Antibody informatics for drug discovery.

More and more antibody therapeutics are being approved every year, mainly due to their high efficacy and antigen selectivity. However, it is still difficult to identify the antigen, and thereby the function, of an antibody if no other information is available. There are obstacles inherent to the antibody science in every project in antibody drug discovery. Recent experimental technologies allow for the rapid generation of large-scale data on antibody sequences, affinity, potency, structures, and biological functions; this should accelerate drug discovery research. Therefore, a robust bioinformatic infrastructure for these large data sets has become necessary. In this article, we first identify and discuss the typical obstacles faced during the antibody drug discovery process. We then summarize the current status of three sub-fields of antibody informatics as follows: (i) recent progress in technologies for antibody rational design using computational approaches to affinity and stability improvement, as well as ab-initio and homology-based antibody modeling; (ii) resources for antibody sequences, structures, and immune epitopes and open drug discovery resources for development of antibody drugs; and (iii) antibody numbering and IMGT. Here, we review "antibody informatics," which may integrate the above three fields so that bridging the gaps between industrial needs and academic solutions can be accelerated. This article is part of a Special Issue entitled: Recent advances in molecular engineering of antibody.

High-throughput analysis system of interaction kinetics for data-driven antibody design.

Surface plasmon resonance (SPR) is widely used for antigen-antibody interaction kinetics analysis. However, it has not been used in the screening phase because of the low throughput of measurement and analysis. Herein, we proposed a high-throughput SPR analysis system named "BreviA" using the Brevibacillus expression system. Brevibacillus was transformed using a plasmid library containing various antibody sequences, and single colonies were cultured in 96-well plates. Sequence analysis was performed using bacterial cells, and recombinant antibodies secreted in the supernatant were immobilized on a sensor chip to analyze their interactions with antigens using high-throughput SPR. Using this system, the process from the transformation to 384 interaction analyses can be performed within a week. This system utility was tested using an interspecies specificity design of an anti-human programmed cell death protein 1 (PD-1) antibody. A plasmid library containing alanine and tyrosine mutants of all complementarity-determining region residues was generated. A high-throughput SPR analysis was performed against human and mouse PD-1, showing that the mutation in the specific region enhanced the affinity for mouse PD-1. Furthermore, deep mutational scanning of the region revealed two mutants with > 100-fold increased affinity for mouse PD-1, demonstrating the potential efficacy of antibody design using data-driven approach.

Computational design, construction, and characterization of a set of specificity determining residues in protein-protein interactions.

Proteins interact with different partners to perform different functions and it is important to elucidate the determinants of partner specificity in protein complex formation. Although methods for detecting specificity determining positions have been developed previously, direct experimental evidence for these amino acid residues is scarce, and the lack of information has prevented further computational studies. In this article, we constructed a dataset that is likely to exhibit specificity in protein complex formation, based on available crystal structures and several intuitive ideas about interaction profiles and functional subclasses. We then defined a "structure-based specificity determining position (sbSDP)" as a set of equivalent residues in a protein family showing a large variation in their interaction energy with different partners. We investigated sequence and structural features of sbSDPs and demonstrated that their amino acid propensities significantly differed from those of other interacting residues and that the importance of many of these residues for determining specificity had been verified experimentally.

Novel TLR2xTLR4 Bispecific Antibody Inhibits Bacterial Sepsis.

Toll-like receptors (TLRs) sense microbial infection through recognition of pathogen-associated molecular patterns. For example, TLR4 responds to the lipopolysaccharide of gram-negative bacteria, whereas TLR2 recognizes a broad range of microbial ligands. Both receptors are, therefore, compelling targets for treating sepsis. Here, we developed a TLR2xTLR4 tetravalent bispecific antibody designated ICU-1, which inhibits both receptors. The inhibitory activity of ICU-1 was comparable to that of the parental antibodies in neutralization assays using a human monocyte cell line. Moreover, ICU-1 completely blocked stimulation of human peripheral blood mononuclear cells by clinically relevant bacterial species. These findings provide convincing evidence that ICU-1 offers a novel approach for treating bacterial sepsis.

Generation and Characterization of a Novel Anti-Rat TLR4/MD2 Antibody with Potent Neutralizing Activity In Vivo.

Toll-like receptor 4 (TLR4) plays a critical role in the innate immune system and is involved in the pathogenesis of multiple diseases. Here, we report the antagonistic and ratized antibody, 52-1H4 e2 (e2), which completely inhibited lipopolysaccharide-induced interleukin-6 secretion in vitro. The average serum drug concentration was above 10 µg/mL for 28 days in rats injected with e2. The novel anti-rat TLR4/myeloid differentiation factor 2 antibody, e2, may be a useful tool for investigating the role of TLR4 in rat disease models.

Exploring designability of electrostatic complementarity at an antigen-antibody interface directed by mutagenesis, biophysical analysis, and molecular dynamics simulations.

Antibodies protect organisms from a huge variety of foreign antigens. Antibody diversity originates from both genetic and structural levels. Antigen recognition relies on complementarity between antigen-antibody interfaces. Recent methodological advances in structural biology and the accompanying rapid increase of the number of crystal structures of proteins have enabled atomic-level manipulation of protein structures to effect alterations in function. In this study, we explored the designability of electrostatic complementarity at an antigen-antibody interface on the basis of a crystal structure of the complex. We designed several variants with altered charged residues at the interface and characterized the designed variants by surface plasmon resonance, circular dichroism, differential scanning calorimetry, and molecular dynamics simulations. Both successes and failures of the structure-based design are discussed. The variants that compensate electrostatic interactions can restore the interface complementarity, enabling the cognate antigen-antibody binding. Retrospectively, we also show that these mutational effects could be predicted by the simulations. Our study demonstrates the importance of charged residues on the physical properties of this antigen-antibody interaction and suggests that computational approaches can facilitate design of antibodies that recognize a weakly immunogenic antigen.

Affinity improvement of a therapeutic antibody by structure-based computational design: generation of electrostatic interactions in the transition state stabilizes the antibody-antigen complex.

The optimization of antibodies is a desirable goal towards the development of better therapeutic strategies. The antibody 11K2 was previously developed as a therapeutic tool for inflammatory diseases, and displays very high affinity (4.6 pM) for its antigen the chemokine MCP-1 (monocyte chemo-attractant protein-1). We have employed a virtual library of mutations of 11K2 to identify antibody variants of potentially higher affinity, and to establish benchmarks in the engineering of a mature therapeutic antibody. The most promising candidates identified in the virtual screening were examined by surface plasmon resonance to validate the computational predictions, and to characterize their binding affinity and key thermodynamic properties in detail. Only mutations in the light-chain of the antibody are effective at enhancing its affinity for the antigen in vitro, suggesting that the interaction surface of the heavy-chain (dominated by the hot-spot residue Phe101) is not amenable to optimization. The single-mutation with the highest affinity is L-N31R (4.6-fold higher affinity than wild-type antibody). Importantly, all the single-mutations showing increase affinity incorporate a charged residue (Arg, Asp, or Glu). The characterization of the relevant thermodynamic parameters clarifies the energetic mechanism. Essentially, the formation of new electrostatic interactions early in the binding reaction coordinate (transition state or earlier) benefits the durability of the antibody-antigen complex. The combination of in silico calculations and thermodynamic analysis is an effective strategy to improve the affinity of a matured therapeutic antibody.

Use of amino acid composition to predict epitope residues of individual antibodies.

We identified specific amino acid propensities at the interfaces of antigen-antibody interactions in non-redundant qualified antigen-antibody complex structures from Protein Data Bank. Propensities were expressed by the frequency of each of the 20 x 20 standard amino acid pairs that appeared at the interfaces of the complexes and were named the antibody-specific epitope propensity (ASEP) index. Using this index, we developed a novel method of predicting epitope residues for individual antibodies by narrowing down candidate epitope residues which was predicted by the conventional method. The 74 benchmarked antigens were used in ASEP prediction. The efficiency of this method was assessed using the leave-one-out approach. On elimination of residues with ASEP indices in the lowest 10% of all measured, true positives were enriched for 49 antigens. On subsequent elimination of residues with ASEP indices in the lowest 50%, true positives were enriched for 40 of the 74 antigens assessed. The ASEP index is the first benchmark proposed to predict epitope residues for an individual antibody. Used in combination with mutation experiments, this index has the potential to markedly increase the success ratio of epitope analysis.

Chemocavity: specific concavity in protein reserved for the binding of biologically functional small molecules.

The idea that there should be a specific site on a protein for a particular functional small molecule is widespread. It is, however, usually not so easy to understand what characteristics of the site determine the binding ability of the functional small molecule. We have focused on the concurrence rate of the 20 standard amino acids at such binding sites. In order to correlate the concurrence rate and the specific binding site, we have analyzed high-quality X-ray structures of complexes between proteins and small molecules. A novel index characterizing the binding site based on the concurrency rate has been introduced. Using this index we have identified that there is a specific concavity designated as a chemocavity where a specific group of small molecules, i.e., canonical molecular group, is highly inclined to be bound. This study has demonstrated that a chemocavity is reserved for a specific canonical molecular group, and the prevalent idea has been confirmed.

Use of amino acid composition to predict ligand-binding sites.

A novel method for predicting the binding sites for druglike compounds on the surface of proteins was developed on the basis of the specific amino acid composition observed at the ligand-binding sites of ligand-protein complexes determined by X-ray analysis. A profile representing the preference of each of the 20 standard amino acids at the binding sites of druglike molecules was obtained for a small set of high-quality complex structures. An index termed propensity for ligand binding (PLB) was created from these profiles. The PLB index was used to predict the propensity of binding for 804 ligands at all potential binding sites on the proteins whose structures were determined by X-ray analysis. If the sites with the first two highest PLB indices are taken into consideration, the successfully predicted sites reached a high percentage of 86. The PLB prediction is relatively simple, but the validation study showed that it is both fast and accurate to detect ligand-binding sites, especially the binding sites of druglike molecules. Therefore, the PLB index can be used to predict the ligand-binding sites of uncharacterized protein structures and also to identify novel drug-binding sites of known drug targets.

Identification of the druggable concavity in homology models using the PLB index.

Identification of the druggable concavity, in which drug-like molecules are highly inclined to bind, is an important step in structure-based drug design. We previously proposed an index named PLB (propensity for ligand binding), which is based on the amino acid composition characteristically observed at the small molecule binding sites in the X-ray structures of the complexes between proteins and drug-like small molecules. The PLB index was proven to be useful in identifying the druggable concavities in the quality X-ray structures of proteins. Here, we apply the PLB to predicting the druggable concavity in target proteins using the structures of homologous proteins constructed by homology modeling. In this study, we assembled a set of reference proteins that were accurately determined by X-ray analysis in forms of complexes with drug-like small molecules. Homology models for the reference protein were constructed using multiple homologous proteins as templates. The PLB index was then used to predict the druggable concavity. If the template protein in a complex with a drug-like small molecule was used, the druggable concavity was predicted well, with a prediction rate of 78%. When only the apo protein was available as the template, the practical prediction rate was 71%. Interestingly, even when the percent sequence identity between the reference and template proteins was lower than 30, the PLB index could successfully identify the druggable concavity in some cases. This study demonstrates the practical value of applying the PLB index to identifying the drugabble concavity in the homology model.