Explore the vast range of titles available.

Artificial Intelligence (AI) techniques have made great advances in assisting antibody design. However, antibody design still heavily relies on isolating antigen-specific antibodies from serum, which is a resource-intensive and time-consuming process. To address this issue, we propose a Pre-trained Antibody generative large Language Model (PALM-H3) for the de novo generation of artificial antibodies heavy chain complementarity-determining region 3 (CDRH3) with desired antigen-binding specificity, reducing the reliance on natural antibodies. We also build a high-precision model antigen-antibody binder (A2binder) that pairs antigen epitope sequences with antibody sequences to predict binding specificity and affinity. PALM-H3-generated antibodies exhibit binding ability to SARS-CoV-2 antigens, including the emerging XBB variant, as confirmed through in-silico analysis and in-vitro assays. The in-vitro assays validate that PALM-H3-generated antibodies achieve high binding affinity and potent neutralization capability against spike proteins of SARS-CoV-2 wild-type, Alpha, Delta, and the emerging XBB variant. Meanwhile, A2binder demonstrates exceptional predictive performance on binding specificity for various epitopes and variants. Furthermore, by incorporating the attention mechanism inherent in the Roformer architecture into the PALM-H3 model, we improve its interpretability, providing crucial insights into the fundamental principles of antibody design. Antibody design still heavily relies on isolating antigen-specific antibodies from serum. Here the authors report a Pre-trained Antibody generative large Language Model (PALM-H3) for the de novo generation of artificial antibodies heavy chain complementarity-determining region 3 with desired antigen-binding specificity.

The vertebrate adaptive immune system modifies the genome of individual B cells to encode antibodies that bind particular antigens 1 . 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 antigens 2 – 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 clonotypes 23 , 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.  Among naturally occurring antibodies that have adapted to antigen, those with similar heavy chains usually have similar light chains.

Defects in T cell receptor (TCR) repertoire are proposed to predispose to autoimmunity. Here we show, by analyzing >2 × 10 8 TCRB sequences of circulating naive, central memory, regulatory and stem cell-like memory CD4 + T cell subsets from patients with type 1 diabetes and healthy donors, that patients have shorter TCRB complementarity-determining region 3s (CDR3), in all cell subsets, introduced by increased deletions/reduced insertions during VDJ rearrangement. High frequency of short CDR3s is also observed in unproductive TCRB sequences, which are not subjected to thymic culling, suggesting that the shorter CDR3s arise independently of positive/negative selection. Moreover, TCRB CDR3 clonotypes expressed by autoantigen-specific CD4 + T cells are shorter compared with anti-viral T cells, and with those from healthy donors. Thus, early events in thymic T cell development and repertoire generation are abnormal in type 1 diabetes, which suggest that short CDR3s increase the potential for self-recognition, conferring heightened risk of autoimmune disease. T cell receptors are generated by somatic gene recombination, and are normally selected against autoreactivity. Here the authors show that CD4 T cells from patients with autoimmune type 1 diabetes have shorter TCRβ sequences, broader repertoire diversity, and more repertoire sharing than those from healthy individuals.

Immune recognition of protein antigens relies on the combined interaction of multiple antibody loops, which provide a fairly large footprint and constrain the size and shape of protein surfaces that can be targeted. Single protein loops can mediate extremely high-affinity binding, but it is unclear whether such a mechanism is available to antibodies. Here we report the isolation and characterization of an antibody called C05, which neutralizes strains from multiple subtypes of influenza A virus, including H1, H2 and H3. X-ray and electron microscopy structures show that C05 recognizes conserved elements of the receptor-binding site on the haemagglutinin surface glycoprotein. Recognition of the haemagglutinin receptor-binding site is dominated by a single heavy-chain complementarity-determining region 3 loop, with minor contacts from heavy-chain complementarity-determining region 1, and is sufficient to achieve nanomolar binding with a minimal footprint. Thus, binding predominantly with a single loop can allow antibodies to target small, conserved functional sites on otherwise hypervariable antigens. The crystal structure of an influenza antibody that recognizes a small, conserved site in the variable receptor-binding domain of HA is described; this antibody shows broad neutralization across multiple subtypes of influenza A virus through an antibody–antigen interaction dominated by a single heavy-chain complementarity-determining region 3 loop. An effective anti-influenza antibody This manuscript reports the identification and structural characterization of a novel anti-influenza antibody, C05, that recognizes a small conserved site in the variable receptor-binding domain of haemagglutinin. The antibody achieves broad neutralization by the insertion of a single loop of the heavy-chain complementarity-determining region 3 into the small conserved site amplified by the avidity of additional binding interactions. This finding highlights loop insertion into the receptor-binding pocket of haemagglutinin as a possible strategy to achieve broad neutralization of influenza by vaccines and therapeutic antibodies.

Since the COVID-19 pandemic onset, the antibody response to SARS-CoV-2 has been extensively characterized. Antibodies to the receptor binding domain (RBD) on the spike protein are frequently encoded by IGHV3-53/3-66 with a short complementarity-determining region (CDR) H3. Germline-encoded sequence motifs in heavy chain CDRs H1 and H2 have a major function, but whether any common motifs are present in CDR H3, which is often critical for binding specificity, is not clear. Here, we identify two public clonotypes of IGHV3-53/3-66 RBD antibodies with a 9-residue CDR H3 that pair with different light chains. Distinct sequence motifs on CDR H3 are present in the two public clonotypes that seem to be related to differential light chain pairing. Additionally, we show that Y58F is a common somatic hypermutation that results in increased binding affinity of IGHV3-53/3-66 RBD antibodies with a short CDR H3. These results advance understanding of the antibody response to SARS-CoV-2. Public antibody clonotypes that recognize SARS-CoV-2 spike protein are important for protection against COVID-19. Here, the authors characterize sequence motifs in the heavy chain complementarity-determining region (CDR) H3s of two public clonotypes and their association with light chain identity.

Natural proteins must both fold into a stable conformation and exert their molecular function. To date, computational design has successfully produced stable and atomically accurate proteins by using so-called “ideal” folds rich in regular secondary structures and almost devoid of loops and destabilizing elements, such as cavities. Molecular function, such as binding and catalysis, however, often demands nonideal features, including large and irregular loops and buried polar interaction networks, which have remained challenging for fold design. Through five design/experiment cycles, we learned principles for designing stable and functional antibody variable fragments (Fvs). Specifically, we (i) used sequence-design constraints derived from antibody multiple-sequence alignments, and (ii) during backbone design, maintained stabilizing interactions observed in natural antibodies between the framework and loops of complementarity-determining regions (CDRs) 1 and 2. Designed Fvs bound their ligands with midnanomolar affinities and were as stable as natural antibodies, despite having >30 mutations from mammalian antibody germlines. Furthermore, crystallographic analysis demonstrated atomic accuracy throughout the framework and in four of six CDRs in one design and atomic accuracy in the entire Fv in another. The principles we learned are general, and can be implemented to design other nonideal folds, generating stable, specific, and precise antibodies and enzymes.

Antibodies capable of neutralizing HIV-1 often target variable regions 1 and 2 (V1V2) of the HIV-1 envelope, but the mechanism of their elicitation has been unclear. Here we define the developmental pathway by which such antibodies are generated and acquire the requisite molecular characteristics for neutralization. Twelve somatically related neutralizing antibodies (CAP256-VRC26.01–12) were isolated from donor CAP256 (from the Centre for the AIDS Programme of Research in South Africa (CAPRISA)); each antibody contained the protruding tyrosine-sulphated, anionic antigen-binding loop (complementarity-determining region (CDR) H3) characteristic of this category of antibodies. Their unmutated ancestor emerged between weeks 30–38 post-infection with a 35-residue CDR H3, and neutralized the virus that superinfected this individual 15 weeks after initial infection. Improved neutralization breadth and potency occurred by week 59 with modest affinity maturation, and was preceded by extensive diversification of the virus population. HIV-1 V1V2-directed neutralizing antibodies can thus develop relatively rapidly through initial selection of B cells with a long CDR H3, and limited subsequent somatic hypermutation. These data provide important insights relevant to HIV-1 vaccine development. A longitudinal study of an individual patient developing neutralizing antibodies against HIV-1 (targeting the V1V2 region of gp120) reveals how such neutralizing antibodies develop and evolve over time, providing important insights relevant to vaccine development. HIV-neutralizing antibody formation examined A better understanding of how HIV-1-neutralizing antibodies are generated could be a useful contribution to the design of improved AIDS vaccines. John Mascola and colleagues have now elucidated the immunological pathway of an important category of HIV-1-neutralizing antibody — those that target the variable V1V2 region of the viral envelope. These antibodies are more frequently elicited than CD4-binding site antibodies in the early stages of HIV infection and feature modest affinity maturation, a process that favours mutations in antibody variable domains that enhance antigen binding.

Antibody-based therapies are at the forefront of modern medicine, addressing diverse challenges across oncology, autoimmune diseases, infectious diseases, and beyond. The ability to design antibodies with enhanced functionality and specificity is critical for advancing next-generation therapeutics. Recent advances in artificial intelligence (AI) have propelled the field of antibody engineering, particularly through inverse folding models for Complementarity-Determining Region (CDR) sequence design. These models aim to generate novel antibody sequences that fold into desired structures with high antigen-binding affinity. However, current evaluation metrics, such as amino acid recovery rates, are limited in their ability to assess the structural and functional accuracy of designed sequences. This study benchmarks state-of-the-art inverse folding models—ProteinMPNN, ESM-IF, LM-Design, and AntiFold—using comprehensive datasets and alternative evaluation metrics like sequence similarity. By systematically analyzing recovery rates, mutation prediction capabilities, and amino acid composition biases, we identify strengths and limitations across models. AntiFold exhibits superior performance in Fab antibody design, whereas LM-Design demonstrates adaptability across diverse antibody types, including VHH antibodies. In contrast, models trained on general protein datasets (e.g., ProteinMPNN and ESM-IF) struggle with antibody-specific nuances. Key insights include the models’ varying reliance on antigen structure and their distinct capabilities in capturing critical residues for antigen binding. Our findings highlight the need for enhanced training datasets, integration of functional data, and refined evaluation metrics to advance antibody design tools. By addressing these challenges, future models can unlock the full potential of AI-driven antibody engineering, paving the way for innovative therapeutic applications.

Monoclonal antibodies are playing an increasing role in both human and animal health. Different strategies of protein and chemical engineering, including humanization techniques of non-human antibodies were applied successfully to optimize clinical performances of antibodies. Despite the emergence of techniques allowing the development of fully human antibodies such as transgenic Xeno-mice, antibody humanization remains a standard procedure for therapeutic antibodies. An important prerequisite for antibody humanization requires standardized numbering methods to define precisely complementary determining regions (CDR), frameworks and residues from the light and heavy chains that affect the binding affinity and/or specificity of the antibody-antigen interaction. The recently generated deep-sequencing data and the increasing number of solved three-dimensional structures of antibodies from human and non-human origins have led to the emergence of numerous databases. However, these different databases use different numbering conventions and CDR definitions. In addition, the large fluctuation of the variable chain lengths, especially in CDR3 of heavy chains (CDRH3), hardly complicates the comparison and analysis of antibody sequences and the identification of the antigen binding residues. This review compares and discusses the different numbering schemes and \"CDR\" definition that were established up to date. Furthermore, it summarizes concepts and strategies used for numbering residues of antibodies and CDR residues identification. Finally, it discusses the importance of specific sets of residues in the binding affinity and/or specificity of immunoglobulins.

The αβ T cell receptor (TCR) repertoire on mature T cells is selected in the thymus, but the basis for thymic selection of MHC-restricted TCRs from a randomly generated pre-selection repertoire is not known. Here we perform comparative repertoire sequence analyses of pre-selection and post-selection TCR from multiple MHC-sufficient and MHC-deficient mouse strains, and find that MHC-restricted and MHC-independent TCRs are primarily distinguished by features in their non-germline CDR3 regions, with many pre-selection CDR3 sequences not compatible with MHC-binding. Thymic selection of MHC-independent TCR is largely unconstrained, but the selection of MHC-specific TCR is restricted by both CDR3 length and specific amino acid usage. MHC-restriction disfavors TCR with CDR3 longer than 13 amino acids, limits positively charged and hydrophobic amino acids in CDR3β, and clonally deletes TCRs with cysteines in their CDR3 peptide-binding regions. Together, these MHC-imposed structural constraints form the basis to shape VDJ recombination sequences into MHC-restricted repertoires. For T cells, functional antigen receptors are selected in the thymus from a pre-selection repertoire by interaction with self MHCs. Here the authors show that specific, non-germline coded features located in the complementarity determining region 3 of the pre-selection antigen receptors are essential for the selection of MHC-restricted repertoire.