Designing antibodies as therapeutics.

Antibody therapeutics are a large and rapidly expanding drug class providing major health benefits. We provide a snapshot of current antibody therapeutics including their formats, common targets, therapeutic areas, and routes of administration. Our focus is on selected emerging directions in antibody design where progress may provide a broad benefit. These topics include enhancing antibodies for cancer, antibody delivery to organs such as the brain, gastrointestinal tract, and lungs, plus antibody developability challenges including immunogenicity risk assessment and mitigation and subcutaneous delivery. Machine learning has the potential, albeit as yet largely unrealized, for a transformative future impact on antibody discovery and engineering.

Glycoengineering of Antibodies for Modulating Functions.

Antibodies are immunoglobulins that play essential roles in immune systems. All antibodies are glycoproteins that carry at least one or more conserved N-linked oligosaccharides (N-glycans) at the Fc domain. Many studies have demonstrated that both the presence and fine structures of the attached glycans can exert a profound impact on the biological functions and therapeutic efficacy of antibodies. However, antibodies usually exist as mixtures of heterogeneous glycoforms that are difficult to separate in pure glycoforms. Recent progress in glycoengineering has provided useful methods that enable production of glycan-defined and site-selectively modified antibodies for functional studies and for improved therapeutic efficacy. This review highlights major approaches in glycoengineering of antibodies with a focus on recent advances in three areas: glycoengineering through glycan biosynthetic pathway manipulation, glycoengineering through in vitro chemoenzymatic glycan remodeling, and glycoengineering of antibodies for site-specific antibody-drug conjugation.

Deep Profiling of Mouse Splenic Architecture with CODEX Multiplexed Imaging.

A highly multiplexed cytometric imaging approach, termed co-detection by indexing (CODEX), is used here to create multiplexed datasets of normal and lupus (MRL/lpr) murine spleens. CODEX iteratively visualizes antibody binding events using DNA barcodes, fluorescent dNTP analogs, and an in situ polymerization-based indexing procedure. An algorithmic pipeline for single-cell antigen quantification in tightly packed tissues was developed and used to overlay well-known morphological features with de novo characterization of lymphoid tissue architecture at a single-cell and cellular neighborhood levels. We observed an unexpected, profound impact of the cellular neighborhood on the expression of protein receptors on immune cells. By comparing normal murine spleen to spleens from animals with systemic autoimmune disease (MRL/lpr), extensive and previously uncharacterized splenic cell-interaction dynamics in the healthy versus diseased state was observed. The fidelity of multiplexed spatial cytometry demonstrated here allows for quantitative systemic characterization of tissue architecture in normal and clinically aberrant samples.

A Method for the Acute and Rapid Degradation of Endogenous Proteins.

Methods for the targeted disruption of protein function have revolutionized science and greatly expedited the systematic characterization of genes. Two main approaches are currently used to disrupt protein function: DNA knockout and RNA interference, which act at the genome and mRNA level, respectively. A method that directly alters endogenous protein levels is currently not available. Here, we present Trim-Away, a technique to degrade endogenous proteins acutely in mammalian cells without prior modification of the genome or mRNA. Trim-Away harnesses the cellular protein degradation machinery to remove unmodified native proteins within minutes of application. This rapidity minimizes the risk that phenotypes are compensated and that secondary, non-specific defects accumulate over time. Because Trim-Away utilizes antibodies, it can be applied to a wide range of target proteins using off-the-shelf reagents. Trim-Away allows the study of protein function in diverse cell types, including non-dividing primary cells where genome- and RNA-targeting methods are limited.

Accurate structure prediction of biomolecular interactions with AlphaFold 3.

The introduction of AlphaFold 21 has spurred a revolution in modelling the structure of proteins and their interactions, enabling a huge range of applications in protein modelling and design2-6. Here we describe our AlphaFold 3 model with a substantially updated diffusion-based architecture that is capable of predicting the joint structure of complexes including proteins, nucleic acids, small molecules, ions and modified residues. The new AlphaFold model demonstrates substantially improved accuracy over many previous specialized tools: far greater accuracy for protein-ligand interactions compared with state-of-the-art docking tools, much higher accuracy for protein-nucleic acid interactions compared with nucleic-acid-specific predictors and substantially higher antibody-antigen prediction accuracy compared with AlphaFold-Multimer v.2.37,8. Together, these results show that high-accuracy modelling across biomolecular space is possible within a single unified deep-learning framework.

Nanobodies: natural single-domain antibodies.

Sera of camelids contain both conventional heterotetrameric antibodies and unique functional heavy (H)-chain antibodies (HCAbs). The H chain of these homodimeric antibodies consists of one antigen-binding domain, the VHH, and two constant domains. HCAbs fail to incorporate light (L) chains owing to the deletion of the first constant domain and a reshaped surface at the VHH side, which normally associates with L chains in conventional antibodies. The genetic elements composing HCAbs have been identified, but the in vivo generation of these antibodies from their dedicated genes into antigen-specific and affinity-matured bona fide antibodies remains largely underinvestigated. However, the facile identification of antigen-specific VHHs and their beneficial biochemical and economic properties (size, affinity, specificity, stability, production cost) supported by multiple crystal structures have encouraged antibody engineering of these single-domain antibodies for use as a research tool and in biotechnology and medicine.

Functional antibodies exhibit light chain coherence.

The vertebrate adaptive immune system modifies the genome of individual B cells to encode antibodies that bind particular antigens1. In most mammals, antibodies are composed of heavy and light chains that are generated sequentially by recombination of V, D (for heavy chains), J and C gene segments. Each chain contains three complementarity-determining regions (CDR1-CDR3), which contribute to antigen specificity. Certain heavy and light chains are preferred for particular antigens2-22. Here we consider pairs of B cells that share the same heavy chain V gene and CDRH3 amino acid sequence and were isolated from different donors, also known as public clonotypes23,24. We show that for naive antibodies (those not yet adapted to antigens), the probability that they use the same light chain V gene is around 10%, whereas for memory (functional) antibodies, it is around 80%, even if only one cell per clonotype is used. This property of functional antibodies is a phenomenon that we call light chain coherence. We also observe this phenomenon when similar heavy chains recur within a donor. Thus, although naive antibodies seem to recur by chance, the recurrence of functional antibodies reveals surprising constraint and determinism in the processes of V(D)J recombination and immune selection. For most functional antibodies, the heavy chain determines the light chain.

Structural basis of malaria RIFIN binding by LILRB1-containing antibodies.

Some Plasmodium falciparum repetitive interspersed families of polypeptides (RIFINs)-variant surface antigens that are expressed on infected erythrocytes1-bind to the inhibitory receptor LAIR1, and insertion of DNA that encodes LAIR1 into immunoglobulin genes generates RIFIN-specific antibodies2,3. Here we address the general relevance of this finding by searching for antibodies that incorporate LILRB1, another inhibitory receptor that binds to ß2 microglobulin and RIFINs through their apical domains4,5. By screening plasma from a cohort of donors from Mali, we identified individuals with LILRB1-containing antibodies. B cell clones isolated from three donors showed large DNA insertions in the switch region that encodes non-apical LILRB1 extracellular domain 3 and 4 (D3D4) or D3 alone in the variable-constant (VH-CH1) elbow. Through mass spectrometry and binding assays, we identified a large set of RIFINs that bind to LILRB1 D3. Crystal and cryo-electron microscopy structures of a RIFIN in complex with either LILRB1 D3D4 or a D3D4-containing antibody Fab revealed a mode of RIFIN-LILRB1 D3 interaction that is similar to that of RIFIN-LAIR1. The Fab showed an unconventional triangular architecture with the inserted LILRB1 domains opening up the VH-CH1 elbow without affecting VH-VL or CH1-CL pairing. Collectively, these findings show that RIFINs bind to LILRB1 through D3 and illustrate, with a naturally selected example, the general principle of creating novel antibodies by inserting receptor domains into the VH-CH1 elbow.

Multiplexed transcriptome discovery of RNA-binding protein binding sites by antibody-barcode eCLIP.

Ultraviolet crosslinking and immunoprecipitation (CLIP) methodologies enable the identification of RNA binding sites of RNA-binding proteins (RBPs). Despite improvements in the library preparation of RNA fragments, the enhanced CLIP (eCLIP) protocol requires 4 days of hands-on time and lacks the ability to process several RBPs in parallel. We present a new method termed antibody-barcode eCLIP that utilizes DNA-barcoded antibodies and proximity ligation of the DNA oligonucleotides to RBP-protected RNA fragments to interrogate several RBPs simultaneously. We observe performance comparable with that of eCLIP with the advantage of dramatically increased scaling while maintaining the same material requirement of a single eCLIP experiment.

Accurate prediction of antibody function and structure using bio-inspired antibody language model.

In recent decades, antibodies have emerged as indispensable therapeutics for combating diseases, particularly viral infections. However, their development has been hindered by limited structural information and labor-intensive engineering processes. Fortunately, significant advancements in deep learning methods have facilitated the precise prediction of protein structure and function by leveraging co-evolution information from homologous proteins. Despite these advances, predicting the conformation of antibodies remains challenging due to their unique evolution and the high flexibility of their antigen-binding regions. Here, to address this challenge, we present the Bio-inspired Antibody Language Model (BALM). This model is trained on a vast dataset comprising 336 million 40% nonredundant unlabeled antibody sequences, capturing both unique and conserved properties specific to antibodies. Notably, BALM showcases exceptional performance across four antigen-binding prediction tasks. Moreover, we introduce BALMFold, an end-to-end method derived from BALM, capable of swiftly predicting full atomic antibody structures from individual sequences. Remarkably, BALMFold outperforms those well-established methods like AlphaFold2, IgFold, ESMFold and OmegaFold in the antibody benchmark, demonstrating significant potential to advance innovative engineering and streamline therapeutic antibody development by reducing the need for unnecessary trials. The BALMFold structure prediction server is freely available at https://beamlab-sh.com/models/BALMFold.

Antibody design using deep learning: from sequence and structure design to affinity maturation.

Deep learning has achieved impressive results in various fields such as computer vision and natural language processing, making it a powerful tool in biology. Its applications now encompass cellular image classification, genomic studies and drug discovery. While drug development traditionally focused deep learning applications on small molecules, recent innovations have incorporated it in the discovery and development of biological molecules, particularly antibodies. Researchers have devised novel techniques to streamline antibody development, combining in vitro and in silico methods. In particular, computational power expedites lead candidate generation, scaling and potential antibody development against complex antigens. This survey highlights significant advancements in protein design and optimization, specifically focusing on antibodies. This includes various aspects such as design, folding, antibody-antigen interactions docking and affinity maturation.

Development and experimental validation of computational methods for human antibody affinity enhancement.

High affinity is crucial for the efficacy and specificity of antibody. Due to involving high-throughput screens, biological experiments for antibody affinity maturation are time-consuming and have a low success rate. Precise computational-assisted antibody design promises to accelerate this process, but there is still a lack of effective computational methods capable of pinpointing beneficial mutations within the complementarity-determining region (CDR) of antibodies. Moreover, random mutations often lead to challenges in antibody expression and immunogenicity. In this study, to enhance the affinity of a human antibody against avian influenza virus, a CDR library was constructed and evolutionary information was acquired through sequence alignment to restrict the mutation positions and types. Concurrently, a statistical potential methodology was developed based on amino acid interactions between antibodies and antigens to calculate potential affinity-enhanced antibodies, which were further subjected to molecular dynamics simulations. Subsequently, experimental validation confirmed that a point mutation enhancing 2.5-fold affinity was obtained from 10 designs, resulting in the antibody affinity of 2 nM. A predictive model for antibody-antigen interactions based on the binding interface was also developed, achieving an Area Under the Curve (AUC) of 0.83 and a precision of 0.89 on the test set. Lastly, a novel approach involving combinations of affinity-enhancing mutations and an iterative mutation optimization scheme similar to the Monte Carlo method were proposed. This study presents computational methods that rapidly and accurately enhance antibody affinity, addressing issues related to antibody expression and immunogenicity.

Predictability of antigen binding based on short motifs in the antibody CDRH3.

Adaptive immune receptors, such as antibodies and T-cell receptors, recognize foreign threats with exquisite specificity. A major challenge in adaptive immunology is discovering the rules governing immune receptor-antigen binding in order to predict the antigen binding status of previously unseen immune receptors. Many studies assume that the antigen binding status of an immune receptor may be determined by the presence of a short motif in the complementarity determining region 3 (CDR3), disregarding other amino acids. To test this assumption, we present a method to discover short motifs which show high precision in predicting antigen binding and generalize well to unseen simulated and experimental data. Our analysis of a mutagenesis-based antibody dataset reveals 11 336 position-specific, mostly gapped motifs of 3-5 amino acids that retain high precision on independently generated experimental data. Using a subset of only 178 motifs, a simple classifier was made that on the independently generated dataset outperformed a deep learning model proposed specifically for such datasets. In conclusion, our findings support the notion that for some antibodies, antigen binding may be largely determined by a short CDR3 motif. As more experimental data emerge, our methodology could serve as a foundation for in-depth investigations into antigen binding signals.

AttABseq: an attention-based deep learning prediction method for antigen-antibody binding affinity changes based on protein sequences.

The optimization of therapeutic antibodies through traditional techniques, such as candidate screening via hybridoma or phage display, is resource-intensive and time-consuming. In recent years, computational and artificial intelligence-based methods have been actively developed to accelerate and improve the development of therapeutic antibodies. In this study, we developed an end-to-end sequence-based deep learning model, termed AttABseq, for the predictions of the antigen-antibody binding affinity changes connected with antibody mutations. AttABseq is a highly efficient and generic attention-based model by utilizing diverse antigen-antibody complex sequences as the input to predict the binding affinity changes of residue mutations. The assessment on the three benchmark datasets illustrates that AttABseq is 120% more accurate than other sequence-based models in terms of the Pearson correlation coefficient between the predicted and experimental binding affinity changes. Moreover, AttABseq also either outperforms or competes favorably with the structure-based approaches. Furthermore, AttABseq consistently demonstrates robust predictive capabilities across a diverse array of conditions, underscoring its remarkable capacity for generalization across a wide spectrum of antigen-antibody complexes. It imposes no constraints on the quantity of altered residues, rendering it particularly applicable in scenarios where crystallographic structures remain unavailable. The attention-based interpretability analysis indicates that the causal effects of point mutations on antibody-antigen binding affinity changes can be visualized at the residue level, which might assist automated antibody sequence optimization. We believe that AttABseq provides a fiercely competitive answer to therapeutic antibody optimization.

Antibody discovery identifies regulatory mechanisms of protein arginine deiminase 4.

Unlocking the potential of protein arginine deiminase 4 (PAD4) as a drug target for rheumatoid arthritis requires a deeper understanding of its regulation. In this study, we use unbiased antibody selections to identify functional antibodies capable of either activating or inhibiting PAD4 activity. Through cryogenic-electron microscopy, we characterized the structures of these antibodies in complex with PAD4 and revealed insights into their mechanisms of action. Rather than steric occlusion of the substrate-binding catalytic pocket, the antibodies modulate PAD4 activity through interactions with allosteric binding sites adjacent to the catalytic pocket. These binding events lead to either alteration of the active site conformation or the enzyme oligomeric state, resulting in modulation of PAD4 activity. Our study uses antibody engineering to reveal new mechanisms for enzyme regulation and highlights the potential of using PAD4 agonist and antagonist antibodies for studying PAD4-dependency in disease models and future therapeutic development.

Lysosome-targeting chimaeras for degradation of extracellular proteins.

The majority of therapies that target individual proteins rely on specific activity-modulating interactions with the target protein-for example, enzyme inhibition or ligand blocking. However, several major classes of therapeutically relevant proteins have unknown or inaccessible activity profiles and so cannot be targeted by such strategies. Protein-degradation platforms such as proteolysis-targeting chimaeras (PROTACs)1,2 and others (for example, dTAGs3, Trim-Away4, chaperone-mediated autophagy targeting5 and SNIPERs6) have been developed for proteins that are typically difficult to target; however, these methods involve the manipulation of intracellular protein degradation machinery and are therefore fundamentally limited to proteins that contain cytosolic domains to which ligands can bind and recruit the requisite cellular components. Extracellular and membrane-associated proteins-the products of 40% of all protein-encoding genes7-are key agents in cancer, ageing-related diseases and autoimmune disorders8, and so a general strategy to selectively degrade these proteins has the potential to improve human health. Here we establish the targeted degradation of extracellular and membrane-associated proteins using conjugates that bind both a cell-surface lysosome-shuttling receptor and the extracellular domain of a target protein. These initial lysosome-targeting chimaeras, which we term LYTACs, consist of a small molecule or antibody fused to chemically synthesized glycopeptide ligands that are agonists of the cation-independent mannose-6-phosphate receptor (CI-M6PR). We use LYTACs to develop a CRISPR interference screen that reveals the biochemical pathway for CI-M6PR-mediated cargo internalization in cell lines, and uncover the exocyst complex as a previously unidentified-but essential-component of this pathway. We demonstrate the scope of this platform through the degradation of therapeutically relevant proteins, including apolipoprotein E4, epidermal growth factor receptor, CD71 and programmed death-ligand 1. Our results establish a modular strategy for directing secreted and membrane proteins for lysosomal degradation, with broad implications for biochemical research and for therapeutics.

Examining the Roles of H3K4 Methylation States with Systematically Characterized Antibodies.

Histone post-translational modifications (PTMs) are important genomic regulators often studied by chromatin immunoprecipitation (ChIP), whereby their locations and relative abundance are inferred by antibody capture of nucleosomes and associated DNA. However, the specificity of antibodies within these experiments has not been systematically studied. Here, we use histone peptide arrays and internally calibrated ChIP (ICeChIP) to characterize 52 commercial antibodies purported to distinguish the H3K4 methylforms (me1, me2, and me3, with each ascribed distinct biological functions). We find that many widely used antibodies poorly distinguish the methylforms and that high- and low-specificity reagents can yield dramatically different biological interpretations, resulting in substantial divergence from the literature for numerous H3K4 methylform paradigms. Using ICeChIP, we also discern quantitative relationships between enhancer H3K4 methylation and promoter transcriptional output and can measure global PTM abundance changes. Our results illustrate how poor antibody specificity contributes to the "reproducibility crisis," demonstrating the need for rigorous, platform-appropriate validation.

The Patent and Literature Antibody Database (PLAbDab): an evolving reference set of functionally diverse, literature-annotated antibody sequences and structures.

Antibodies are key proteins of the adaptive immune system, and there exists a large body of academic literature and patents dedicated to their study and concomitant conversion into therapeutics, diagnostics, or reagents. These documents often contain extensive functional characterisations of the sets of antibodies they describe. However, leveraging these heterogeneous reports, for example to offer insights into the properties of query antibodies of interest, is currently challenging as there is no central repository through which this wide corpus can be mined by sequence or structure. Here, we present PLAbDab (the Patent and Literature Antibody Database), a self-updating repository containing over 150,000 paired antibody sequences and 3D structural models, of which over 65 000 are unique. We describe the methods used to extract, filter, pair, and model the antibodies in PLAbDab, and showcase how PLAbDab can be searched by sequence, structure, or keyword. PLAbDab uses include annotating query antibodies with potential antigen information from similar entries, analysing structural models of existing antibodies to identify modifications that could improve their properties, and facilitating the compilation of bespoke datasets of antibody sequences/structures that bind to a specific antigen. PLAbDab is freely available via Github (https://github.com/oxpig/PLAbDab) and as a searchable webserver (https://opig.stats.ox.ac.uk/webapps/plabdab/).

Characterization and decontamination of background noise in droplet-based single-cell protein expression data with DecontPro.

Assays such as CITE-seq can measure the abundance of cell surface proteins on individual cells using antibody derived tags (ADTs). However, many ADTs have high levels of background noise that can obfuscate down-stream analyses. In an exploratory analysis of PBMC datasets, we find that some droplets that were originally called 'empty' due to low levels of RNA contained high levels of ADTs and likely corresponded to neutrophils. We identified a novel type of artifact in the empty droplets called a 'spongelet' which has medium levels of ADT expression and is distinct from ambient noise. ADT expression levels in the spongelets correlate to ADT expression levels in the background peak of true cells in several datasets suggesting that they can contribute to background noise along with ambient ADTs. We then developed DecontPro, a novel Bayesian hierarchical model that can decontaminate ADT data by estimating and removing contamination from these sources. DecontPro outperforms other decontamination tools in removing aberrantly expressed ADTs while retaining native ADTs and in improving clustering specificity. Overall, these results suggest that identification of empty drops should be performed separately for RNA and ADT data and that DecontPro can be incorporated into CITE-seq workflows to improve the quality of downstream analyses.

Multiplexed in situ protein imaging using DNA-barcoded antibodies with extended hybridization chain reactions.

Antibodies have long served as vital tools in biological and clinical laboratories for the specific detection of proteins. Conventional methods employ fluorophore or horseradish peroxidase-conjugated antibodies to detect signals. More recently, DNA-conjugated antibodies have emerged as a promising technology, capitalizing on the programmability and amplification capabilities of DNA to enable highly multiplexed and ultrasensitive protein detection. However, the nonspecific binding of DNA-conjugated antibodies has impeded the widespread adoption of this approach. Here, we present a novel DNA-conjugated antibody staining protocol that addresses these challenges and demonstrates superior performance in suppressing nonspecific signals compared to previously published protocols. We further extend the utility of DNA-conjugated antibodies for signal-amplified in situ protein imaging through the hybridization chain reaction (HCR) and design a novel HCR DNA pair to expand the HCR hairpin pool from the previously published 5 pairs to 13, allowing for flexible hairpin selection and higher multiplexing. Finally, we demonstrate highly multiplexed in situ protein imaging using these techniques in both cultured cells and tissue sections.