1196 Streamlining T cell engager development with a diverse panel of fully human CD3-binding antibodies, bispecific engineering technology, and an integrated discovery engine

Lindsay DeVorkin, Juntao Mai, Kate Caldwell, Timothy M. Jacobs, Raffi Tonikian, Karine Hervé, Yuri Hwang, Cristina Faralla, Wei Wei, Emma Lathouwers, Rhys Chappell, Stefan Hannie, Katherine Lam, Harveer Dhupar, Tran Tran, Melissa Cid, Lena M. Bolten, Tova Pinsky, Ping Xiang, Courtenay Lai, Ahn Lee, Patrick Chan, Jasmine Chin, Aaron Yamniuk, Kush Dalal, Bryan C. BarnhartRegular 2022 · DOI

Introduction

T cell engagers (TCEs) represent a transformative class of immuno-oncology therapeutics designed to bridge the gap between the immune system and malignancy. By physically linking a patient’s T cells to tumor-associated antigens, TCEs trigger targeted cytotoxic activity. However, the development of these bispecific antibodies is a complex balancing act. The primary challenge lies in the CD3-binding arm: if the binding is too potent, it risks systemic over-activation and a dangerous "cytokine storm"; if it is too weak, the therapeutic effect is lost.

Historically, the industry has relied on a narrow pool of existing CD3 antibodies, necessitating extensive and often inefficient downstream protein engineering to achieve the desired pharmacological profile. This research presents an integrated discovery engine that bypasses these limitations by starting with a massive, diverse library of fully human CD3-binding antibodies, allowing for the selection of "naturally" optimal candidates.

Methods and Discovery Pipeline

The researchers at AbCellera utilized a high-throughput microfluidic screening platform to evaluate over 3.5 million single cells from humanized mice. This process identified more than 200 unique, fully human CD3-specific antibodies.

Following identification, the antibodies underwent rigorous characterization:

  • Affinity and Kinetics: Binders were assessed for affinity to CD3εδ and CD3εγ, with binding kinetics and epitope binning used to ensure structural diversity.
  • Cross-Reactivity: Essential for translational safety, the team confirmed cross-reactivity between human and cynomolgus monkey primary T cells.
  • OrthoMab™ Engineering: To create functional TCEs, the CD3 binders were paired with an EGFR-targeting arm using the OrthoMab™ bispecific platform.
  • Developability Profiling: To ensure manufacturability, candidates were screened for hydrophobicity (aHIC), self-association (AC-SINS), polyspecificity (BVP-ELISA), thermal stability (nanoDSF), and aggregation (aSEC).
  • Functional Validation: The final bispecifics were tested using NFAT reporter assays for T cell activation, xCELLigence assays for real-time tumor cell killing, and FLEXMAP CD for cytokine release profiling.

Results: A Diverse and Validated CD3 Toolbox

The discovery engine yielded a novel panel of over 200 CD3-specific antibodies targeting diverse epitopes, including several distinct from previously documented cross-reactive binders. These antibodies exhibited a wide spectrum of binding affinities ($K_D$ ranging from ~1 nM to 1 μM) and T cell activation potencies ($EC_{50}$ from ~6 to 190 nM).

This diversity proved critical during the proof-of-concept phase. By integrating these binders into CD3 x EGFR bispecifics, the team generated a library of engagers with highly tunable performance. The resulting bispecifics demonstrated:

  • Activation Potency: $EC_{50}$ values ranging from ~2 pM to 2 nM.
  • Tumor Killing: $EC_{50}$ values from ~0.01 to 1 nM, with top candidates achieving >90% tumor cell killing.
  • Optimized Safety Profiles: Most significantly, the platform identified specific variants capable of high-potency tumor destruction while maintaining minimal cytokine release—a key requirement for improving the safety margin of TCEs in the clinic.

Discussion and Conclusion

The study demonstrates that starting with a large, diverse pool of fully human antibodies significantly reduces the need for reactive protein engineering. By selecting candidates that inherently possess the desired affinity and developability traits, the discovery timeline is streamlined.

While this proof-of-concept utilized EGFR as a model target, the implications for the broader field of immunotherapy are substantial. The ability to decouple tumor-killing efficacy from cytokine-related toxicity through precise CD3 fine-tuning addresses one of the most significant hurdles in TCE development. Future work will focus on validating these "high-potency, low-cytokine" profiles in vivo to confirm that the favorable lab results translate to improved safety and efficacy in living systems. By integrating high-throughput screening, bispecific engineering, and early developability assessment, this engine provides a robust framework for the next generation of immuno-oncology treatments.

Summary updated

Abstract

<h3>Background</h3> CD3 T cell engagers have the potential to be a cornerstone of immuno-oncology. However, a limited pool of CD3-binding antibodies and technological challenges in engineering bispecifics have hindered development. Discovering effective T cell engagers requires two target-binding arms — a CD3 arm that fine-tunes T cell activation and a tumor arm with high specificity for cancer cells — optimized as a whole to work in concert with each other. Beginning with diverse panels of antibodies increases the probability of finding appropriately potent and developable T cell engagers and reduces the need for downstream engineering. <h3>Methods</h3> We used microfluidic technology to screen more than 3.5 million single cells from humanized mice and identified &gt;200 CD3-specific antibodies. Using high-throughput assays, we determined affinity for CD3εδ and CD3εγ, cross-reactivity to human and cyno primary T cells, CD3 binding kinetics, and epitope bins. We assessed T cell activation by measuring CD25 and CD69 expression by flow cytometry. We then used our bispecific engineering platform, OrthoMab<sup>TM</sup>, to generate a proof-of-concept panel of CD3 x EGFR bispecific antibodies. Developability properties were assessed, including hydrophobicity (aHIC), self-association (AC-SINS), polyspecificity (BVP-ELISA), stability (nanoDSF), and aggregation (aSEC). CD3 T cell engager potencies were measured using an NFAT reporter T cell activation assay and an xCELLigence tumor cell killing assay, and cytokine release was assessed by FLEXMAP CD. <h3>Results</h3> We identified hundreds of fully human CD3-specific antibodies that are diverse, developable, and validated. The antibodies displayed a wide range of CD3 binding affinities (K<sub>D</sub> ~1 nM to 1 μM), binding kinetics, and T cell activation potencies (EC<sub>50</sub> ~6 to 190 nM). Data on this novel panel includes epitope binning, which revealed human and cyno CD3-binders that are distinct from previously described cross-reactive antibodies. The antibodies were assessed using a range of biophysical assays and have favorable developability properties. In an expanded proof-of-concept study, we used OrthoMab<sup>TM</sup> to generate a panel of CD3 x EGFR bispecific antibodies. The resulting bispecifics had favorable developability properties, and displayed a wide range of antigen-dependent T cell activation (EC<sub>50</sub> ~2 pM to 2 nM) and tumor cell killing potencies (EC<sub>50</sub> ~0.01 to 1 nM). From this panel, we identified potent T cell engagers that achieved &gt;90% tumor cell killing with low levels of cytokine release. <h3>Conclusions</h3> By integrating our panel of CD3-binding antibodies with our bispecific engineering and high-throughput antibody assessment capabilities, we identified developable CD3 T cell engagers with potent tumor cell-killing activity and minimal cytokine release.
BibTeX
@inproceedings{devorkin2022streamlining,
  author = {Lindsay DeVorkin and Juntao Mai and Kate Caldwell and Timothy M. Jacobs and Raffi Tonikian and Karine Hervé and Yuri Hwang and Cristina Faralla and Wei Wei and Emma Lathouwers and Rhys Chappell and Stefan Hannie and Katherine Lam and Harveer Dhupar and Tran Tran and Melissa Cid and Lena M. Bolten and Tova Pinsky and Ping Xiang and Courtenay Lai and Ahn Lee and Patrick Chan and Jasmine Chin and Aaron Yamniuk and Kush Dalal and Bryan C. Barnhart},
  title = {{1196 Streamlining T cell engager development with a diverse panel of fully human CD3-binding antibodies, bispecific engineering technology, and an integrated discovery engine}},
  booktitle = {Regular},
  year = {2022},
  doi = {10.1136/jitc-2022-sitc2022.1196},
}
Figure