Abstract 4230: 1st-in-human CAR T targets MUC1 transmembrane cleavage product

Purpose: To develop a MUC1-targeted CAR T that recognizes the growth factor receptor form, MUC1*, does not bind full-length MUC1, hits a wide range of cancers but binds to little or no normal tissues, and effectively kills tumor cells. We developed a novel CAR T, huMNC2-CAR44, that targets MUC1*, which is the transmembrane cleavage product of MUC1. A 1st-in-human clinical trial for metastatic breast cancers opened in Q4, 2019 at the Fred Hutchinson Cancer Research Center. MUC1* is a growth factor receptor that is activated when onco-embryonic growth factor NME7AB dimerizes its truncated extra cellular domain. The binding site for NME7AB is ectopic and is only unmasked after the tandem repeat domain of MUC1 is cleaved and shed from the cell surface. Unlike full-length MUC1, MUC1* has no sites for O-linked glycosylation, so antibodies such as 5E5 that bind to aberrant, trapped glycans cannot bind to the MUC1*. Methods: We developed a novel antibody screen that identifies monoclonal antibodies that bind to a conformational epitope of MUC1* that is created when MUC1 is cleaved by MMP9, which has been linked to poor prognosis and metastasis. The cancer selectivity of conformation-specific anti-MUC1* antibodies, anti-MUC1* antibodies that recognize linear epitopes, or antibodies that bind to the tandem repeat domain of full-length MUC1 was assessed in IHC studies of cancer vs normal tissue arrays. Conformation specific anti-MUC1* antibodies were incorporated into CARs, transduced into human T cells and tested in vitro and in vivo for their ability to kill MUC1* positive, but not MUC1* negative, tumor cells. Results: Anti-MUC1* antibody huMNC2 binds to the conformational epitope that is unmasked when MUC1 is cleaved to MUC1* by MMP9. huMNC2 competes with onco-embryonic growth factor NME7AB for this same conformational epitope on MUC1*. Neither huMNC2 nor NME7AB binds to full-length MUC1. IHC studies of tissue micro arrays showed that antibodies that bound to full-length MUC1 did not react with 29% of the breast cancer tissues in several arrays. Conversely, anti-MUC1* antibody MNC2 bound robustly to 95% of the breast cancer tissues in serial sections of the same arrays. Further, huMNC2-scFv bound to 83% ovarian, 78% pancreatic and 71% of lung cancer tissues, with little to no binding to normal tissues. In vivo, huMNC2-CAR44 T cells inhibited or completely obliterated a variety of MUC1* positive solid tumors in NSG mice (ng 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 4230.