Characterization of anti-CD3 antibodies in clinically available bispecific T cell engagers.

Abstract

Bispecific T cell engagers (bispecific TCEs) are engineered antibodies that redirect T cells to mediate tumor cell killing by simultaneously binding to CD3 on T cells and tumor-associated antigens. As of July 2025, ten bispecific TCEs are clinically available. The CD3-binding antibodies in these bispecific TCEs can be classified into 6 groups based on the amino acid sequence similarity across their 6 complementarity-determining regions (CDRs). Specifically, antibodies were assigned to the same family if their six CDRs-HCDR1-3 and LCDR1-3-exhibited ≥80% pairwise sequence identity upon multiple sequence alignment. Family 1, derived from OKT3-a mouse hybridoma generated by immunizing BALB/c mice with human T cells-includes only blinatumomab; Family 2, derived from SP34-a rhesus monkey (Macaca mulatta) derived hybridoma specific for human T cells-comprises 5 antibodies; and Family 6, derived from UCHT1-a mouse hybridoma generated by immunizing mice with human T cells-contains only tebentafusp. The origin of the remaining 3 antibodies has not been disclosed and they possess unique CD3-binding sequences. We classified them into their own distinct families (Families 3, 4, and 5). Interestingly, mosunetuzumab (Family 4) showed remarkably lower incidence of adverse events such as cytokine release syndrome (CRS), immune effector cell-associated neurotoxicity syndrome (ICANS), and infection compared to other bispecific TCEs even though its affinity for CD3ε was not significantly different. The epitopes of 4 antibodies in Family 2, teclistamab, talquetamab, glofitamab, and tarlatamab were previously defined to be located at the N-terminal region of CD3ε via hydrogen-deuterium exchange mass spectrometry (HDX-MS) analysis. In our in silico epitope prediction analysis, the N-terminal region was included in the epitope region of all bispecific TCEs regardless of their family. Blinatumomab (Family 1) and tebentafusp (Family 6) did not bind to the CD3ε homolog of the cynomolgus monkey, whereas the other 8 bispecific TCEs did. This lack of cross-reactivity poses clear disadvantages in their preclinical development, particularly for toxicity and safety evaluation in nonhuman primate models.