Antibody-Dependent Cellular Cytotoxicity (ADCC) Prediction

The foundational step in ADCC prediction is the in silico estimation of the binding affinity (KD) between the antibody's constant region (Fc) and the extracellular domains of activating Fc-gamma receptors, primarily FcγRIIIa (CD16a). This involves physics-based simulations or deep learning models that analyze the protein-protein interface at the CH2-CH3 domain hinge. A critical variable is the V158F polymorphism in FcγRIIIa, where a single amino acid change dramatically alters binding affinity and clinical response. Models must accurately capture the energetic contributions of hydrophobic interactions, hydrogen bonds, and salt bridges at the interface to rank-order antibody variants.

The asparagine-linked (N-linked) glycan at position N297 in the CH2 domain is a critical determinant of ADCC potency. The core fucose moiety sterically hinders the carbohydrate-carbohydrate interaction between the antibody and FcγRIIIa. Afucosylated antibodies exhibit up to 100-fold enhanced ADCC. Predictive models must represent the three-dimensional conformation of complex biantennary glycans, specifically quantifying the solvent-accessible surface area occluded by the fucose residue. Techniques include molecular dynamics simulations with explicit glycan force fields and graph neural networks trained on glycan-sequencing data to predict the impact of specific glycoforms on receptor engagement.

ADCC is not governed by a monovalent 1:1 interaction but by a multivalent immune synapse between the effector cell and the opsonized target cell. Effective models move beyond simple affinity to incorporate avidity—the accumulated strength of multiple Fc-FcγR interactions. Key parameters include:

• Antigen density on the target cell surface
• Antibody opsonization level and epitope accessibility
• FcγR clustering and nanoscale spatial organization on the effector cell membrane Agent-based models and spatial stochastic simulations are used to predict the threshold of receptor cross-linking required to trigger intracellular signaling cascades.

Binding of the Fc region to FcγRIIIa triggers an intracellular immunoreceptor tyrosine-based activation motif (ITAM) phosphorylation cascade via the associated CD3ζ or FcεRIγ adaptor proteins. Downstream, this activates Syk kinase, leading to calcium flux and cytotoxic granule exocytosis. Predictive models integrate ordinary differential equations (ODEs) to simulate the kinetics of this signaling network. The key output is the predicted threshold antibody concentration required to trigger degranulation, measured by surrogate markers like CD107a mobilization. These models help distinguish between antibodies that bind strongly but fail to agonize the receptor effectively.

Rational Fc engineering aims to fine-tune ADCC activity by introducing point mutations in the CH2 domain. Computational models predict the effect of these mutations on the conformational dynamics of the Fc loop regions (residues 233-239 and 265-270). Common strategies include:

• S239D/I332E: Increases FcγRIIIa affinity for enhanced ADCC
• L234A/L235A (LALA): Silences FcγR binding to abrogate ADCC
• G236A/S239D/A330L/I332E (GAALIE): Selectively enhances FcγRIIa activation Machine learning models trained on deep mutational scanning data can predict the ADCC phenotype of novel combinatorial mutations, enabling multi-objective optimization alongside pharmacokinetics and thermostability.

ADCC efficacy is a function of both the antibody's effector engagement and the target cell's intrinsic susceptibility to perforin/granzyme B-mediated apoptosis. Predictive models incorporate target cell-specific features:

• MHC Class I expression levels (low expression sensitizes to NK-mediated killing)
• Expression of death receptors (Fas, TRAIL-R)
• Anti-apoptotic protein expression (Bcl-2, XIAP) Integrative models combine antibody binding simulations with transcriptomic signatures of the target cell to predict a composite ADCC sensitivity score. This is particularly relevant for predicting heterogeneous responses across tumor subclones in solid tumors versus hematological malignancies.

The computational estimation of the binding strength between an antibody's Fc region and Fc-gamma receptors on immune effector cells. Critical parameters include:

• Binding kinetics (kon/koff rates) for activating receptors (FcγRIIIa) versus inhibitory receptors (FcγRIIb)
• Allotype dependence: FcγRIIIa V158 vs. F158 polymorphism significantly impacts binding affinity
• pH-dependent binding profiles that influence endosomal trafficking and receptor recycling

The computational engineering of targeted cancer therapies linking a cytotoxic payload to an antibody. ADCC prediction is relevant because:

• The native Fc region of the antibody scaffold may retain effector function capabilities alongside payload delivery
• Bystander killing mechanisms can complement ADCC-mediated tumor clearance
• DAR (drug-to-antibody ratio) optimization must balance payload potency with potential disruption of FcγR binding

A physics-based computational method for simulating the atomic movements of an antibody-FcγR complex over time. Applications in ADCC prediction include:

• Conformational sampling of the Fc glycan to assess its impact on receptor binding geometry
• Binding free energy calculations (MM-GBSA/MM-PBSA) to quantify mutational effects on FcγR affinity
• Allosteric communication analysis between the Fab and Fc domains to identify long-range effects on effector function

A computational framework that simultaneously optimizes an antibody for multiple properties, including ADCC potency. The Pareto frontier approach balances:

• ADCC enhancement via Fc mutations against immunogenicity risk from introduced non-germline residues
• FcγRIIIa affinity against FcRn binding to maintain long half-life
• Effector function against developability metrics like thermal stability and aggregation propensity

The design of antibodies binding two distinct antigens, often with one arm engaging a tumor antigen and the other engaging CD16 (FcγRIIIa) on NK cells. ADCC prediction is central to:

• CD16-binding arm optimization to recruit NK cells without off-target activation
• Heterodimerization strategies (knobs-into-holes) that preserve Fc effector function
• T-cell redirecting bispecifics that complement NK-cell-mediated ADCC with CD8+ T-cell cytotoxicity