Antibody Affinity Maturation by Mutagenesis: How to Make Antibodies "Bind Tighter"?

In antibody drug development, affinity is one of the core parameters determining druggability. The binding strength between an antibody and its target antigen directly modulates drug efficacy, dosage, and safety window. Antibodies generated by the natural immune system generally suffer from insufficient affinity: lead antibodies from hybridoma screening typically exhibit dissociation constants (Kd) in the nanomolar to sub-nanomolar range, while antibodies from initial synthetic immune library screens may have affinities as low as the micromolar level. Most clinical therapeutic antibodies, however, require high-affinity binding at the sub-nanomolar or even picomolar level.

Antibody affinity maturation by mutagenesis is precisely the core technical approach to addressing the affinity deficiencies of lead antibodies. It achieves systematic enhancement of binding capacity through targeted molecular engineering of the antibody complementarity-determining regions (CDRs) and key framework regions, while preserving specificity and developability. This article systematically reviews the mainstream methods, technological evolution, and cutting-edge trends in antibody affinity maturation by mutagenesis.

I. Molecular Basis of Affinity Mutagenesis

Molecular basis & multi-objective constraints for antibody affinity maturation

The binding affinity between an antibody and its antigen is primarily determined by the amino acid sequence and spatial conformation of the CDRs. CDR residues form intermolecular contacts with the antigenic epitope through hydrogen bonds, hydrophobic interactions, electrostatic interactions, and van der Waals forces. Some framework region residues can indirectly influence affinity by maintaining CDR conformation and are therefore also potential targets for engineering. The essence of antibody affinity maturation by mutagenesis is to optimize the spatial complementarity and interaction energy of the binding interface by engineering interface contact residues.

It is worth emphasizing that antibody affinity optimization is not a unidirectional pursuit of "the stronger, the better." Mutagenesis design must balance multiple dimensions, including affinity, specificity, stability, and developability. Some mutations that enhance affinity may be accompanied by negative effects such as decreased thermal stability, increased aggregation risk, and enhanced non-specific binding. Therefore, efficient affinity design requires systematic multi-objective co-optimization.

II. Classical Methods: From Random Trial-and-Error to Structure-Guided Rational Design

Traditional antibody affinity maturation technologies have gradually evolved from early random screening to structure-guided rational design.

Site-directed mutagenesis is the most fundamental rational engineering approach. Relying on antibody-antigen complex crystal structures, researchers identify residues in the CDRs that form critical contacts with the antigen and introduce targeted substitutions to enhance interactions. In a study on EGFR×PD-L1 bispecific antibodies, a single point mutation in the CDR3 region achieved a dozens-fold improvement in affinity. Another study, using computational-assisted rational design, restored the binding of pertuzumab to the HER2 S310F mutant antigen with just two amino acid changes.

Saturation mutagenesis further expands the exploration space by systematically introducing all 20 natural amino acids at key CDR positions to screen for optimal residue combinations.

Directed evolution constructs large-capacity random mutagenesis libraries through error-prone PCR and DNA shuffling, combined with high-throughput screening using phage display or yeast display. In 2026, a domestic university team developed the high-frequency antibody editor HAE1, which combines gene editing with antibody evolution to achieve targeted and precise mutagenesis of antibody variable regions, successfully accomplishing affinity maturation and neutralization activity recovery of neutralizing antibodies against SARS-CoV-2 escape mutants.

CDR grafting is a classic strategy for antibody humanization, typically used in conjunction with affinity optimization: the CDR sequences of a functional antibody from a heterologous source are grafted onto a human framework region with higher stability and lower immunogenicity, followed by framework back-mutations and fine CDR mutations to restore or even enhance affinity, achieving coordinated optimization of specificity, stability, low immunogenicity, and affinity.

Technical evolution of antibody affinity-maturation design

III. Computational-Aided Design: From Experience-Driven to Data-Driven

With advances in structural biology and computational power, computational-aided design has gradually become a mainstream approach for affinity maturation optimization.

Structure-based binding free energy calculation is the most fundamental strategy. By resolving or predicting antibody-antigen complex structures, molecular mechanics force fields are used to calculate changes in binding free energy upon mutagenesis to screen for beneficial mutations. The LAffAb strategy systematically introduces combinatorial mutations to optimize binding energy based on antibody-antigen crystal structures; a combinatorial library of 7,000 variants (containing up to 9 mutations) can achieve up to 1,000-fold affinity enhancement, and in a therapeutic antibody test case, achieved a 30-fold affinity improvement while also improving antibody stability.

Molecular dynamics (MD) simulations can capture dynamic conformational changes at the binding interface, identifying hotspot residues that are difficult to discover in static structures. In an affinity maturation study of an scFv targeting Neisseria meningitidis fHbp, researchers integrated MD simulations to guide mutagenesis design, significantly improving engineering efficiency.

RosettaAntibodyDesign (RAbD) is a computational toolkit specifically designed for antibody engineering that systematically searches the sequence space of CDR regions to predict optimal mutation combinations. Since its publication in 2018, this tool has become a classic method in the field of computational antibody design.

IV. AI-Driven Design: From "Computational-Aided" to "Intelligent Generation"

AI-driven paradigm for antibody affinity mutation design

Deep learning and generative AI are fundamentally reshaping the paradigm of antibody affinity maturation by mutagenesis.

Deep learning-based prediction of mutation effects is the most direct application direction. A 2025 study published in Briefings in Bioinformatics constructed an antibody optimization system driven by the synergy of deep learning and computational biology; on specific test sets, deep learning models designed for this purpose performed excellently in predicting the affinity effects of single-point antibody mutations.

Generative AI models are achieving breakthroughs in "designing the unknown." AffinityFlow, published at ICML 2025, proposed a sequence-only affinity maturation scheme that performs alternating optimization using only antibody and antigen sequences. Specialized antibody sequence generators based on inverse folding models, such as ESM-IF, can efficiently generate novel sequences with high antigen-binding activity through preference optimization training.

Ab-SELDON is a modular, automated computational pipeline for antibody design that iteratively optimizes interactions across five different modification steps, including CDR graftingg, framework grafting, and mutagenesis.

IgDesign is a deep learning model focused on antibody CDR design that has successfully generated binders validated by in vitro experiments for eight therapeutic targets.

Explainable AI is an important frontier direction. Researchers are exploring the combination of deep learning with physics-based mutagenesis analysis to not only predict the effects of mutations on affinity and stability but also explain "why" a particular mutation is effective.

V. Next-Generation Technology Platforms

With the convergence of gene editing, synthetic biology, and AI, a new generation of high-throughput antibody affinity maturation platforms has emerged.

High-frequency antibody editor HAE1 (Peking University, 2026) integrates the PAM-less Cas9 variant SpRY with cytidine deaminase and adenine deaminase to achieve targeted and precise mutagenesis of antibody variable regions, broadening the pathways of antibody evolution.

SAMPLER is a directed evolution strategy that rapidly enhances affinity by systematically introducing mutations into CDRs. In the optimization of MERS-CoV neutralizing antibodies, it successfully selected variants with both potent neutralization and broad-spectrum activity.

The OrthoRep continuous hypermutation system enables sustained autonomous mutagenesis of target genes, driving rapid affinity maturation of yeast-displayed antibodies. Integrated studies of computational design and OrthoRep are leveraging evolutionary trajectories to elucidate the molecular mechanisms of affinity maturation.

At the level of synergy between structural biology and AI, researchers have systematically engineered PD-1 antibodies through computational saturation mutagenesis, AlphaFold prediction, and MD simulations, enhancing affinity by reshaping the binding interface and introducing additional salt bridges and hydrophobic interactions.

VI. Industry Deployment: MatwingsVenus™ (Xiaowu™) by Matwings Technology

The industrialization of antibody affinity maturation by mutagenesis is accelerating. MatwingsVenus™ (Xiaowu™), a conversational protein R&D agent independently developed by Matwings Technology, has integrated AI-driven antibody optimization capabilities into a unified platform.

In July 2026, MatwingsVenus™ (Xiaowu™) stood out from hundreds of exhibited products to be selected for the World Artificial Intelligence Conference (WAIC) "Treasure of the Hall" list, as the sole AI for Science product to receive this honor.

In antibody affinity optimization scenarios, the platform leverages billion-scale real labeled protein data and over 200 protein design tools to directly predict the effects of mutations on affinity from antibody sequences, completing mutagenesis design, performance prediction, and candidate screening in one stop. The platform integrates two core capabilities—AI-directed evolution and AI-driven enzyme mining—which complement each other in antibody optimization: AI-directed evolution can rapidly predict beneficial mutation combinations based on limited data, while AI-driven enzyme mining can discover novel antibody scaffolds from massive datasets.

The platform features a natural language conversational interface, enabling researchers to complete the entire workflow from sequence analysis to affinity optimization solution design without requiring programming or bioinformatics backgrounds. According to public reports, Xiaowu™ can compress the traditional 2–5 year protein R&D cycle to 2–6 months, reduce experimental samples from tens of thousands to hundreds, and increase the success rate from 5% to 30%.

VII. Conclusion

Antibody affinity maturation by mutagenesis is undergoing a profound transformation from empirical trial-and-error to intelligent design.

From the "point-to-point" engineering of site-directed mutagenesis, to the "random exploration" of saturation mutagenesis and directed evolution, and then to the "precise prediction and generation" driven by AI—each evolution signifies higher efficiency, lower costs, and greater design space. Traditional methods excel at "engineering the known"—local optimization based on existing antibody structures; AI methods are achieving "designing the unknown"—not only telling us "how to modify" but also predicting "where to modify" and "what to modify into."

The future of antibody affinity design will not only pursue higher affinity but also achieve precise multi-objective co-optimization across affinity, specificity, stability, and developability. With the continued integration of protein language models, generative AI, and structural biology, antibody affinity maturation by mutagenesis will provide an increasingly powerful technological foundation for the rapid development of high-affinity therapeutic antibodies.