Analysis of conformational stability of interacting residues in protein binding interfaces.

After approximately 60 years of work, the protein folding problem has recently seen rapid advancement thanks to the inventions of AlphaFold and RoseTTAFold, which are machine-learning algorithms capable of reliably predicting protein structures from their sequences. A key component in their success was the inclusion of pairwise interaction information between residues. As research focus shifts towards developing algorithms to design and engineer binding proteins, it is likely that knowledge of interaction features at protein interfaces can improve predictions. Here, 574 protein complexes were analyzed to identify the stability features of their pairwise interactions, revealing that interactions between pre-stabilized residues are a selected feature in protein binding interfaces. In a retrospective analysis of 475 de novo designed binding proteins with an experimental success rate of 19%, inclusion of pairwise interaction pre-stabilization parameters increased the frequency of identifying experimentally successful binders to 40%.

Developing similarity matrices for antibody-protein binding interactions.

The inventions of AlphaFold and RoseTTAFold are revolutionizing computational protein science due to their abilities to reliably predict protein structures. Their unprecedented successes are due to the parallel consideration of several types of information, one of which is protein sequence similarity information. Sequence homology has been studied for many decades and depends on similarity matrices to define how similar or different protein sequences are to one another. A natural extension of predicting protein structures is predicting the interactions between proteins, but similarity matrices for protein-protein interactions do not exist. This study conducted a mutational analysis of 384 non-redundant antibody-protein antigen complexes to calculate antibody-protein interaction similarity matrices. Every important residue in each antibody and each antigen was mutated to each of the other 19 commonly occurring amino acids and the percentage changes in interaction energies were calculated using three force fields: CHARMM, Amber, and Rosetta. The data were used to construct six interaction similarity matrices, one for antibodies and another for antigens using each force field. The matrices exhibited both commonalities, such as mutations of aromatic and charged residues being the most detrimental, and differences, such as Rosetta predicting mutations of serines to be better tolerated than either Amber or CHARMM. A comparison to nine previously published similarity matrices for protein sequences revealed that the new interaction matrices are more similar to one another than they are to any of the previous matrices. The created similarity matrices can be used in force field specific applications to help guide decisions regarding mutations in protein-protein binding interfaces.

MutDock: A computational docking approach for fixed-backbone protein scaffold design.

Despite the successes of antibodies as therapeutic binding proteins, they still face production and design challenges. Alternative binding scaffolds of smaller size have been developed to overcome these issues. A subset of these alternative scaffolds recognizes target molecules through mutations to a set of surface resides, which does not alter their backbone structures. While the computational design of antibodies for target epitopes has been explored in depth, the same has not been done for alternative scaffolds. The commonly used dock-and-mutate approach for binding proteins, including antibodies, is limited because it uses a constant sequence and structure representation of the scaffold. Docking fixed-backbone scaffolds with a varied group of surface amino acids increases the chances of identifying superior starting poses that can be improved with subsequent mutations. In this work, we have developed MutDock, a novel computational approach that simultaneously docks and mutates fixed backbone scaffolds for binding a target epitope by identifying a minimum number of hydrogen bonds. The approach is broadly divided into two steps. The first step uses pairwise distance alignment of hydrogen bond-forming areas of scaffold residues and compatible epitope atoms. This step considers both native and mutated rotamers of scaffold residues. The second step mutates clashing variable interface residues and thermodynamically unfavorable residues to create additional strong interactions. MutDock was used to dock two scaffolds, namely, Affibodies and DARPins, with ten randomly selected antigens. The energies of the docked poses were minimized and binding energies were compared with docked poses from ZDOCK and HADDOCK. The top MutDock poses consisted of higher and comparable binding energies than the top ZDOCK and HADDOCK poses, respectively. This work contributes to the discovery of novel binders based on smaller-sized, fixed-backbone protein scaffolds.

Antibody epitope repertoire analysis enables rapid antigen discovery and multiplex serology.

The detection of pathogen-specific antibodies remains a cornerstone of clinical diagnostics. Yet, many test exhibit undesirable performance or are completely lacking. Given this, we developed serum epitope repertoire analysis (SERA), a method to rapidly discover conserved, pathogen-specific antigens and their epitopes, and applied it to develop an assay for Chagas disease caused by the protozoan parasite Trypanosoma cruzi. Antibody binding peptide motifs were identified from 28 Chagas repertoires using a bacterial display random 12-mer peptide library and next-generation sequencing (NGS). Thirty-three motifs were selected and mapped to candidate Chagas antigens. In a blinded validation set (n = 72), 30/30 Chagas were positive, 30/30 non-Chagas were negative, and 1/12 Leishmania sp. was positive. After unblinding, a Leishmania cross-reactive epitope was identified and removed from the panel. The Chagas assay exhibited 100% sensitivity (30/30) and specificity (90/90) in a second blinded validation set including individuals with other parasitic infections. Amongst additional epitope repertoires with unknown Chagas serostatus, assay specificity was 99.8% (998/1000). Thus, the Chagas assay achieved a combined sensitivity and specificity equivalent or superior to diagnostic algorithms that rely on three separate tests to achieve high specificity. NGS-based serology via SERA provides an effective approach to discover antigenic epitopes and develop high performance multiplex serological assays.

De novo design of antibody complementarity determining regions binding a FLAG tetra-peptide.

Computational antibody engineering efforts to date have focused on improving binding affinities or biophysical characteristics. De novo design of antibodies binding specific epitopes could greatly accelerate discovery of therapeutics as compared to conventional immunization or synthetic library selection strategies. Here, we employed de novo complementarity determining region (CDR) design to engineer targeted antibody-antigen interactions using previously described in silico methods. CDRs predicted to bind the minimal FLAG peptide (Asp-Tyr-Lys-Asp) were grafted onto a single-chain variable fragment (scFv) acceptor framework. Fifty scFvs comprised of designed heavy and light or just heavy chain CDRs were synthesized and screened for peptide binding by phage ELISA. Roughly half of the designs resulted in detectable scFv expression. Four antibodies, designed entirely in silico, bound the minimal FLAG sequence with high specificity and sensitivity. When reformatted as soluble antigen-binding fragments (Fab), these clones expressed well, were predominantly monomeric and retained peptide specificity. In both formats, the antibodies bind the peptide only when present at the amino-terminus of a carrier protein and even conservative peptide amino acid substitutions resulted in a complete loss of binding. These results support in silico CDR design of antibody specificity as an emerging antibody engineering strategy.

Identification of disease-specific motifs in the antibody specificity repertoire via next-generation sequencing.

Disease-specific antibodies can serve as highly effective biomarkers but have been identified for only a relatively small number of autoimmune diseases. A method was developed to identify disease-specific binding motifs through integration of bacterial display peptide library screening, next-generation sequencing (NGS) and computational analysis. Antibody specificity repertoires were determined by identifying bound peptide library members for each specimen using cell sorting and performing NGS. A computational algorithm, termed Identifying Motifs Using Next- generation sequencing Experiments (IMUNE), was developed and applied to discover disease- and healthy control-specific motifs. IMUNE performs comprehensive pattern searches, identifies patterns statistically enriched in the disease or control groups and clusters the patterns to generate motifs. Using celiac disease sera as a discovery set, IMUNE identified a consensus motif (QPEQPF[PS]E) with high diagnostic sensitivity and specificity in a validation sera set, in addition to novel motifs. Peptide display and sequencing (Display-Seq) coupled with IMUNE analysis may thus be useful to characterize antibody repertoires and identify disease-specific antibody epitopes and biomarkers.

The Iterative Protein Redesign and Optimization (IPRO) suite of programs.

Proteins are an important class of biomolecules with applications spanning across biotechnology and medicine. In many cases, native proteins must be redesigned to improve various performance metrics by changing their amino acid sequences. Algorithms can help sharpen protein library design by focusing the library on sequences that optimize computationally accessible proxies. The Iterative Protein Redesign and Optimization (IPRO) suite of programs offers an integrated environment for (1) altering protein binding affinity and specificity, (2) grafting a binding pocket into an existing protein scaffold, (3) predicting an antibody's tertiary structure based on its sequence, (4) enhancing enzymatic activity, and (5) assessing the structure and binding energetics for a specific mutant. This manuscript provides an overview of the methods involved in IPRO, input language terminology, algorithmic details, software implementation specifics and application highlights. IPRO can be downloaded at http://maranas.che.psu.edu.

OptMAVEn--a new framework for the de novo design of antibody variable region models targeting specific antigen epitopes.

Antibody-based therapeutics provides novel and efficacious treatments for a number of diseases. Traditional experimental approaches for designing therapeutic antibodies rely on raising antibodies against a target antigen in an immunized animal or directed evolution of antibodies with low affinity for the desired antigen. However, these methods remain time consuming, cannot target a specific epitope and do not lead to broad design principles informing other studies. Computational design methods can overcome some of these limitations by using biophysics models to rationally select antibody parts that maximize affinity for a target antigen epitope. This has been addressed to some extend by OptCDR for the design of complementary determining regions. Here, we extend this earlier contribution by addressing the de novo design of a model of the entire antibody variable region against a given antigen epitope while safeguarding for immunogenicity (Optimal Method for Antibody Variable region Engineering, OptMAVEn). OptMAVEn simulates in silico the in vivo steps of antibody generation and evolution, and is capable of capturing the critical structural features responsible for affinity maturation of antibodies. In addition, a humanization procedure was developed and incorporated into OptMAVEn to minimize the potential immunogenicity of the designed antibody models. As case studies, OptMAVEn was applied to design models of neutralizing antibodies targeting influenza hemagglutinin and HIV gp120. For both HA and gp120, novel computational antibody models with numerous interactions with their target epitopes were generated. The observed rates of mutations and types of amino acid changes during in silico affinity maturation are consistent with what has been observed during in vivo affinity maturation. The results demonstrate that OptMAVEn can efficiently generate diverse computational antibody models with both optimized binding affinity to antigens and reduced immunogenicity.

OptZyme: computational enzyme redesign using transition state analogues.

OptZyme is a new computational procedure for designing improved enzymatic activity (i.e., kcat or kcat/KM) with a novel substrate. The key concept is to use transition state analogue compounds, which are known for many reactions, as proxies for the typically unknown transition state structures. Mutations that minimize the interaction energy of the enzyme with its transition state analogue, rather than with its substrate, are identified that lower the transition state formation energy barrier. Using Escherichia coli ß-glucuronidase as a benchmark system, we confirm that KM correlates (R(2) = 0.960) with the computed interaction energy between the enzyme and the para-nitrophenyl- ß, D-glucuronide substrate, kcat/KM correlates (R(2) = 0.864) with the interaction energy of the transition state analogue, 1,5-glucarolactone, and kcat correlates (R(2) = 0.854) with a weighted combination of interaction energies with the substrate and transition state analogue. OptZyme is subsequently used to identify mutants with improved KM, kcat, and kcat/KM for a new substrate, para-nitrophenyl- ß, D-galactoside. Differences between the three libraries reveal structural differences that underpin improving KM, kcat, or kcat/KM. Mutants predicted to enhance the activity for para-nitrophenyl- ß, D-galactoside directly or indirectly create hydrogen bonds with the altered sugar ring conformation or its substituents, namely H162S, L361G, W549R, and N550S.

MAPs: a database of modular antibody parts for predicting tertiary structures and designing affinity matured antibodies.

BACKGROUND: The de novo design of a novel protein with a particular function remains a formidable challenge with only isolated and hard-to-repeat successes to date. Due to their many structurally conserved features, antibodies are a family of proteins amenable to predictable rational design. Design algorithms must consider the structural diversity of possible naturally occurring antibodies. The human immune system samples this design space (2 1012) by randomly combining variable, diversity, and joining genes in a process known as V-(D)-J recombination. DESCRIPTION: By analyzing structural features found in affinity matured antibodies, a database of Modular Antibody Parts (MAPs) analogous to the variable, diversity, and joining genes has been constructed for the prediction of antibody tertiary structures. The database contains 929 parts constructed from an analysis of 1168 human, humanized, chimeric, and mouse antibody structures and encompasses all currently observed structural diversity of antibodies. CONCLUSIONS: The generation of 260 antibody structures shows that the MAPs database can be used to reliably predict antibody tertiary structures with an average all-atom RMSD of 1.9 Å. Using the broadly neutralizing anti-influenza antibody CH65 and anti-HIV antibody 4E10 as examples, promising starting antibodies for affinity maturation are identified and amino acid changes are traced as antibody affinity maturation occurs.

Recent advances in computational protein design.

Proteins are the molecules cells primarily rely on for catalysis, recognition, signaling, defense, locomotion, and structural integrity. Engineering proteins for improved function or new applications is a fast-growing segment of biotechnology and biomedicine. Experimental efforts based on the screening of large mutant libraries have led to many successes but they do not provide quantitative design principles and/or insight into the structural features that underpin the desired function. The computational de novo design of proteins promises to bridge this gap; however, it requires reliable structure prediction, provisions for protein stability, and accurate descriptions of inter-molecule interactions. Studies that successfully meet all these criteria are beginning to emerge including the design of an O(2)-binding protein and a novel enzyme that catalyzes a Diels-Alder reaction.

Computational design of Candida boidinii xylose reductase for altered cofactor specificity.

In this study we introduce a computationally-driven enzyme redesign workflow for altering cofactor specificity from NADPH to NADH. By compiling and comparing data from previous studies involving cofactor switching mutations, we show that their effect cannot be explained as straightforward changes in volume, hydrophobicity, charge, or BLOSUM62 scores of the residues populating the cofactor binding site. Instead, we find that the use of a detailed cofactor binding energy approximation is needed to adequately capture the relative affinity towards different cofactors. The implicit solvation models Generalized Born with molecular volume integration and Generalized Born with simple switching were integrated in the iterative protein redesign and optimization (IPRO) framework to drive the redesign of Candida boidinii xylose reductase (CbXR) to function using the non-native cofactor NADH. We identified 10 variants, out of the 8,000 possible combinations of mutations, that improve the computationally assessed binding affinity for NADH by introducing mutations in the CbXR binding pocket. Experimental testing revealed that seven out of ten possessed significant xylose reductase activity utilizing NADH, with the best experimental design (CbXR-GGD) being 27-fold more active on NADH. The NADPH-dependent activity for eight out of ten predicted designs was either completely abolished or significantly diminished by at least 90%, yielding a greater than 10(4)-fold change in specificity to NADH (CbXR-REG). The remaining two variants (CbXR-RTT and CBXR-EQR) had dual cofactor specificity for both nicotinamide cofactors.

Optimal protein library design using recombination or point mutations based on sequence-based scoring functions.

In this paper, we introduce and test two new sequence-based protein scoring systems (i.e. S1, S2) for assessing the likelihood that a given protein hybrid will be functional. By binning together amino acids with similar properties (i.e. volume, hydrophobicity and charge) the scoring systems S1 and S2 allow for the quantification of the severity of mismatched interactions in the hybrids. The S2 scoring system is found to be able to significantly functionally enrich a cytochrome P450 library over other scoring methods. Given this scoring base, we subsequently constructed two separate optimization formulations (i.e. OPTCOMB and OPTOLIGO) for optimally designing protein combinatorial libraries involving recombination or mutations, respectively. Notably, two separate versions of OPTCOMB are generated (i.e. model M1, M2) with the latter allowing for position-dependent parental fragment skipping. Computational benchmarking results demonstrate the efficacy of models OPTCOMB and OPTOLIGO to generate high scoring libraries of a prespecified size.