Depletion of the helper/inducer (memory) T cell subset using a bispecific antibody-toxin conjugate directed against CD4 and CD29.

We have developed a bispecific antibody that recognizes the CD4 and CD29 antigens simultaneously and that was examined for its ability to target CD4+CD29bright T cells. The premise for using bispecific antibody-toxin conjugates was that monovalent binding to one antigen would not result in internalization while binding through both antigens would predispose toward endocytosis and delivery of the toxin to the cell interior. In this study, we show that the bispecific antibody binds monovalently to CD4 and CD29 and bivalently to both antigens. Both monovalent and bivalent binding rendered the target cells sensitive to complement-mediated lysis, demonstrating that this effector modality cannot take advantage of the dual specificity of the antibody. Bivalent binding, however, allowed modulation of more than 60% of the bound antibody off the surface of CD4+CD29bright cells, which we further exploited to deliver a toxin moiety preferentially to CD4+CD29bright cells. Immunoconjugates incorporating blocked ricin (a ricin holotoxin that has its intrinsic galactose-binding sites blocked by chemically linked affinity ligands) preferentially killed CD4+CD29bright cells in vitro by a factor of 25 in comparison with killing of total CD4+ cells in functional assays. Assays on resting PBL demonstrated a similar specificity of the bispecific immunotoxin for CD4+CD29bright cells with a concomitant relative survival of CD4+CD29dim (CD45RA+). We demonstrated that the potency of the immunotoxin is not only a function of its affinity but also of its propensity to internalize, and that this property is influenced by the degree of bivalent binding. These results open up the possibility of engineering bispecific antibody toxin conjugates for use as therapeutic immunotoxins for selective removal of restricted T cell subsets.