A novel and highly effective truncated BCMA safety switch for adoptive cell therapy

Abstract Introduction Adoptive cell therapies using engineered immune cells such as chimeric antigen receptor (CAR) T cells can be highly effective but may also cause severe acute toxicities including cytokine release syndrome and neurotoxicity or long-term deleterious effects such as B-cell aplasia with CD19-targeting CAR T cell therapy due to on-target, off-tumor toxicity. In addition, risk of tumorigenicity exists with any genetically engineered cell therapy. Incorporation of a safety switch that allows elimination of the engineered cells could potentially provide a strategy to mitigate those risks. Here, we developed a novel safety switch that could be activated using therapeutic agents readily available in the clinic. We selected BCMA as a safety switch as i) it is not naturally expressed on immune cells commonly used for adoptive cell therapies including T cells, NK cell, NKT cells, gd T cells, and macrophages, ii) its expression is restricted to plasma cells and a small subset of B cells amongst normal human tissues, and iii) highly effective BCMA-targeting bispecific T-cell engagers, such as teclistamab, elranatamab, and linvoseltamab are readily available in the clinic and used for the treatment of multiple myeloma. Methods and Results We developed and tested two formats of the BCMA safety switches: a transmembrane domain-containing chimeric truncated BCMA (ctBCMA) and a glycosylphosphatidylinositol (GPI)-anchored BCMA. ctBCMA was generated by fusing two extracellular domains of BCMA with hinge and transmembrane domain of PD-L1 and cytoplasmic domain of Trop2. GPI-BCMA was generated by fusing two extracellular domains of BCMA with GPI anchor signals derived from CD52 or CD55. Both ctBCMA and GPI-BCMA lack g-secretase cleavage site and the cytoplasmic domain of BCMA to prevent signaling in response to BCMA ligands, but still permit endocytosis of agents such as belantamab mafodotin, an anti-BCMA antibody-drug conjugate. Both formats also include two CD34 epitopes as the linker joining the two BCMA extracellular domains to allow for enrichment of transduced cells using readily available GMP-grade anti-CD34 microbeads. To evaluate these safety switches, CD19-targeting CAR T cells co-expressing either ctBCMA or GPI-BCMA were generated from healthy donor T cells by transducing them with a bicistronic lentiviral vector. Consistent with the design of our construct, cell surface level of ctBCMA was stable in g-secretase-expressing cells, and soluble BCMA levels in the culture supernatants were not altered in the presence or absence of a g-secretase inhibitor indicating that ctBCMA is resistant to cleavage by g-secretase. To assess the efficacy of the safety switch, CD19 CAR T cells co-expressing either ctBCMA or GPI-BCMA were transduced with GFP and cocultured with a B-cell lymphoma cell line at an effector to target ratio of 1:1 in the absence or presence of teclistamab, elranatamab or belantamab mafodotin or their respective isotype antibody controls. Using Incucyte® Live Cell Imaging System, we found that all three agents but not isotype antibodies rapidly and efficiently eliminated ctBCMA- and GPI-BCMA-expressing CAR T cells at concentrations >10-fold lower than the serum Cmax observed in humans with standard dosing of these agents (P<0.0001). No significant cytotoxicity was observed in CAR T cells without BCMA safety switch. We also assessed the efficacy of the safety switches in vivo in B-cell lymphoma tumor-bearing NSG mice using luciferase-labeled CD19 CAR T cells. While CAR T cell expansion was observed by bioluminescence imaging in control mice in response to tumor, they were rapidly and durably cleared in teclistamab-treated mice (total flux P<0.001). In agreement with this, mice treated with teclistamab had significantly higher numbers of B-cell lymphoma tumor cells in spleen and bone marrow at necropsy as compared to controls (P<0.01). Similar results were observed when belantamab mafodotin was used to target the BCMA-expressing CAR T cells. Conclusion In conclusion, our results indicate that either ctBCMA or GPI-BCMA may serve as a novel and highly effective safety switch for engineered cell therapies and they could be activated using commercially available anti-BCMA agents. They may be used with various autologous or allogeneic CAR T cell and other adoptive immune cell therapy approaches and could also be broadly applicable for use with iPSC-derived therapeutic products being evaluated in regenerative medicine.