Abstract 1850: A novel anti-CD137 antibody recognizing the membrane-proximal CD137 domain elicits potent anti-tumor T cell activity in a bispecific antibody format

Abstract

The activation of CD137, also known as 4-1BB, has been identified as a promising strategy for next-generation immune therapeutics. However, cancer therapies based on activating antibodies to CD137 were discontinued by adverse events such as liver toxicity. To overcome those limitations, next generation antibody therapeutics using bispecific antibody approach has been developed by adjusting affinity, epitope, and valency. In this study, the anti-CD137 antibody with the binding epitope in the membrane-proximal region did not have CD137 cross-linking or T cell activating activity alone. However, when the anti-CD137 scFv was linked into a bispecific antibody with a tumor-associated antigen (TAA) targeting antibody, the anti-CD137 antibody strongly augmented T cell activation. We mined the anti-CD137 antibody, 1A10, which induces clustering-dependent agonism using screening of a phage library. 1A10 bound to the CD137 with a high affinity and induced signaling only in the presence of the Fc cross-linking anti-human IgG antibody. Various TAAx1A10 bispecific antibodies were generated by introducing 1A10 scFv into human IgGs targeting TAA. When the 1A10 was linked to TAA-specific antibodies in bispecific antibody format, 1A10 induced potent T cell activation and tumor-killing in a TAA dependent manner according to CD137 bioassays and PBMC based assays. To evaluate the competition on CD137 binding, we did epitope binning analysis with other anti-CD137 antibodies and CD137L. 1A10 binding inhibited further binding of utomilumab to CD137, but not the binding of urelumab or CD137L to CD137, suggesting that the 1A10 binding site is different from urelumab or CD137L. Further detailed epitope mapping study using a library of surface arginine or glutamic acid mutations on the CD137 via high-throughput FACS and NGS analysis revealed the binding sites of 1A10 mapped to CRD4. The site did not overlap with CD137L binding region, which explains why 1A10 did not compete with CD137L. The epitope site of 1A10 is also quite distinct from the urelumab binding site, which locates on the apical region of CD137. The 1A10 epitope site is partially overlapping with utomilumab binding sites on the CRD4 but does not bind to the CRD3, where utomilumab binds and competes with CD137L. In summary, 1A10, the anti-CD137 antibody with a unique epitope exhibited clustering dependent CD137 activation and potent anti-tumor activity via tumor-specific activation of CD137 in bispecific antibody format. Citation Format: Yangsoon Lee, Suyoun Lee, Yeunju Kim, Hyejin Chung, Kyungjin Park, Eunyoung Park, Kyeong-Su Park, Jinwon Jung, Byungje Sung, Jonghwa Won. A novel anti-CD137 antibody recognizing the membrane-proximal CD137 domain elicits potent anti-tumor T cell activity in a bispecific antibody format [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 1850.

The emerging landscape of novel 4-1BB (CD137) agonistic drugs for cancer immunotherapy

“…Most molecules still display an antibody framework or include antibody-derived binding domains, such as scFv (ABL503/TJ-L14B, ABL111/Tj-CD4B, ABL105/YH32367, ATG-101/YN051, LBL-024), 64 , 65 , 71 , 77 , 78 VH (CB307), 67 VH/VL (ND021/NM21-1480), 58 sdAb (INBRIX-105/ES101) 43 or VHH (PM1003, PM1032). 49 In contrast, FAP-4-1BBL (RG7827) and CD19-4-1BBL (RG6076) are antibody fusion proteins based on human 4–1BBL ectodomains fused to an IgG1 framework 32 and DSP107 is a trimeric fusion protein without the implementation of antibody components.…”

“…4–1BB agonists implementing at least two non-4-1BBL blocking binders by binding to CRD1 (e.g., urelumab, HOT1030) or to CRD4 (e.g., ABL50, ABL111, 65 AGEN2373, 51 PM1003, PM1032, 49 CTX-471 35 ) may lead to systemic 4–1BB activation, especially at high soluble 4–1BBL levels. ND-021/NM21-1480, however, can only bind one 4–1BB receptor and therefore cannot hyper-crosslink 4–1BB receptors in the absence of simultaneous PD-L1 binding.…”