRNAs bearing the encephalomyocarditis internal ribosome entry enabled robust expression and persistence compared with linear capped RNAs in cardiomyocytes and neurons, which highlights the utility of cRNAs in these non-dividing cells. We also describe genome targeting via deimmunized Cas9 delivered as cRNA and a long-range multiplexed protein engineering methodology for the combinatorial screening of deimmunized protein variants that enables compatibility between persistence of expression and immunogenicity in cRNA-delivered proteins. The cRNA toolset will aid research and the development of therapeutics.

Design of Efficient Artificial Enzymes Using Crystallographically Enhanced Conformational Sampling

The ability to create efficient artificial enzymes for any chemical reaction is of great interest. Here, we describe a computational design method for increasing the catalytic efficiency of de novo enzymes by several orders of magnitude without relying on directed evolution and high-throughput screening. Using structural ensembles generated from dynamics-based refinement against X-ray diffraction data collected from crystals of Kemp eliminases HG3 (kcat/KM 125 M-1 s-1) and KE70 (kcat/KM 57 M-1 s-1), we design from each enzyme ≤10 sequences predicted to catalyze this reaction more efficiently. The most active designs display kcat/KM values improved by 100-250-fold, comparable to mutants obtained after screening thousands of variants in multiple rounds of directed evolution. Crystal structures show excellent agreement with computational models, with catalytic contacts present as designed and transition-state root-mean-square deviations of ≤0.65 Å. Our work shows how ensemble-based design can generate efficient artificial enzymes by exploiting the true conformational ensemble to design improved active sites.

Design of efficient artificial enzymes using crystallographically-enhanced conformational sampling

The ability to create efficient artificial enzymes for any chemical reaction is of great interest. Here, we describe a computational design method for increasing catalytic efficiency of de novo enzymes to a level comparable to their natural counterparts without relying on directed evolution. Using structural ensembles generated from dynamics-based refinement against X-ray diffraction data collected from crystals of Kemp eliminases HG3 (kcat/KM 125 M-1 s-1) and KE70 (kcat/KM 57 M-1 s-1), we design from each enzyme ≤10 sequences predicted to catalyze this reaction more efficiently. The most active designs display kcat/KM values improved by 100-250-fold, comparable to mutants obtained after screening thousands of variants in multiple rounds of directed evolution. Crystal structures show excellent agreement with computational models. Our work shows how computational design can generate efficient artificial enzymes by exploiting the true conformational ensemble to more effectively stabilize the transition state.

Computational remodeling of an enzyme conformational landscape for altered substrate selectivity

Structural plasticity of enzymes dictates their function. Yet, our ability to rationally remodel enzyme conformational landscapes to tailor catalytic properties remains limited. Here, we report a computational procedure for tuning conformational landscapes that is based on multistate design of hinge-mediated domain motions. Using this method, we redesign the conformational landscape of a natural aminotransferase to preferentially stabilize a less populated but reactive conformation and thereby increase catalytic efficiency with a non-native substrate, resulting in altered substrate selectivity. Steady-state kinetics of designed variants reveals activity increases with the non-native substrate of approximately 100-fold and selectivity switches of up to 1900-fold. Structural analyses by room-temperature X-ray crystallography and multitemperature nuclear magnetic resonance spectroscopy confirm that conformational equilibria favor the target conformation. Our computational approach opens the door to targeted alterations of conformational states and equilibria, which should facilitate the design of biocatalysts with customized activity and selectivity.

Benchmarking TriadAb using targets from the second antibody modeling assessment

Computational modeling and design of antibodies has become an integral part of today's research and development in antibody therapeutics. Here we describe the Triad Antibody Homology Modeling (TriadAb) package, a functionality of the Triad protein design platform that predicts the structure of any heavy and light chain sequences of an antibody Fv domain using template-based modeling. To gauge the performance of TriadAb, we benchmarked against the results of the Second Antibody Modeling Assessment (AMA-II). On average, TriadAb produced main-chain carbonyl root-mean-square deviations between models and experimentally determined structures at 1.10 Å, 1.45 Å, 1.41 Å, 3.04 Å, 1.47 Å, 1.27 Å, 1.63 Å in the framework and the six complementarity-determining regions (H1, H2, H3, L1, L2, L3), respectively. The inaugural results are comparable to those reported in AMA-II, corroborating with our internal bench-based experiences that models generated using TriadAb are sufficiently accurate and useful for antibody engineering using the sequence design capabilities provided by Triad.

Structural effects of HIV-1 subtype C integrase mutations on the activity of integrase strand transfer inhibitors in South African patients

HIV-1 integrase enzyme is responsible for the integration of viral DNA into the host genomic DNA. Integrase strand transfer inhibitors (INSTIs) are highly potent antiretroviral agents that inhibit this process, and are internationally approved for the treatment of both naïve and treated HIV-1 patients. However, their long-term efficacy is threatened by development of drug resistance strains resulting in resistance mutations. This work aimed to examine the effect of INSTI resistance-associated mutations (RAMs) and polymorphisms on the structure of HIV-1 subtype C (HIV-1C) integrase. Genetic analysis was performed on seven HIV-1C infected individuals with virologic failure after at least 6 months of INSTI-based antiretroviral therapy, presenting at the King Edward VIII hospital in Durban, South Africa. These were compared with sequences from 41 INSTI-naïve isolates. Integrase structures of selected isolates were modeled on the SWISS model online server. Molecular docking and dynamics simulations were also conducted using AutoDock-Vina and AMBER 18 force fields, respectively. Only one INSTI-treated isolate (14.28%) harboured major mutations (G140A + Q148R) as well as the E157Q minor mutation. Interestingly, S119T and V151I were only found in patients failing raltegravir (an INSTI drug). Molecular modeling and docking showed that RAMs and polymorphisms associated with INSTI-based therapy affect protein stability and this is supported by their weakened hydrogen-bond interactions compared to the wild-type. To the best of our knowledge, this is the first study to identify a double mutant in the 140's loop region from South African HIV-1C isolates and study its effects on Raltegravir, Elvitegravir, and Dolutegravir binding.Communicated by Ramaswamy H. Sarma.

Recent advances in user-friendly computational tools to engineer protein function

Progress in technology and algorithms throughout the past decade has transformed the field of protein design and engineering. Computational approaches have become well-engrained in the processes of tailoring proteins for various biotechnological applications. Many tools and methods are developed and upgraded each year to satisfy the increasing demands and challenges of protein engineering. To help protein engineers and bioinformaticians navigate this emerging wave of dedicated software, we have critically evaluated recent additions to the toolbox regarding their application for semi-rational and rational protein engineering. These newly developed tools identify and prioritize hotspots and analyze the effects of mutations for a variety of properties, comprising ligand binding, protein-protein and protein-nucleic acid interactions, and electrostatic potential. We also discuss notable progress to target elusive protein dynamics and associated properties like ligand-transport processes and allosteric communication. Finally, we discuss several challenges these tools face and provide our perspectives on the further development of readily applicable methods to guide protein engineering efforts.

Revolutionizing enzyme engineering through artificial intelligence and machine learning

The combinatorial space of an enzyme sequence has astronomical possibilities and exploring it with contemporary experimental techniques is arduous and often ineffective. Multi-target objectives such as concomitantly achieving improved selectivity, solubility and activity of an enzyme have narrow plausibility under approaches of restricted mutagenesis and combinatorial search. Traditional enzyme engineering approaches have a limited scope for complex optimization due to the requirement of a priori knowledge or experimental burden of screening huge protein libraries. The recent surge in high-throughput experimental methods including Next Generation Sequencing and automated screening has flooded the field of molecular biology with big-data, which requires us to re-think our concurrent approaches towards enzyme engineering. Artificial Intelligence (AI) and Machine Learning (ML) have great potential to revolutionize smart enzyme engineering without the explicit need for a complete understanding of the underlying molecular system. Here, we portray the role and position of AI techniques in the field of enzyme engineering along with their scope and limitations. In addition, we explain how the traditional approaches of directed evolution and rational design can be extended through AI tools. Recent successful examples of AI-assisted enzyme engineering projects and their deviation from traditional approaches are highlighted. A comprehensive picture of current challenges and future avenues for AI in enzyme engineering are also discussed.

Attempts to develop an enzyme converting DHIV to KIV

Dihydroxy-acid dehydratase (DHAD) catalyzes the dehydration of R-2,3-dihydroxyisovalerate (DHIV) to 2-ketoisovalerate (KIV) using an Fe-S cluster as a cofactor, which is sensitive to oxidation and expensive to synthesize. In contrast, sugar acid dehydratases catalyze the same chemical reactions using a magnesium ion. Here, we attempted to substitute the high-cost DHAD with a cost-efficient engineered sugar acid dehydratase using computational protein design (CPD). First, we tried without success to modify the binding pocket of a sugar acid dehydratase to accommodate the smaller, more hydrophobic DHIV. Then, we used a chemically activated substrate analog to react with sugar acid dehydratases or other enolase superfamily enzymes. Mandelate racemase from Pseudomonas putida (PpManR) and the putative sugar acid dehydratase from Salmonella typhimurium (StPutD) showed beta-elimination activity towards chlorolactate (CLD). CPD combined with medium-throughput selection improved the PpManR kcat/KM for CLD by four-fold. However, these enzyme variants did not show dehydration activity towards DHIV. Lastly, assuming phosphorylation could also be a good activation mechanism, we found that mevalonate-3-kinase (M3K) from Picrophilus torridus (PtM3K) exhibited adenosine triphosphate (ATP) hydrolysis activity when mixed with DHIV, indicating phosphorylation activity towards DHIV. Engineering PpManR or StPutD to accept 3-phospho-DHIV as a substrate was performed, but no variants with the desired activity were obtained.

To Improve Protein Sequence Profile Prediction through Image Captioning on Pairwise Residue Distance Map

Protein sequence profile prediction aims to generate multiple sequences from structural information to advance the protein design. Protein sequence profile can be computationally predicted by energy-based or fragment-based methods. By integrating these methods with neural networks, our previous method, SPIN2, has achieved a sequence recovery rate of 34%. However, SPIN2 employed only one-dimensional (1D) structural properties that are not sufficient to represent three-dimensional (3D) structures. In this study, we represented 3D structures by 2D maps of pairwise residue distances and developed a new method (SPROF) to predict protein sequence profiles based on an image captioning learning frame. To our best knowledge, this is the first method to employ a 2D distance map for predicting protein properties. SPROF achieved 39.8% in sequence recovery of residues on the independent test set, representing a 5.2% improvement over SPIN2. We also found the sequence recovery increased with the number of their neighbored residues in 3D structural space, indicating that our method can effectively learn long-range information from the 2D distance map. Thus, such network architecture using a 2D distance map is expected to be useful for other 3D structure-based applications, such as binding site prediction, protein function prediction, and protein interaction prediction. The online server and the source code is available at http://biomed.nscc-gz.cn and https://github.com/biomed-AI/SPROF , respectively.

Protein stability engineering insights revealed by domain-wide comprehensive mutagenesis

The accurate prediction of protein stability upon sequence mutation is an important but unsolved challenge in protein engineering. Large mutational datasets are required to train computational predictors, but traditional methods for collecting stability data are either low-throughput or measure protein stability indirectly. Here, we develop an automated method to generate thermodynamic stability data for nearly every single mutant in a small 56-residue protein. Analysis reveals that most single mutants have a neutral effect on stability, mutational sensitivity is largely governed by residue burial, and unexpectedly, hydrophobics are the best tolerated amino acid type. Correlating the output of various stability-prediction algorithms against our data shows that nearly all perform better on boundary and surface positions than for those in the core and are better at predicting large-to-small mutations than small-to-large ones. We show that the most stable variants in the single-mutant landscape are better identified using combinations of 2 prediction algorithms and including more algorithms can provide diminishing returns. In most cases, poor in silico predictions were tied to compositional differences between the data being analyzed and the datasets used to train the algorithm. Finally, we find that strategies to extract stabilities from high-throughput fitness data such as deep mutational scanning are promising and that data produced by these methods may be applicable toward training future stability-prediction tools.

Positive multistate protein design

Motivation

Structure-based computational protein design (CPD) plays a critical role in advancing the field of protein engineering. Using an all-atom energy function, CPD tries to identify amino acid sequences that fold into a target structure and ultimately perform a desired function. The usual approach considers a single rigid backbone as a target, which ignores backbone flexibility. Multistate design (MSD) allows instead to consider several backbone states simultaneously, defining challenging computational problems.

Results

We introduce efficient reductions of positive MSD problems to Cost Function Networks with two different fitness definitions and implement them in the Pompd (Positive Multistate Protein design) software. Pompd is able to identify guaranteed optimal sequences of positive multistate full protein redesign problems and exhaustively enumerate suboptimal sequences close to the MSD optimum. Applied to nuclear magnetic resonance and back-rubbed X-ray structures, we observe that the average energy fitness provides the best sequence recovery. Our method outperforms state-of-the-art guaranteed computational design approaches by orders of magnitudes and can solve MSD problems with sizes previously unreachable with guaranteed algorithms.

Availability and implementation

https://forgemia.inra.fr/thomas.schiex/pompd as documented Open Source.

Supplementary information

Supplementary data are available at Bioinformatics online.

Integrating linear optimization with structural modeling to increase HIV neutralization breadth

Computational protein design has been successful in modeling fixed backbone proteins in a single conformation. However, when modeling large ensembles of flexible proteins, current methods in protein design have been insufficient. Large barriers in the energy landscape are difficult to traverse while redesigning a protein sequence, and as a result current design methods only sample a fraction of available sequence space. We propose a new computational approach that combines traditional structure-based modeling using the Rosetta software suite with machine learning and integer linear programming to overcome limitations in the Rosetta sampling methods. We demonstrate the effectiveness of this method, which we call BROAD, by benchmarking the performance on increasing predicted breadth of anti-HIV antibodies. We use this novel method to increase predicted breadth of naturally-occurring antibody VRC23 against a panel of 180 divergent HIV viral strains and achieve 100% predicted binding against the panel. In addition, we compare the performance of this method to state-of-the-art multistate design in Rosetta and show that we can outperform the existing method significantly. We further demonstrate that sequences recovered by this method recover known binding motifs of broadly neutralizing anti-HIV antibodies. Finally, our approach is general and can be extended easily to other protein systems. Although our modeled antibodies were not tested in vitro, we predict that these variants would have greatly increased breadth compared to the wild-type antibody.

In this article, we report a new approach for protein design, which combines traditional structural modeling with machine learning and integer programming. Using this method, we are able to design antibodies that are predicted to bind large panels of antigenically diverse HIV variants. The combination of methods from these fields allows us to surpass protein design limitations that have been seen up to this point. We predict that if we tested these modified antibodies against HIV variants they would have greater neutralization breadth than any antibodies seen to this point.

Enriching Peptide Libraries for Binding Affinity and Specificity Through Computationally Directed Library Design

Peptide reagents with high affinity or specificity for their target protein interaction partner are of utility for many important applications. Optimization of peptide binding by screening large libraries is a proven and powerful approach. Libraries designed to be enriched in peptide sequences that are predicted to have desired affinity or specificity characteristics are more likely to yield success than random mutagenesis. We present a library optimization method in which the choice of amino acids to encode at each peptide position can be guided by available experimental data or structure-based predictions. We discuss how to use analysis of predicted library performance to inform rounds of library design. Finally, we include protocols for more complex library design procedures that consider the chemical diversity of the amino acids at each peptide position and optimize a library score based on a user-specified input model.

Nature-Inspired Construction of Two-Dimensionally Self-Assembled Peptide on Pristine Graphene

Peptide assemblies have received significant attention because of their important role in biology and applications in bionanotechnology. Despite recent efforts to elucidate the principles of peptide self-assembly for developing novel functional devices, peptide self-assembly on two-dimensional nanomaterials has remained challenging. Here, we report nature-inspired two-dimensional peptide self-assembly on pristine graphene via optimization of peptide-peptide and peptide-graphene interactions. Two-dimensional peptide self-assembly was designed based on statistical analyses of >10⁴ protein structures existing in nature and atomistic simulation-based structure predictions. We characterized the structures and surface properties of the self-assembled peptide formed on pristine graphene. Our study provides insights into the formation of peptide assemblies coupled with two-dimensional nanomaterials for further development of nanobiocomposite devices.

Rosetta:MSF: a modular framework for multi-state computational protein design

Computational protein design (CPD) is a powerful technique to engineer existing proteins or to design novel ones that display desired properties. Rosetta is a software suite including algorithms for computational modeling and analysis of protein structures and offers many elaborate protocols created to solve highly specific tasks of protein engineering. Most of Rosetta’s protocols optimize sequences based on a single conformation (i. e. design state). However, challenging CPD objectives like multi-specificity design or the concurrent consideration of positive and negative design goals demand the simultaneous assessment of multiple states. This is why we have developed the multi-state framework MSF that facilitates the implementation of Rosetta’s single-state protocols in a multi-state environment and made available two frequently used protocols. Utilizing MSF, we demonstrated for one of these protocols that multi-state design yields a 15% higher performance than single-state design on a ligand-binding benchmark consisting of structural conformations. With this protocol, we designed de novo nine retro-aldolases on a conformational ensemble deduced from a (βα)₈-barrel protein. All variants displayed measurable catalytic activity, testifying to a high success rate for this concept of multi-state enzyme design.

Protein engineering, i. e. the targeted modification or design of proteins has tremendous potential for medical and industrial applications. One generally applicable strategy for protein engineering is rational protein design: based on detailed knowledge of structure and function, computer programs like Rosetta propose the sequence of a protein possessing the desired properties. So far, most computer protocols have used rigid structures for design, which is a simplification because a protein’s structure is more accurately specified by a conformational ensemble. We have now implemented a framework for computational protein design that allows certain design protocols of Rosetta to make use of multiple design states like structural ensembles. An in silico assessment simulating ligand-binding design showed that this new approach generates more reliably native-like sequences than a single-state approach. As a proof-of-concept, we introduced de novo retro-aldolase activity into a scaffold protein and characterized nine variants experimentally, all of which were catalytically active.

Multistate Computational Protein Design with Backbone Ensembles

The ability of computational protein design (CPD) to identify protein sequences possessing desired characteristics in vast sequence spaces makes it a highly valuable tool in the protein engineering toolbox. CPD calculations are typically performed using a single-state design (SSD) approach in which amino-acid sequences are optimized on a single protein structure. Although SSD has been successfully applied to the design of numerous protein functions and folds, the approach can lead to the incorrect rejection of desirable sequences because of the combined use of a fixed protein backbone template and a set of rigid rotamers. This fixed backbone approximation can be addressed by using multistate design (MSD) with backbone ensembles. MSD improves the quality of predicted sequences by using ensembles approximating conformational flexibility as input templates instead of a single fixed protein structure. In this chapter, we present a step-by-step guide to the implementation and analysis of MSD calculations with backbone ensembles. Specifically, we describe ensemble generation with the PertMin protocol, execution of MSD calculations for recapitulation of Streptococcal protein G domain β1 mutant stability, and analysis of computational predictions by sequence binning. Furthermore, we provide a comparison between MSD and SSD calculation results and discuss the benefits of multistate approaches to CPD.

Achievements and Challenges in Computational Protein Design

Computational protein design (CPD), a yet evolving field, includes computer-aided engineering for partial or full de novo designs of proteins of interest. Designs are defined by a requested structure, function, or working environment. This chapter describes the birth and maturation of the field by presenting 101 CPD examples in a chronological order emphasizing achievements and pending challenges. Integrating these aspects presents the plethora of CPD approaches with the hope of providing a "CPD 101". These reflect on the broader structural bioinformatics and computational biophysics field and include: (1) integration of knowledge-based and energy-based methods, (2) hierarchical designated approach towards local, regional, and global motifs and the integration of high- and low-resolution design schemes that fit each such region, (3) systematic differential approaches towards different protein regions, (4) identification of key hot-spot residues and the relative effect of remote regions, (5) assessment of shape-complementarity, electrostatics and solvation effects, (6) integration of thermal plasticity and functional dynamics, (7) negative design, (8) systematic integration of experimental approaches, (9) objective cross-assessment of methods, and (10) successful ranking of potential designs. Future challenges also include dissemination of CPD software to the general use of life-sciences researchers and the emphasis of success within an in vivo milieu. CPD increases our understanding of protein structure and function and the relationships between the two along with the application of such know-how for the benefit of mankind. Applied aspects range from biological drugs, via healthier and tastier food products to nanotechnology and environmentally friendly enzymes replacing toxic chemicals utilized in the industry.

Computationally Designed Bispecific Antibodies using Negative State Repertoires

A challenge in the structure-based design of specificity is modeling the negative states, i.e., the complexes that you do not want to form. This is a difficult problem because mutations predicted to destabilize the negative state might be accommodated by small conformational rearrangements. To overcome this challenge, we employ an iterative strategy that cycles between sequence design and protein docking in order to build up an ensemble of alternative negative state conformations for use in specificity prediction. We have applied our technique to the design of heterodimeric CH3 interfaces in the Fc region of antibodies. Combining computationally and rationally designed mutations produced unique designs with heterodimer purities greater than 90%. Asymmetric Fc crystallization was able to resolve the interface mutations; the heterodimer structures confirmed that the interfaces formed as designed. With these CH3 mutations, and those made at the heavy-/light-chain interface, we demonstrate one-step synthesis of four fully IgG-bispecific antibodies.

Single-Point Mutation with a Rotamer Library Toolkit: Toward Protein Engineering

Protein engineers have long been hard at work to harness biocatalysts as a natural source of regio-, stereo-, and chemoselectivity in order to carry out chemistry (reactions and/or substrates) not previously achieved with these enzymes. The extreme labor demands and exponential number of mutation combinations have induced computational advances in this domain. The first step in our virtual approach is to predict the correct conformations upon mutation of residues (i.e., rebuilding side chains). For this purpose, we opted for a combination of molecular mechanics and statistical data. In this work, we have developed automated computational tools to extract protein structural information and created conformational libraries for each amino acid dependent on a variable number of parameters (e.g., resolution, flexibility, secondary structure). We have also developed the necessary tool to apply the mutation and optimize the conformation accordingly. For side-chain conformation prediction, we obtained overall average root-mean-square deviations (RMSDs) of 0.91 and 1.01 Å for the 18 flexible natural amino acids within two distinct sets of over 3000 and 1500 side-chain residues, respectively. The commonly used dihedral angle differences were also evaluated and performed worse than the state of the art. These two metrics are also compared. Furthermore, we generated a family-specific library for kinases that produced an average 2% lower RMSD upon side-chain reconstruction and a residue-specific library that yielded a 17% improvement. Ultimately, since our protein engineering outlook involves using our docking software, Fitted/Impacts, we applied our mutation protocol to a benchmarked data set for self- and cross-docking. Our side-chain reconstruction does not hinder our docking software, demonstrating differences in pose prediction accuracy of approximately 2% (RMSD cutoff metric) for a set of over 200 protein/ligand structures. Similarly, when docking to a set of over 100 kinases, side-chain reconstruction (using both general and biased conformation libraries) had minimal detriment to the docking accuracy.

Design of Protein Multi-specificity Using an Independent Sequence Search Reduces the Barrier to Low Energy Sequences

Computational protein design has found great success in engineering proteins for thermodynamic stability, binding specificity, or enzymatic activity in a 'single state' design (SSD) paradigm. Multi-specificity design (MSD), on the other hand, involves considering the stability of multiple protein states simultaneously. We have developed a novel MSD algorithm, which we refer to as REstrained CONvergence in multi-specificity design (RECON). The algorithm allows each state to adopt its own sequence throughout the design process rather than enforcing a single sequence on all states. Convergence to a single sequence is encouraged through an incrementally increasing convergence restraint for corresponding positions. Compared to MSD algorithms that enforce (constrain) an identical sequence on all states the energy landscape is simplified, which accelerates the search drastically. As a result, RECON can readily be used in simulations with a flexible protein backbone. We have benchmarked RECON on two design tasks. First, we designed antibodies derived from a common germline gene against their diverse targets to assess recovery of the germline, polyspecific sequence. Second, we design "promiscuous", polyspecific proteins against all binding partners and measure recovery of the native sequence. We show that RECON is able to efficiently recover native-like, biologically relevant sequences in this diverse set of protein complexes.

Using Molecular Dynamics Simulations as an Aid in the Prediction of Domain Swapping of Computationally Designed Protein Variants

In standard implementations of computational protein design, a positive-design approach is used to predict sequences that will be stable on a given backbone structure. Possible competing states are typically not considered, primarily because appropriate structural models are not available. One potential competing state, the domain-swapped dimer, is especially compelling because it is often nearly identical with its monomeric counterpart, differing by just a few mutations in a hinge region. Molecular dynamics (MD) simulations provide a computational method to sample different conformational states of a structure. Here, we tested whether MD simulations could be used as a post-design screening tool to identify sequence mutations leading to domain-swapped dimers. We hypothesized that a successful computationally designed sequence would have backbone structure and dynamics characteristics similar to that of the input structure and that, in contrast, domain-swapped dimers would exhibit increased backbone flexibility and/or altered structure in the hinge-loop region to accommodate the large conformational change required for domain swapping. While attempting to engineer a homodimer from a 51-amino-acid fragment of the monomeric protein engrailed homeodomain (ENH), we had instead generated a domain-swapped dimer (ENH_DsD). MD simulations on these proteins showed increased B-factors derived from MD simulation in the hinge loop of the ENH_DsD domain-swapped dimer relative to monomeric ENH. Two point mutants of ENH_DsD designed to recover the monomeric fold were then tested with an MD simulation protocol. The MD simulations suggested that one of these mutants would adopt the target monomeric structure, which was subsequently confirmed by X-ray crystallography.

Discriminating between stabilizing and destabilizing protein design mutations via recombination and simulation

Accuracy of current computational protein design (CPD) methods is limited by inherent approximations in energy potentials and sampling. These limitations are often used to qualitatively explain design failures; however, relatively few studies provide specific examples or quantitative details that can be used to improve future CPD methods. Expanding the design method to include a library of sequences provides data that is well suited for discriminating between stabilizing and destabilizing design elements. Using thermophilic endoglucanase E1 from Acidothermus cellulolyticus as a model enzyme, we computationally designed a sequence with 60 mutations. The design sequence was rationally divided into structural blocks and recombined with the wild-type sequence. Resulting chimeras were assessed for activity and thermostability. Surprisingly, unlike previous chimera libraries, regression analysis based on one- and two-body effects was not sufficient for predicting chimera stability. Analysis of molecular dynamics simulations proved helpful in distinguishing stabilizing and destabilizing mutations. Reverting to the wild-type amino acid at destabilized sites partially regained design stability, and introducing predicted stabilizing mutations in wild-type E1 significantly enhanced thermostability. The ability to isolate stabilizing and destabilizing elements in computational design offers an opportunity to interpret previous design failures and improve future CPD methods.

Computational Design of the β-Sheet Surface of a Red Fluorescent Protein Allows Control of Protein Oligomerization

Computational design has been used with mixed success for the design of protein surfaces, with directed evolution heretofore providing better practical solutions than explicit design. Directed evolution, however, requires a tractable high-throughput screen because the random nature of mutation does not enrich for desired traits. Here we demonstrate the successful design of the β-sheet surface of a red fluorescent protein (RFP), enabling control over its oligomerization. To isolate the problem of surface design, we created a hybrid RFP from DsRed and mCherry with a stabilized protein core that allows for monomerization without loss of fluorescence. We designed an explicit library for which 93 of 96 (97%) of the protein variants are soluble, stably fluorescent, and monomeric. RFPs are heavily used in biology, but are natively tetrameric, and creating RFP monomers has proven extremely difficult. We show that surface design and core engineering are separate problems in RFP development and that the next generation of RFP markers will depend on improved methods for core design.

Methods for library-scale computational protein design

Faced with a protein engineering challenge, a contemporary researcher can choose from myriad design strategies. Library-scale computational protein design (LCPD) is a hybrid method suitable for the engineering of improved protein variants with diverse sequences. This chapter discusses the background and merits of several practical LCPD techniques. First, LCPD methods suitable for delocalized protein design are presented in the context of example design calculations for cellobiohydrolase II. Second, localized design methods are discussed in the context of an example design calculation intended to shift the substrate specificity of a ketol-acid reductoisomerase Rossmann domain from NADPH to NADH.

Optimization of rotamers prior to template minimization improves stability predictions made by computational protein design

Computational protein design (CPD) predictions are highly dependent on the structure of the input template used. However, it is unclear how small differences in template geometry translate to large differences in stability prediction accuracy. Herein, we explored how structural changes to the input template affect the outcome of stability predictions by CPD. To do this, we prepared alternate templates by Rotamer Optimization followed by energy Minimization (ROM) and used them to recapitulate the stability of 84 protein G domain β1 mutant sequences. In the ROM process, side-chain rotamers for wild-type (WT) or mutant sequences are optimized on crystal or nuclear magnetic resonance (NMR) structures prior to template minimization, resulting in alternate structures termed ROM templates. We show that use of ROM templates prepared from sequences known to be stable results predominantly in improved prediction accuracy compared to using the minimized crystal or NMR structures. Conversely, ROM templates prepared from sequences that are less stable than the WT reduce prediction accuracy by increasing the number of false positives. These observed changes in prediction outcomes are attributed to differences in side-chain contacts made by rotamers in ROM templates. Finally, we show that ROM templates prepared from sequences that are unfolded or that adopt a nonnative fold result in the selective enrichment of sequences that are also unfolded or that adopt a nonnative fold, respectively. Our results demonstrate the existence of a rotamer bias caused by the input template that can be harnessed to skew predictions toward sequences displaying desired characteristics.

SwiftLib: rapid degenerate-codon-library optimization through dynamic programming

Degenerate codon (DC) libraries efficiently address the experimental library-size limitations of directed evolution by focusing diversity toward the positions and toward the amino acids (AAs) that are most likely to generate hits; however, manually constructing DC libraries is challenging, error prone and time consuming. This paper provides a dynamic programming solution to the task of finding the best DCs while keeping the size of the library beneath some given limit, improving on the existing integer-linear programming formulation. It then extends the algorithm to consider multiple DCs at each position, a heretofore unsolved problem, while adhering to a constraint on the number of primers needed to synthesize the library. In the two library-design problems examined here, the use of multiple DCs produces libraries that very nearly cover the set of desired AAs while still staying within the experimental size limits. Surprisingly, the algorithm is able to find near-perfect libraries where the ratio of amino-acid sequences to nucleic-acid sequences approaches 1; it effectively side-steps the degeneracy of the genetic code. Our algorithm is freely available through our web server and solves most design problems in about a second.

Discovery of substrates for a SET domain lysine methyltransferase predicted by multistate computational protein design

Characterization of lysine methylation has proven challenging despite its importance in biological processes such as gene transcription, protein turnover, and cytoskeletal organization. In contrast to other key posttranslational modifications, current proteomics techniques have thus far shown limited success at characterizing methyl-lysine residues across the cellular landscape. To complement current biochemical characterization methods, we developed a multistate computational protein design procedure to probe the substrate specificity of the protein lysine methyltransferase SMYD2. Modeling of substrate-bound SMYD2 identified residues important for substrate recognition and predicted amino acids necessary for methylation. Peptide- and protein- based substrate libraries confirmed that SMYD2 activity is dictated by the motif [LFM]-1-K(∗)-[AFYMSHRK]+1-[LYK]+2 around the target lysine K(∗). Comprehensive motif-based searches and mutational analysis further established four additional substrates of SMYD2. Our methodology paves the way to systematically predict and validate posttranslational modification sites while simultaneously pairing them with their associated enzymes.

Direct prediction of profiles of sequences compatible with a protein structure by neural networks with fragment-based local and energy-based nonlocal profiles

Locating sequences compatible with a protein structural fold is the well-known inverse protein-folding problem. While significant progress has been made, the success rate of protein design remains low. As a result, a library of designed sequences or profile of sequences is currently employed for guiding experimental screening or directed evolution. Sequence profiles can be computationally predicted by iterative mutations of a random sequence to produce energy-optimized sequences, or by combining sequences of structurally similar fragments in a template library. The latter approach is computationally more efficient but yields less accurate profiles than the former because of lacking tertiary structural information. Here we present a method called SPIN that predicts Sequence Profiles by Integrated Neural network based on fragment-derived sequence profiles and structure-derived energy profiles. SPIN improves over the fragment-derived profile by 6.7% (from 23.6 to 30.3%) in sequence identity between predicted and wild-type sequences. The method also reduces the number of residues in low complex regions by 15.7% and has a significantly better balance of hydrophilic and hydrophobic residues at protein surface. The accuracy of sequence profiles obtained is comparable to those generated from the protein design program RosettaDesign 3.5. This highly efficient method for predicting sequence profiles from structures will be useful as a single-body scoring term for improving scoring functions used in protein design and fold recognition. It also complements protein design programs in guiding experimental design of the sequence library for screening and directed evolution of designed sequences. The SPIN server is available at http://sparks-lab.org.

Structure-based redesign of the binding specificity of anti-apoptotic Bcl-x(L)

Many native proteins are multi-specific and interact with numerous partners, which can confound analysis of their functions. Protein design provides a potential route to generating synthetic variants of native proteins with more selective binding profiles. Redesigned proteins could be used as research tools, diagnostics or therapeutics. In this work, we used a library screening approach to reengineer the multi-specific anti-apoptotic protein Bcl-x(L) to remove its interactions with many of its binding partners, making it a high-affinity and selective binder of the BH3 region of pro-apoptotic protein Bad. To overcome the enormity of the potential Bcl-x(L) sequence space, we developed and applied a computational/experimental framework that used protein structure information to generate focused combinatorial libraries. Sequence features were identified using structure-based modeling, and an optimization algorithm based on integer programming was used to select degenerate codons that maximally covered these features. A constraint on library size was used to ensure thorough sampling. Using yeast surface display to screen a designed library of Bcl-x(L) variants, we successfully identified a protein with ~1000-fold improvement in binding specificity for the BH3 region of Bad over the BH3 region of Bim. Although negative design was targeted only against the BH3 region of Bim, the best redesigned protein was globally specific against binding to 10 other peptides corresponding to native BH3 motifs. Our design framework demonstrates an efficient route to highly specific protein binders and may readily be adapted for application to other design problems.

Computational Enzyme Design: Advances, hurdles and possible ways forward

This mini review addresses recent developments in computational enzyme design. Successful protocols as well as known issues and limitations are discussed from an energetic perspective. It will be argued that improved results can be obtained by including a dynamic treatment in the design protocol. Finally, a molecular dynamics-based approach for evaluating and refining computational designs is presented.

Recent advances in rational approaches for enzyme engineering

Enzymes are an attractive alternative in the asymmetric syntheses of chiral building blocks. To meet the requirements of industrial biotechnology and to introduce new functionalities, the enzymes need to be optimized by protein engineering. This article specifically reviews rational approaches for enzyme engineering and de novo enzyme design involving structure-based approaches developed in recent years for improvement of the enzymes’ performance, broadened substrate range, and creation of novel functionalities to obtain products with high added value for industrial applications.

Multistate approaches in computational protein design

Computational protein design (CPD) is a useful tool for protein engineers. It has been successfully applied towards the creation of proteins with increased thermostability, improved binding affinity, novel enzymatic activity, and altered ligand specificity. Traditionally, CPD calculations search and rank sequences using a single fixed protein backbone template in an approach referred to as single-state design (SSD). While SSD has enjoyed considerable success, certain design objectives require the explicit consideration of multiple conformational and/or chemical states. Cases where a "multistate" approach may be advantageous over the SSD approach include designing conformational changes into proteins, using native ensembles to mimic backbone flexibility, and designing ligand or oligomeric association specificities. These design objectives can be efficiently tackled using multistate design (MSD), an emerging methodology in CPD that considers any number of protein conformational or chemical states as inputs instead of a single protein backbone template, as in SSD. In this review article, recent examples of the successful design of a desired property into proteins using MSD are described. These studies employing MSD are divided into two categories--those that utilized multiple conformational states, and those that utilized multiple chemical states. In addition, the scoring of competing states during negative design is discussed as a current challenge for MSD.

Designing specific protein-protein interactions using computation, experimental library screening, or integrated methods

Given the importance of protein-protein interactions for nearly all biological processes, the design of protein affinity reagents for use in research, diagnosis or therapy is an important endeavor. Engineered proteins would ideally have high specificities for their intended targets, but achieving interaction specificity by design can be challenging. There are two major approaches to protein design or redesign. Most commonly, proteins and peptides are engineered using experimental library screening and/or in vitro evolution. An alternative approach involves using protein structure and computational modeling to rationally choose sequences predicted to have desirable properties. Computational design has successfully produced novel proteins with enhanced stability, desired interactions and enzymatic function. Here we review the strengths and limitations of experimental library screening and computational structure-based design, giving examples where these methods have been applied to designing protein interaction specificity. We highlight recent studies that demonstrate strategies for combining computational modeling with library screening. The computational methods provide focused libraries predicted to be enriched in sequences with the properties of interest. Such integrated approaches represent a promising way to increase the efficiency of protein design and to engineer complex functionality such as interaction specificity.

Predictive Bcl-2 family binding models rooted in experiment or structure

Proteins of the Bcl-2 family either enhance or suppress programmed cell death and are centrally involved in cancer development and resistance to chemotherapy. BH3 (Bcl-2 homology 3)-only Bcl-2 proteins promote cell death by docking an α-helix into a hydrophobic groove on the surface of one or more of five pro-survival Bcl-2 receptor proteins. There is high structural homology within the pro-death and pro-survival families, yet a high degree of interaction specificity is nevertheless encoded, posing an interesting and important molecular recognition problem. Understanding protein features that dictate Bcl-2 interaction specificity is critical for designing peptide-based cancer therapeutics and diagnostics. In this study, we present peptide SPOT arrays and deep sequencing data from yeast display screening experiments that significantly expand the BH3 sequence space that has been experimentally tested for interaction with five human anti-apoptotic receptors. These data provide rich information about the determinants of Bcl-2 family specificity. To interpret and use the information, we constructed two simple data-based models that can predict affinity and specificity when evaluated on independent data sets within a limited sequence space. We also constructed a novel structure-based statistical potential, called STATIUM, which is remarkably good at predicting Bcl-2 affinity and specificity, especially considering it is not trained on experimental data. We compare the performance of our three models to each other and to alternative structure-based methods and discuss how such tools can guide prediction and design of new Bcl-2 family complexes.

Computational design of a PDZ domain peptide inhibitor that rescues CFTR activity

The cystic fibrosis transmembrane conductance regulator (CFTR) is an epithelial chloride channel mutated in patients with cystic fibrosis (CF). The most prevalent CFTR mutation, ΔF508, blocks folding in the endoplasmic reticulum. Recent work has shown that some ΔF508-CFTR channel activity can be recovered by pharmaceutical modulators (“potentiators” and “correctors”), but ΔF508-CFTR can still be rapidly degraded via a lysosomal pathway involving the CFTR-associated ligand (CAL), which binds CFTR via a PDZ interaction domain. We present a study that goes from theory, to new structure-based computational design algorithms, to computational predictions, to biochemical testing and ultimately to epithelial-cell validation of novel, effective CAL PDZ inhibitors (called “stabilizers”) that rescue ΔF508-CFTR activity. To design the “stabilizers”, we extended our structural ensemble-based computational protein redesign algorithm to encompass protein-protein and protein-peptide interactions. The computational predictions achieved high accuracy: all of the top-predicted peptide inhibitors bound well to CAL. Furthermore, when compared to state-of-the-art CAL inhibitors, our design methodology achieved higher affinity and increased binding efficiency. The designed inhibitor with the highest affinity for CAL (kCAL01) binds six-fold more tightly than the previous best hexamer (iCAL35), and 170-fold more tightly than the CFTR C-terminus. We show that kCAL01 has physiological activity and can rescue chloride efflux in CF patient-derived airway epithelial cells. Since stabilizers address a different cellular CF defect from potentiators and correctors, our inhibitors provide an additional therapeutic pathway that can be used in conjunction with current methods.

Cystic fibrosis (CF) is an inherited disease that causes the body to produce thick mucus that clogs the lungs and obstructs the breakdown and absorption of food. The cystic fibrosis transmembrane conductance regulator (CFTR) is mutated in CF patients, and the most common mutation causes three defects in CFTR: misfolding, decreased function, and rapid degradation. Drugs are currently being studied to correct the first two CFTR defects, but the problem of rapid degradation remains. Recently, key protein-protein interactions have been discovered that implicate the protein CAL in CFTR degradation. Here we have developed new computational protein design algorithms and used them to successfully predict peptide inhibitors of the CAL-CFTR interface. Our algorithm uses a structural ensemble-based evaluation of protein sequences and conformations to calculate accurate predictions of protein-peptide binding affinities. The algorithm is general and can be applied to a wide variety of protein-protein interface designs. All of our designed inhibitors bound CAL with high affinity. We tested our top binding peptide and observed that the inhibitor could successfully rescue CFTR function in CF patient-derived epithelial cells. Our designed inhibitors provide a novel therapeutic path which could be used in combination with existing CF therapeutics for additive benefit.

Protein stability by number: high-throughput and statistical approaches to one of protein science's most difficult problems

Most proteins are only barely stable, which impedes research, complicates therapeutic applications, and makes proteins susceptible to pathologically destabilizing mutations. Our ability to predict the thermodynamic consequences of even single point mutations is still surprisingly limited, and established methods of measuring stability are slow. Recent advances are bringing protein stability studies into the high-throughput realm. Some methods are based on inferential read-outs such as activity, proteolytic resistance or split-protein fragment reassembly. Other methods use miniaturization of direct measurements, such as intrinsic fluorescence, H/D exchange, cysteine reactivity, aggregation and hydrophobic dye binding (DSF). Protein engineering based on statistical analysis (consensus and correlated occurrences of amino acids) is promising, but much work remains to understand and implement these methods.

Computational design of protein-ligand interfaces: potential in therapeutic development

Computational design of protein-ligand interfaces finds optimal amino acid sequences within a small-molecule binding site of a protein for tight binding of a specific small molecule. It requires a search algorithm that can rapidly sample the vast sequence and conformational space, and a scoring function that can identify low energy designs. This review focuses on recent advances in computational design methods and their application to protein-small molecule binding sites. Strategies for increasing affinity, altering specificity, creating broad-spectrum binding, and building novel enzymes from scratch are described. Future prospects for applications in drug development are discussed, including limitations that will need to be overcome to achieve computational design of protein therapeutics with novel modes of action.

Generation of longer emission wavelength red fluorescent proteins using computationally designed libraries

The longer emission wavelengths of red fluorescent proteins (RFPs) make them attractive for whole-animal imaging because cells are more transparent to red light. Although several useful RFPs have been developed using directed evolution, the quest for further red-shifted and improved RFPs continues. Herein, we report a structure-based rational design approach to red-shift the fluorescence emission of RFPs. We applied a combined computational and experimental approach that uses computational protein design as an in silico prescreen to generate focused combinatorial libraries of mCherry mutants. The computational procedure helped us identify residues that could fulfill interactions hypothesized to cause red-shifts without destabilizing the protein fold. These interactions include stabilization of the excited state through H-bonding to the acylimine oxygen atom, destabilization of the ground state by hydrophobic packing around the charged phenolate, and stabilization of the excited state by a π-stacking interaction. Our methodology allowed us to identify three mCherry mutants (mRojoA, mRojoB, and mRouge) that display emission wavelengths > 630 nm, representing red-shifts of 20-26 nm. Moreover, our approach required the experimental screening of a total of ∼5,000 clones, a number several orders of magnitude smaller than those previously used to achieve comparable red-shifts. Additionally, crystal structures of mRojoA and mRouge allowed us to verify fulfillment of the interactions hypothesized to cause red-shifts, supporting their contribution to the observed red-shifts.