Non-Cognate Cytotoxic T Cells Can Be Retargeted Against CD20 Positive Tumour Cells Using [Fab' × MHC Class I/Peptide] Conjugates.

Abstract 2838

To circumvent cytotoxic T lymphocyte (CTL) tolerance of tumour-associated antigens, the concept of redirecting CTLs against non-cognate targets has developed. One way of doing this is to use bispecific antibodies comprising anti-CD3 and anti-tumour antigen moieties. Unfortunately, this is frequently associated with unacceptable toxicity due to inflammatory cytokine release. As an alternative our approach has been to use a bivalent conjugate recognising a tumour antigen (through an antibody fragment) and a defined population of CTLs (specific for a single antigenic peptide e.g. viral epitope) through peptide presented in the context of recombinant MHC class I. We have produced a conjugate consisting of an anti-human CD20 Fab' fragment joined via a chemical crosslinker (succinimidyl-4-(N-maleimidomethyl) cyclohexane-1-carboxylate) to murine MHC class I/peptide (K^bα₁-α₃ domains/β₂microglobulin presenting the ovalbumin-derived peptide SIINFEKL; expressed bacterially as a continuous polypeptide single chain trimer after Yu et al, J Immunol 2002). Size exclusion chromatography allowed purification of conjugates with [Fab':MHC class I/peptide] ratios of 1:1 and 2:1 (F2 and F3 respectively). In vitro both constructs were able to redirect the transgenic murine CTL line OT-1 (specific for K^bSIINFEKL) to lyse human CD20⁺ tumour cells (lymphoblastoid Daudi cell line) at effector: target ratios of 10:1. This lysis could be blocked by the addition of 100 fold excess of either anti-CD20 F(ab')₂ or the K^b/SIINFEKL-specific antibody 25D1. The constructs were also able to cause in vitro proliferation of naïve OT-1 cells (but not irrelevant CD8⁺ T cells) in the presence of human CD20⁺ cells in both thymidine incorporation and CFSE dilution assays. Using a human CD20 transgenic mouse model (Ahuja et al, J Immunol 2007) we have evaluated both constructs in vivo for their ability to redirect adoptively transferred OT-1 cells to deplete B cells from the peripheral blood. A single dose of 1 nmole F3 and 2 nmole F2 caused respectively up to 95% and 85% B cell depletion at day 7. The efficacy of lower doses suggested a dose: response relationship. As a marker of toxicity, we have measured cytokine levels at 2, 8 and 24 hours following a dose of 1 nmole F3 and compared them to those seen after administration of an [anti-CD3 × anti-CD20] bispecific F(ab')₂ at a dose (0.5 nmole) which produced similar day 7 peripheral blood B cell depletion: phosphate-buffered saline was given as a negative control. Maximal cytokine release was seen at 2 hours with the levels of IL-4, IL-5, KC, IL-2 and IL-10 being lower after administration of the F3 than after the bispecific F(ab')₂. However, interestingly, the F3 resulted in greater IL-12 release. Overall these data suggest that [Fab' × MHC class I/peptide] constructs have the potential to redirect non-cognate CTLs to deplete CD20⁺ malignant B cells from the peripheral blood and that this is associated with a lower level of cytokine release than a similarly efficacious dose of an anti-CD3-containing bispecific F(ab')₂. Furthermore, the ability of [Fab' × MHC class I/peptide] constructs to cause proliferation of OT-1 cells in vitro suggests it may be possible to use a single molecule to both generate a secondary cytotoxic T cell response and subsequently to retarget it, increasing the viability of the approach if adopted in the clinic.