Abstract 7319: AI-guided engineering and development of anti-DLL3-targeting T cell engagers (TCEs)

Abstract The aggressive nature of small cell lung cancer (SCLC) and other neuroendocrine cancers (NECs), coupled with their poor prognosis and limited therapeutic options, represent significant challenges in clinical settings. Current first-line treatments, which often include platinum-based chemotherapy combined with programmed cell death protein 1 (PD-1) axis blockade, are associated with short-lived remission and rapid development of resistance, underscoring the critical need for novel therapeutic strategies. Delta-like ligand 3 (DLL3) has emerged as a promising tumor-associated biomarker, with specific overexpression on NECs, particularly SCLC. An anti-DLL3 x CD3 bispecific T-cell engager (TCE), tarlatamab, has demonstrated significant clinical benefits and was recently approved by the US FDA for the treatment of DLL3-expressing SCLC. In this study, we have generated several novel DLL3/CD3-targeting TCEs through the application of our streamlined AI-guided multi-specific development platform. Our AI-driven approach facilitates the engineering and optimization of multiple parameters, including antibody binding epitope, affinity, expression level, stability, and further developability of the candidates in a high-throughput fashion. Our biparatopic TCEs were designed to simultaneously engage two distinct epitopes on DLL3 on tumor cells, and mono-valency binding to CD3 on T cells with moderate to low affinity, in an attempt to enhance T cell-mediated specific cytotoxicity while minimizing potential side effects due to T cell over activation and cytokine production. The affinities and activities of the TCE building-blocks, i.e., the individual anti-DLL3 and CD3 antibodies, and the overall TCE molecules, were fine-tuned through both sequence and molecular format engineering under the guidance of our in-house AI algorithm. Our leading TCE molecule demonstrated a robust DLL3-specific T cell-dependent cellular cytotoxicity towards several tumor cell lines with various DLL3 expression levels, including both cisplatin-sensitive (e.g., NCI-H209, NCI-H526, DMS153) and cisplatin-resistant (e.g., SHP-77) cell lines. The biparatopic DLL3-target TCE also showed significant tumor inhibitory activity in tumor xenograft models including SHP-77, NCI-H82, and DMS-53. Taken together, our data underscore the significant utility of AI-guided platform in antibody/protein engineering, and support further development of our biparatopic anti-DLL3 x CD3 TCE in the treatment of DLL3-expressing SCLC and other NECs. Citation Format: Chuan Chen, Yue Wu, Liang Tian, Chenpeng Su, Zhaohui Chen, Dandan Liu, Jiyuan Tian, Xiaoou Xu, Xiaoqian Chen, Yang He, Yongxin Shang, Jian Peng, Zhenping Zhu. AI-guided engineering and development of anti-DLL3-targeting T cell engagers (TCEs) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 7319.