Background: Acute myeloid leukemia (AML) is one of the most common and highly heterogeneous hematological malignancy. The prognosis is poor especially for the elderly patients. For Relapsed/Refractory(R/R) AML patients, the one-year overall survival rate is merely 20%. CAR-T cells have shown tremendous therapeutic benefits for ALL, DLBCL and MM. One of the limitations of current CAR-T therapies is the use of autologous T cells. This can be limiting due to the cost, complicated manufacturing process, rapid disease progression, and manufacturing failures etc. Therefore, there is a need to develop the ready-to-use allogeneic cell product. Double negative T (DNT) cells are the ideal candidate for AML cell therapy as they naturally kill AML cells, and have been successfully administered to AML patients with therapeutic benefit without causing GvHD. Relying on their own natural immune activated receptors, DNT cells could lyse a wide range of tumors independent of TAA expression, which have more advantages compared to the conventional CAR-T cells. In our study, we developed the allogeneic product of CAR-DNT cells targeting C-type lectin-like molecule-1 (CLL1), which is a promising target due to its over-expression in primary AML cells and leukemia stem cells, while absence on hematopoietic stem/progenitor cells.Methods: We developed a humanized single chain variable fragment (scFv) with specific recognition of human CLL1 antigen by screening from 62 murine-derived CLL1 antibodies. Then, we designed a second-generation CLL1 CAR structure containing 4-1BB-CD3ζ and transduced it into the DNT cells by lentivirus. To clarify the GvH and HvG reactions of DNT cells, we performed the mixed lymphocyte reaction experiments in vitro and compared DNT cells with TCR and B2M double knockout T(DKO) cells and TCR knockout T(TKO) cells. We evaluated the anti-tumor function of CLL1 CAR-DNT cells through killing experiments by different effector-to-target ratios and multiple stimulations with AML cell lines (THP1 and HL60). Furthermore, we speculated that the CAR structure plays a critical role in both tumoricidal function and persistence of CAR-DNT cells. We designed nearly 20 enhanced CAR structures based on the signaling pathway of DNT cells. These newly designed structures have been tested in vitro and will be confirmed in vivo.Results: The humanized anti-CLL1 scFv candidate H1-3Q2 demonstrated high specificity toward human CLL1 antigen at the Membrane Protein Array. H1-3Q2 based CAR-T cells showed good tumor suppression in vivo. However, due to the heterogenous expression of CLL1 in AML, CLL1 CAR-T cells may not be able to eliminate AML cancer cells completely. While DNT cells could naturally lyse tumor cells through the receptors like NKG2D and DNAM-1, providing them with lots of advantages. Moreover, DNT cells have the potential to be used as allogenic cell therapy product. Co-culturing allogenic PBMCs or umbilical cord blood with DNT cells for up to 72 hours resulted in little or no killing reaction, indicated a low risk of GvHD. Conversely, the PBMCs from three different donors were used to attack allogenic DNT cells or DKO/TKO T cells in vitro. The killing effect on DKO/TKO T cells gradually increased over time, while no effect on DNT cells was observed even after 192 hours. This suggested that DNT cells are not easily cleared by host immune cells. DNT cells could efficiently lyse various hematological and solid tumor cell lines, particularly showing high sensitivity towards AML cells. The introduction of the CAR moiety further enhanced the specificity of DNT cells. The in vitro killing results demonstrated that CLL1 CAR-DNT cells exhibited stronger and faster AML resistance compared with CAR-T cells, and even eliminated tumor cells that have lost TAA. However, CLL1 CAR-DNT cells showed limited proliferation in the multiple-stimulation assays. To address this, we designed enhanced CAR structures. Screening by multiple-challenge assays, the fifth-generation CAR exhibited outstanding proliferation and superior tumor lysis ability in vitro.Conclusions: Our findings demonstrate that CAR-DNT product is an ideal candidate for allogeneic adoptive immunotherapy and exhibits superior tumoricidal function. This study underscores the potential of the “off-the-shelf” CLL1 CAR-DNT cell product for treating R/R AML.