AUTHORS
Jordan S. Kesner, PhD, Jonathan Liu2, Michelle Saetersmoen, PhD3, Yonghong Yao, MD4, Michel Sadelain, MD, PhD5, Sascha Haubner, MD6
AFFILIATIONS
1 Columbia Institute for Cell Engineering and Therapy, Vagelos College of Physicians and Surgeons, Columbia University Irving Medical Center, New York, New York, United States
2 Columbia Institute for Cell Engineering and Therapy, Vagelos College of Physicians and Surgeons, Columbia University Irving Medical Center, New York, New York, United States
3 H. Lee Moffitt Cancer Center and Research Institute, Tampa, Florida, United States
4 Columbia Institute for Cell Engineering and Therapy, Vagelos College of Physicians and Surgeons, Columbia University Irving Medical Center, New York, New York, United States
5 Columbia Institute for Cell Engineering and Therapy, Vagelos College of Physicians and Surgeons, Columbia University Irving Medical Center, New York, New York, United States
6 Columbia Institute for Cell Engineering and Therapy, Vagelos College of Physicians and Surgeons, Columbia University Irving Medical Center, New York, New York, United States
ABSTRACT
CAR T cell therapy in AML is complicated by inflammatory and hematological toxicities. We previously developed a CAR strategy to enhance CAR sensitivity against AML while sparing normal hematopoietic stem and progenitor cells (HSPCs), via T cells co-expressing a signal-calibrated CAR and CCR: ADGRE2-28z1XX and CLEC12A-BB (ADCLEC.syn1), which is being tested in a phase 1 clinical trial (NCT05748197). Beyond on-target toxicities, off-target toxicities remain a challenge: Immune effector cell-associated hematotoxicity (ICAHT) contributes to morbidity and mortality and correlates with an inflammatory bone marrow microenvironment.
Here, we use pre-clinical models to mechanistically evaluate off-target hematological toxicities and apply T cell engineering for cytokine calibration, uncoupling CAR T effector function from inflammatory cytokine release to limit ICAHT.
To assess effects of the CAR T secretome on normal HSPCs, we exposed CD19-28z1XX and ADCLEC.syn1 CAR T cells to antigen-positive AML cell lines, generating conditioned media containing cytokines released by CAR T cells, IFN-γ and TNF being among the top 10 most abundant cytokines in a 45-plex protein panel. Normal HSPCs exposed to conditioned media from antigen-engaged CAR T cells showed reduction of CD34+CD38- primitive HSPCs. Transcriptionally, these HSPCs were enriched for pathways of inflammatory cytokine responses (IL-1, TNF, IFN-γ, IL-10, IL-17), NF-κB and ERK signaling, myeloid/hematopoietic progenitor differentiation, and ligand-independent apoptosis, while losing expression of stemness-associated genes (HLF, MEIS1). These pathways were found to be similarly enriched in a publicly accessible transcriptional dataset of CD34+ cells from patients with ICAHT upon commercial CD19 CAR T therapy.
Given the association between pro-inflammatory cytokine exposure and loss of stem cell reserve through differentiation and apoptosis, we reasoned that excessive cytokine release by CAR T cells may drive ICAHT. Rather than genetically depleting ICAHT-inducing cytokines in T cells, we hypothesized that cytokine attenuation could reduce hyperinflammatory events while preserving CAR T functionality. We thus designed a single-vector CAR platform with multiplexed shRNA elements under the same promoter as the CAR transgene, directly linking cytokine attenuation with CAR expression. This generalizable strategy is applied here to ADCLEC.syn1, a clinically investigated CAR therapy in AML.
ADCLEC.syn2 integrates an IFNG-shRNA for IFN-γ calibration and was evaluated in vitro and in vivo for efficacy and safety. In a MOLM13 xenograft model, ADCLEC.syn2 (IFNG-KD) was as efficacious as ADCLEC.syn1 (IFNG-WT) and outperformed ADCLEC.syn1 (IFNG-KO), suggesting that cytokine down-modulation is preferable to complete ablation. We engrafted mice with IFNGR1-WT or IFNGR1-KO MOLM13 cells, then treated with TRAC-KO or TRAC-KO/IFNG-KO T cells expressing ADCLEC.syn1 or IFN-γ-calibrated ADCLEC.syn2. Responses in conditions with IFNGR1-KO MOLM13 or TRAC-KO/IFNG-KO ADCLEC.syn1 T cells were significantly reduced, whereas TRAC-KO ADCLEC.syn2 T cells achieved the best response, demonstrating the role of IFN-γ in CAR T activity against AML, further supporting the benefit of calibrating rather than depleting IFN-γ in T cells. In our hematotoxicity models, conditioned media from ADCLEC.syn2 or IFN-γ-calibrated 19-28z1XX CAR T cells did not cause loss of primitive HSPCs and preserved stem cell-associated gene expression equivalent to IFNG-KO T cell or IFNGR1-KO HSPC controls.
Our single-vector platform allows for simultaneous calibration of multiple inflammatory mediators: ADCLEC.syn3 co-expresses shIFNG and shTNF to control two major inflammatory axes. Post-antigen engagement in a 14-day repetitive AML re-challenge assay, ADCLEC.syn3 T cells were enriched in Tscm phenotype, had higher proliferative capacity (Ki-67), led to higher net CAR T expansion and improved per-cell cytolytic capacity relative to ADCLEC.syn1, indicating that multiplexed cytokine calibration favorably impacts T cell differentiation, proliferation and functional outcomes. In an in vivo MOLM13 re-challenge xenograft, ADCLEC.syn2 and ADCLEC.syn3 showed equivalent anti-leukemic efficacy to ADCLEC.syn1.
Overall, we demonstrate that CAR T-derived inflammatory cytokines directly impair HSPC stemness and that multiplexable cytokine calibration rather than depletion enhances CAR T cell hematological safety while supporting anti-leukemic efficacy.
DISCLOSURES
Jordan S. Kesner, PhD: No relevant conflicts of interest to declare.