A humanised mouse model of cytokine release: comparison of CD3-specific antibody fragments

Abstract

CD3-specific antibodies have shown clinical efficacy in both transplantation and autoimmunity. However, targeting CD3 in this way can lead to T-cell activation and a serious cytokine release syndrome mediated by Fcγ receptor binding. An in vivo mouse model has been developed using severe combined immunodeficient (SCID) mice to detect human T-cell depletion and cytokine release into the circulation after administration of OKT3. This system has been used to evaluate OKT3 antibody fragments lacking the entire Fc region alongside whole antibody constructs. These data clearly show that cytokine release is detected with all OKT3 antibody constructs and fragments tested and these can be ranked from highest to lowest as follows: mIgG2a>hIgG1 (Ala-Ala)>hIgG1 diFab' maleimide (DFM)>hIgG1 F(ab')₂>mIgG2a F(ab')₂>hIgG1 Fab'. Furthermore, the monovalent hIgG1 Fab' fragment gives the least cytokine release but it does not deplete human T-cells in this assay format. This suggests that T-cell activation may be playing a role in the mechanism of action of anti-CD3 antibodies and consequently the unwanted cytokine release is potentially unavoidable for this class of molecules. This model system provides a useful tool to aid in understanding and reducing the potential risks of cytokine release following antibody therapy.