Yile Dai, Joseph Brouillard, Jillian R. Jaycox, Sung M. Yeon, John D. Huck, Soumya S. Yandamuri, Zhe Zhong, Kevin C. O’Connor, Irina Burnina, Cameron Henkel, Allie K. LeMay, Asparouh Lilov, Elizabeth McGurk, Morgan Morrill, Liz Parker, Tricia Sackett, Chanita Sandberg, Arvind Sivasubramanian, James C. Geoghegan, Aaron M. Ring.  

Structure 34, 1–11. https://doi.org/10.1016/j.str.2026.02.012 

Overview 

Therapeutic antibodies are widely regarded as being highly specific, but there is an increasing body of evidence suggesting that specific off-target interactions occur and may pose risks for safety and efficacy. This study set out to systematically assess the prevalence of off-target reactivity in a set of approved and clinical-stage monoclonal antibodies (mAbs) using rapid extracellular antigen profiling (REAP), and to determine whether any identified off-target binding could be eliminated through antibody engineering. Overall, the study reinforces the importance of early-stage specificity screening, particularly for novel therapeutic modalities where weak off-target reactivity can lead to enhanced biological effects. 

Key hypotheses and objectives 

Hypothesis: Therapeutic antibodies may exhibit specific off-target reactivity, which might pose risks for safety, efficacy, and pharmacokinetics. 

Objectives: 

• Evaluate specific off-target reactivity in approved and clinical-stage antibodies. 

• Investigate the molecular basis and functional effects of these interactions. 

• Assess the utility of a novel platform for early-stage specificity screening. 

• Determine whether off-target reactivity can be eliminated through engineering. 

Approach and techniques 

The prevalence of off-target reactivity in a set of 174 approved and clinical-stage mAbs was systematically assessed using REAP. REAP is a yeast display-based screening system, developed by the Ring lab, that enables conformational epitope interaction analysis against >6000 human extracellular proteins and protein domains.  

Key aspects of the approach include: 

• Proteome-scale screening to candidate off-target interactions. 

• Employing orthogonal techniques such as Octet BLI binding assays to validate the identified off-target binding. 

• Epitope mapping to determine the binding site of the antibodies on the off-target proteins.  

• Investigating the potential of the off-target binding to interfere with natural receptor-ligand binding. 

• Applying antibody engineering to make minimal changes within binding regions and selectively screening to retain on-target activity while eliminating off-target interactions. 

• Using established developability tests to confirm that the engineered antibodies maintain favorable biophysical properties. 

Major findings and impact 

• Prevalence: 28% of the 174 approved and clinical-stage mAbs were found to have at least one off-target interaction. 

• Validation: A select set of nine predicted off-target interactions were validated via Octet BLI binding of the respective IgGs to recombinant antigen. 

• Functional relevance: In one instance, the off-target binding site was found to overlap with the natural receptor-ligand binding interface, indicating the potential to disrupt normal biological signaling. 

• Mechanisms: Antibody cross-reactivity could be potentially explained by epitope structural similarity in the case of homologous target/off-target pairs, and off-target epitope physicochemical properties (charge, hydrophobicity) in unrelated pairs.   

• Engineering resolution: Minimal changes within the complementarity-determining regions were effective in eliminating off-target binding. Single amino acid substitutions removed off-target effects while retaining, and in some cases improving, binding to the target antigen. 

• Correlation with standard assays: Non-specific polyreactivity, as measured by the PSR score, had a modest ability to predict off-target binding. However, several cases of clean PSR antibodies with validated off-target binding point to the need for a dedicated assay. 

Implications for the development of therapeutic antibodies 

Specific off-target binding is an understudied but resolvable problem in the development of therapeutic antibodies. The discovery rate of off-target interactions contradicts the prevailing understanding that antibodies in the clinic are generally specific, especially when the evidence for specificity is incomplete or indirect. New classes of therapeutics, including bispecific antibodies, CAR-T cell therapies, and antibody-drug conjugates, may be particularly affected, as even low-affinity off-target binding may initiate strong biological activities, including unintended cell death or activation of biological pathways. However, the extent to which this may pose a clinical risk depends on binding affinity, epitope location, and tissue expression of the off-target protein. 

The findings point to several considerations for development strategy: 

• Broader assessment of specificity: Assessing off-target binding at the proteome scale during the discovery phase allows for the detection of problems while the drug is still modifiable. 

• Engineering as a solution: Off-target binding does not necessarily exclude an antibody from development, as these interactions may be overcome through modifications to the drug. 

• Specificity-based candidate selection: Where there are several options, specificity may be used as an additional selection criterion, in addition to affinity and developability. 

• Cautious integration of computational design: Though in silico methods are improving, they are not yet able to accurately predict specificity. Empirical screening must be carried out, especially in the absence of natural tolerance in de novo or AI-designed antibodies.