Document resource
Background The efficacy of T cell engager (TCE) therapy for AML and other hematologic malignancies is often limited by high disease burden, non-tumor selective target expression, and the low frequency and exhausted phenotype of T cells. Novel approaches, including simultaneous engagement of CD3 (signal 1) and co-stimulatory molecules such as CD28 (signal 2) on T cells, are being explored to improve clinical response rates. We utilized a previously reported TCR mimetic (TCRm) antibody that potently and specifically recognizes the cathepsin G-derived peptide CG1 presented by HLA-A*02:01 to generate a tri-specific TCE that simultaneously binds CG1/peptide-HLA (pHLA) on leukemic cells and CD3 and CD28 on T cells. The CG1 peptide is selectively presented on AML blasts and leukemic stem cells, with minimal detection on healthy hematopoietic cells, making it an attractive immunotherapeutic target. CD28 is constitutively expressed on most T cells and has a well-documented role in enhancing T cell activity.Methods A panel of proprietary CD28-targeting antibodies was engineered to produce variants with different affinities and incorporated as monovalent binders into tri-specific CG1xCD3xCD28 TCEs. These were screened for enhanced performance versus the bi-specific CG1xCD3 TCE without increasing T cell binding avidity or inducing off-target cytotoxicity. In vitro assays measured cytokine secretion, T cell activation, T cell mediated cytotoxicity and other parameters. Specificity was assessed using Ala/Gly scanning and testing of computationally predicted cross-reactive peptides. In vivo efficacy was evaluated in an aggressive humanized AML xenograft (CDX) model. Off-target effects were further assessed ex vivo using bone marrow colony-forming unit (CFU) assays and testing on primary cells from healthy tissues.Results Compared to the bi-specific CG1xCD3 TCE, the CG1xCD3xCD28 tri-specific TCE induced robust IL-2 secretion and enhanced cytotoxicity against multiple AML cell lines in vitro at low E:T ratios and low pHLA copy numbers. Repeated tumor challenge experiments in vitro revealed improved ability of the tri-specific CG1xCD3xCD28TCE to mediate serial killing of leukemic cells. In vivo, the tri-specific TCE achieved potent tumor control in an aggressive AML CDX model with no observed adverse events. Importantly, specificity for CG1 pHLA was maintained as indicated by Ala/Gly scan and testing of computationally-predicted cross-reactive peptides. No meaningful cytotoxicity was detected in bone marrow progenitor cells in CFU assays or against normal primary cells from vital tissues.Conclusions These findings support the notion that leveraging CD28 co-stimulation could overcome limitations of conventional bispecific TCEs, particularly in AML patients with high disease burden, thus supporting clinical exploration of this approach.