To reduce on-target, off-tumor risk in CAR-T, first determine whether the target is present on essential healthy tissues, then select a binder and receptor design that can distinguish tumor cells from those tissues under realistic conditions. A nanobody with strong binding is not automatically a safer choice: affinity, epitope, antigen density, receptor architecture, and cell-killing thresholds all affect risk. We recommend combining human- or species-relevant tissue expression review with quantitative binding and functional testing before advancing a candidate.
Click here to get more.
This guide is for translational and preclinical teams assessing CAR-T targets and nanobody discovery strategies, including veterinary medicine projects. It outlines practical decision points, assay options, common mistakes, and questions to discuss with a discovery partner. No single assay can establish clinical safety, so findings should be interpreted together and reviewed by qualified research and clinical teams.
On-target, off-tumor activity occurs when a CAR recognizes the intended antigen on healthy cells as well as tumor cells. The biological risk depends on which tissues express the antigen, how much is present, and whether CAR engagement can trigger damaging effector activity. Off-target binding, by contrast, involves unintended recognition of a different molecule, which requires its own specificity assessment.
For that reason, a tumor-positive result is not enough to establish that a target is suitable. Teams need to ask whether healthy cells display the same accessible epitope, whether expression varies among tissues or donors, and whether low-level expression can still activate the CAR. These questions should guide target selection and assay design before a candidate binder is treated as a lead.
Start with available expression evidence, then assess whether it reflects the relevant cell types, disease context, and species. Transcript data can help identify where a target may be expressed, but RNA abundance alone does not establish the amount of accessible protein on the cell surface. Where feasible, verify surface expression using validated protein-level methods and appropriate positive and negative controls.
Include healthy tissues that could be clinically important, not only convenient cell lines. When primary cells or tissue-derived models are used, document sample source, handling, and assay limitations. In veterinary research, check whether the target sequence and expression pattern are appropriate for the animal species under study; human-target data should not be assumed to apply directly.
Measure surface antigen levels on tumor and healthy-cell models with a method that allows meaningful comparisons. Report the method, controls, and units used, such as calibrated molecules per cell when the assay supports that interpretation. If absolute quantification is not validated, label results as relative rather than presenting them as a precise density measurement.
Use a panel of models when possible because one cell line may not represent the range found in tumors or healthy tissues. Compare antigen levels with CAR activation and killing readouts, rather than treating expression as a stand-alone safety threshold. There is no universal antigen-density cutoff that guarantees safety across targets, constructs, and assay systems.
Evaluate whether each nanobody recognizes the intended target and whether binding depends on an epitope accessible on intact cells. Include unrelated proteins or cells lacking the target to help identify nonspecific signal. For a veterinary program, test binding against the relevant species’ target and consider whether a cross-reactive binder is desirable or could complicate interpretation.
Goto NBbiolab to know more.
Affinity is one selection factor, not a safety score. A high-affinity binder may improve recognition of low-antigen tumor cells, but it could also recognize low-level antigen on healthy cells; the outcome depends on the complete receptor and cellular context. Compare candidates across a range of target expression and, where appropriate, evaluate whether differences in epitope or binding kinetics change functional responses.
Binding assays show recognition, while functional assays help determine whether recognition activates the engineered cells. Assess relevant outputs, such as activation, cytokine release, and target-cell killing, using matched tumor and healthy-cell models where practical. Include controls that distinguish antigen-dependent activity from background activation.
For example, a team might use a binding assay at 4°C to limit active internalization during the measurement, then evaluate cellular function at 37°C under conditions closer to physiological temperature. A proposed initial coculture readout at 24–48 hours can help capture early activity, but it is a starting point rather than a universal standard; assay duration should be justified for the model and endpoint. These are experimental design examples, not guarantees of safety or efficacy.
If a binder recognizes both tumor and healthy cells, consider whether receptor design or activation tuning can improve discrimination without losing the desired tumor response. Options to investigate may include alternative binder candidates, changes to receptor architecture, or strategies that require additional tumor-associated signals. Each option introduces its own design and validation needs, so changes should be tested rather than assumed to reduce risk.
Define go/no-go criteria before comparing candidates. Criteria can include a minimum functional response in tumor models, acceptable background in healthy-cell models, and reproducibility across relevant donors or batches. A candidate should not advance solely because it ranks first in an affinity screen.
When evaluating a supplier, ask how the discovery plan will address your target’s biology rather than focusing only on library size or screening speed. Useful discussion points include target construct design, screening against native cell-surface antigen, specificity counterscreens, and the plan for confirming binding on cells. Also clarify what sequence, assay, and material data will be delivered at each project stage.
For CAR-T development, ask whether the workflow can connect binder discovery to functional testing, or whether those activities require separate partners. Agree on species requirements, sample availability, acceptance criteria, and how inconclusive results will be reported. NBbiolab can discuss project requirements for CAR-T target nanobody discovery and help teams define a suitable scope; specific methods, deliverables, timelines, and feasibility should be confirmed for each inquiry.
The most reliable way to reduce on-target, off-tumor binding risk is to evaluate the target, binder, and CAR function as one connected system. Begin by mapping healthy-tissue expression, comparing antigen levels in relevant cell models, and screening nanobodies for cell-surface specificity and functional selectivity. Then test the leading candidates in species-appropriate tumor and normal-cell models before making advancement decisions.
For your next project review, prepare the target identity, intended species, available tumor and normal-cell models, and the evidence you already have on tissue expression. Use that information to agree on screening controls, quantitative readouts, and explicit advancement criteria with your internal team or discovery supplier. This approach will not eliminate uncertainty, but it makes the key risks visible and supports more defensible candidate selection.
Are you interested in learning more about How to Reduce On-Target Off-Tumor Binding Risk in CAR-T? Contact us today to secure an expert consultation!