We have developed a bispecific antibody that recognizes the CD4 and CD26 antigens simultaneously and that was examined for its ability to target CD4+CD26+T cells. These latter cells constitute the activated component of the CD4+ CD29highCD45RO+ memory T-cell subset that provides help for B-cell Ig synthesis and help for responses against recall antigens. The purified bispecific antibody exhibited an estimated dissociation constant (kd) of 2.4 x 10(-9) mol/L, on comparison with 1.1 x 10(-9) mol/L for anti-CD26, and 1.6 x 10(-10) mol/L for anti-CD4. Surface plasmon resonance was used to show the bifunctional capacity of the antibody. On binding 125I-bispecific antibody to phytohemagglutinin (PHA)-activated T cells, 54.4% of the bound antibody was internalized. This was the result of bispecific binding, because monovalent fragments of anti-CD4 and anti-CD26 were not able to modulate antigen or induce internalization using both a fluorescent assay and an 125I-internalization assay. The ability of the bispecific antibody to be internalized was used to deliver a toxin, blocked ricin, specifically to cells that are CD4+CD26+. The inability of monovalent fragments to be internalized formed the basis for our hypothesis that monovalent binding by the bispecific immunotoxin would not result in internalization. Against resting E+ T cells, the bispecific immunotoxin developed a minimal effect. On preactivating the same cells, using phorbol myristate acetate (PMA)/ionomycin on concanavalin A (ConA) or especially PHA, levels of CD26 were upregulated and the immunotoxin effectively inhibited the ability to provide help for B-cell Ig synthesis while leaving intact the CD4-CD26+ and CD4+CD26- populations; an effect observed both functionally and by phenotype. The bispecific antibody proved to be most effective at inhibiting a heterologous mixed leukocyte reaction. We propose that this reagent may form the basis for the rational design of toxins designed to modulate activated T cells from, or directed against, tissue grafts.