A protein proximity map guided the selection of two targets for a single antibody
A protein proximity map guided the selection of two targets for a single antibody
On September 9, 2026, Nature published a study describing a method for choosing two targets for a single antitumor molecule on the basis of their proximity on the tumor cell surface. The authors selected the EGFR–CDCP1 pair and tested, in cells and in mice, an antibody that binds both proteins and carries an attached cytotoxic payload.
An antibody carrying a cytotoxic payload must enter the tumor cell. Conventionally, such an antibody recognizes a single protein on the cell surface. EGFR is a receptor through which the cell receives growth signals; it is suitable for targeting, but it is also present in healthy tissues. The authors looked for a second protein that consistently sits near EGFR in tumor cell line membranes, reasoning that this co-localization could sharpen payload delivery.
To build the proximity landscape, the authors directed photosensitive labels at 12 surface receptors. These labels deposited a chemical trace on nearby proteins, and mass spectrometry read that trace. In this way the researchers compiled 248 proximity maps across 28 tumor cell lines. Each map records which proteins regularly appear close together within a local patch of the membrane.
The MetaMap algorithm compared these maps and flagged CDCP1, another surface protein, as a candidate partner for EGFR. In tumor cell lines, the two proteins were repeatedly found in close proximity; in normal epithelial and mesenchymal cells that expressed both proteins, the proximity signal was weaker. The authors corroborated this result with reciprocal maps in which the neighborhoods of EGFR and CDCP1 were labeled in turn, with a separate series of 38 CDCP1 maps in 19 tumor cell lines, and with co-immunoprecipitation of proteins pulled down together with EGFR. Proximity strength varied independently of the abundance of each protein on the cell.
The researchers then assembled a bispecific antibody in which one arm binds EGFR and the other binds CDCP1, while the attached payload kills the cell once internalized. They deliberately made the EGFR arm weaker than the CDCP1 arm, expecting this asymmetry to favor simultaneous engagement of both neighboring receptors. In cell-based assays the bispecific antibody internalized into tumor cells more efficiently than antibodies targeting only one protein; the stronger internalization is consistent with concurrent engagement of EGFR and CDCP1. In primary epithelial cells the bispecific antibody internalized far less efficiently.
In a xenograft model using SW48 tumor cells in mice, a 1 mg/kg dose of the bispecific antibody produced complete tumor growth inhibition, whereas the EGFR-only antibody achieved 43% inhibition. Each group contained eight mice. In a separate experiment every mouse carried two tumors: a standard SW48 xenograft and a variant with reduced CDCP1 expression. After a 3 mg/kg dose, growth inhibition reached 97% in the standard tumor and 46% in the CDCP1-reduced tumor. The difference between 97% and 46% supports a role for CDCP1 in payload delivery in this model.