Abstract 3998: Preclinical efficacy of a synthetic Notch-based combinatorial immunotherapy against colorectal cancer with HER2 amplification

HER2 amplification occurs in about 5% of colorectal cancers (CRC) and is only partially associated with clinical response to combined HER2/EGFR targeted treatment. An alternative approach, based on Adoptive Cell Therapy (ACT) using T cells engineered with anti-HER2 Chimeric Antigen Receptor (CAR), proved toxic due to "on-target off-tumor" activity. Therefore, we developed a combinatorial ACT strategy to safely target HER2 amplification in CRC using a synNotch-based artificial regulatory network. A synthetic Notch receptor was employed in which the extracellular domain is an anti-HER2 scFv and the intracellular domain contains the GAL4VP64 artificial transcription factor. Engagement of the HER2-synNotch by overexpressed HER2 on the surface of target cells induces GAL4VP64 cleavage and translocation to the nucleus, where it drives expression of a CAR under a GAL4-responsive promoter. In this way, only cells co-expressing both HER2 and the CAR target are killed. Carcinoembryonic antigen (CEA) was selected as a colon-specific CAR target: CEA expression is indeed restricted to the digestive tract and is increased in CRC. The natural killer cell line NK-92 was chosen as effector to be engineered. NK-92 cells transduced with HER2-synNotch and inducible CEA-CAR lentiviral vectors were subsequently sorted in the OFF and ON state to select those with the higher CAR induction after synNotch engagement. Subsequently, cloning of sorted cells led to identification of an optimally responsive clone (clone 5F), in terms of specificity and amplitude of CAR induction. The 5F clone demonstrated significant activity in vitro and in vivo specifically against HER2amp CRC models, with no effects on cells with physiological HER2 levels. In vitro, the 5F clone displayed selective cytotoxicity against HER2amp/CEA+ CRC cells, with minimal killing activity against HER2amp/CEA- cells, or HER2norm/CEA+ cells. Additional assays on 3D organoids highlighted better recruitment and infiltration by the 5F clone respect to NK-92 WT cells, only in HER2amp models. In vivo, NK-92.5F significantly impaired tumor growth in two different HER2amp CRC models, with no effect on HER2norm CRC cells. To further improve survival, tumor penetration and in vivo efficacy of the NK-92.5F clone, a more complex system was built, in which HER2 synNotch engagement drives expression of both the CEA-CAR and IL2. 5F-IL2 cells displayed further increased cytotoxicity in vitro. In vivo, 5F-IL2 cells drastically increased survival of mice carrying HER2amp CRC xenografts with respect to the parental 5F clone. The observed selective therapeutic efficacy both in vitro and in vivo of this innovative HER2 synNotch/CEA-CAR system, and its future evolutions, opens perspectives for possible clinical applications in cases of HER2amp CRC resistant or partially responsive to HER2/EGFR blockade.