CD33 KO -CD33-mesothelin Loop CAR design avoids fratricide and improves efficacy of iNK cells against acute myeloid leukemia

Yao Wang, Xiujuan Zheng, Zhiqian Wang, Ziyun Xiao, Yunqing Lin, Fan Zhang, Yanhong Liu, Pengcheng Liu, Qitong Weng, Leqiang Zhang, Chengxiang Xia, Dehao Huang, Lijuan Liu, Yanping Zhu, Qi Zhang, Hanmeng Qi, Yi Chen, Yiyuan Shen, Chenyuan Zhang, Jiacheng Xu, Yaoqin Zhao, Jiaxin Wu, Tongjie Wang, Mengyun Zhang, Minming Li, Wenbin Qian, Aibin Liang, Xin Du, Wenyu Yang, Tianyuan Hu, Quan Chen, Xiaofan Zhu, Fangxiao Hu, Jinyong WangJournal for ImmunoTherapy of Cancer 2025 · DOI

Abstract

Loop CAR empowered both UCB-NK cells and hPSC-iNK cells with superior cytotoxicity against CD33<sup>+</sup>MSLN<sup>+</sup> tumor cells. Genetic disruption of CD33 avoided fratricide and improved efficacy of Loop CAR-iNK cells against AML. This innovative strategy possesses unique advantages and translational potential for treating AML.

Introduction

Acute myeloid leukemia (AML) remains a significant clinical challenge, particularly for elderly patients who cannot tolerate intensive chemotherapy or wait for the lengthy production cycles of autologous CAR-T cell therapies. Natural Killer (NK) cells have emerged as a promising "off-the-shelf" alternative. However, two primary obstacles hinder their effectiveness: fratricide and antigen escape.

Fratricide occurs when CAR-NK cells target markers like CD33, which are also expressed on healthy NK cells, leading to self-destruction. Antigen escape occurs when cancer cells downregulate a single target antigen to evade detection. This study introduces a novel "Loop CAR" design and CRISPR-mediated gene editing to overcome these hurdles.

Methods and Design

The researchers developed a bivalent "Loop CAR" architecture to simultaneously target CD33 and Mesothelin (MSLN). Unlike standard tandem designs, the Loop structure utilizes a specific spatial arrangement of variable light ($V_L$) and heavy ($V_H$) chains connected by a "whitlow" linker to optimize receptor clustering.

Structural design of MSLN CAR, CD33 CAR, and the bivalent CD33-MSLN Loop CAR.

To eliminate fratricide, the team used CRISPR/Cas12i to knock out the CD33 gene in human pluripotent stem cells (hPSCs). These $CD33^{KO}$ hPSCs were then differentiated into mature NK cells (iNKs) using a specialized 3D organoid induction system. This approach ensures a scalable supply of standardized, "invisible" therapeutic cells that do not kill one another. The therapy was validated using established cell lines (Nomo-1, AGS, HL-60) and primary bone marrow samples from AML patients.

Key Results

Enhanced Cytotoxicity and Expansion

The Loop CAR design demonstrated superior receptor clustering and immunological synapse formation compared to single-target or tandem CARs. Crucially, the $CD33^{KO}$ modification successfully restored the expansion capacity of the NK cells, which was otherwise severely limited by self-killing. Without the knockout, population growth was stunted; with it, the cells expanded normally and showed higher specific cytotoxicity.

Efficacy Against Patient Samples

The optimized $CD33^{KO}$ Loop CAR-iNK cells showed significantly higher specific cytotoxicity against primary AML cells from patients compared to control groups, demonstrating the clinical relevance of the dual-targeting approach.

Specific cytotoxicity of various NK cell constructs against primary AML cells from Patient #2 at different E:T ratios.

Improved Survival In Vivo

In mouse models of AML, the $CD33^{KO}$ Loop CAR-iNK treatment resulted in the most robust antitumor response. Mice treated with these optimized cells achieved a median survival of 42–46 days, a significant improvement over untreated mice (28–32 days) and those treated with standard CAR designs (34–38 days).

Discussion

This study highlights the importance of CAR architecture and genetic "shielding" in developing effective NK cell therapies. By targeting both CD33 and MSLN, the Loop CAR reduces the risk of antigen escape. While MSLN shedding remains a potential challenge for clinical translation, the dual-targeting strategy and the use of an hPSC-derived platform offer a potent, scalable solution for AML. Future work will focus on safety testing and exploring the potential of this design in solid tumors where MSLN is highly expressed, such as ovarian or pancreatic cancer.

Summary updated

Abstract

Loop CAR empowered both UCB-NK cells and hPSC-iNK cells with superior cytotoxicity against CD33<sup>+</sup>MSLN<sup>+</sup> tumor cells. Genetic disruption of CD33 avoided fratricide and improved efficacy of Loop CAR-iNK cells against AML. This innovative strategy possesses unique advantages and translational potential for treating AML.
BibTeX
@article{wang2025cd,
  author = {Yao Wang and Xiujuan Zheng and Zhiqian Wang and Ziyun Xiao and Yunqing Lin and Fan Zhang and Yanhong Liu and Pengcheng Liu and Qitong Weng and Leqiang Zhang and Chengxiang Xia and Dehao Huang and Lijuan Liu and Yanping Zhu and Qi Zhang and Hanmeng Qi and Yi Chen and Yiyuan Shen and Chenyuan Zhang and Jiacheng Xu and Yaoqin Zhao and Jiaxin Wu and Tongjie Wang and Mengyun Zhang and Minming Li and Wenbin Qian and Aibin Liang and Xin Du and Wenyu Yang and Tianyuan Hu and Quan Chen and Xiaofan Zhu and Fangxiao Hu and Jinyong Wang},
  title = {{CD33 <sup>KO</sup> -CD33-mesothelin Loop CAR design avoids fratricide and improves efficacy of iNK cells against acute myeloid leukemia}},
  journal = {Journal for ImmunoTherapy of Cancer},
  year = {2025},
  doi = {10.1136/jitc-2025-011887},
}
Figure