Split CAR-T cells targeting CD312 and TIM-3 for acute myeloid leukemia to reduce the risk of antigen escape

To the Editor: A malignant disorder, acute myeloid leukemia (AML), comprises a limited proportion of leukemia-initiating cells or leukemic stem cells (LSCs), which form a clonal population.1 The main challenge in AML chimeric antigen receptor T-cell (CAR-T) therapy is a lack of a suitable targeting antigen. Many preclinical studies have reported sialic acid-binding Ig-like lectin 3 (CD33) and C-type lectin domain family 12 member A (CLL-1) as antigens that are not only expressed on AML cells but also in normal cells.2 With an epidermal growth factor-seven transmembrane (EGF-TM7)-like molecular structure, CD312 is a member of the adhesion G protein-coupled receptor (GPCR) family that consists of an extended extracellular domain (ECD) and a B-class GPCR-like seven transmembrane segment.3 The expression of CD312 has been observed in AML LSCs, but not in normal hematopoietic stem and progenitor cells (HSPCs).4 The T cell immunoglobulin and mucin domain-containing protein 3 (TIM-3) protein is a type I transmembrane protein that belongs to the TIM family. It is expressed in the majority of CD34+CD38– LSCs and CD34+CD38+ leukemic progenitors in AML.5 Additionally, anti-TIM-3 CAR-T cells are efficacious against AML cell lines, primary AML blasts, and murine models.6 Notably, anti-TIM-3 CAR T-cell therapy may enhance the clinical outcomes of AML patients by eradicating the LSCs present in the minimal residual disease (MRD).7 However, CAR-T cells targeting these antigens individually face the risk of off-target consequences. Hence, we used an “AND” gating strategy to develop novel CD312 and TIM-3 dual CAR-T cells and assess the efficacy of these modified split CD312-TIM3 CAR-T cells against AML cells in vitro and in vivo. This study design was approved by the Ethics Committee of Tianjin First Central Hospital (2020N212KY). The informed consent was obtained from the patients. The animal experiment protocol was approved by the Animal Care and Use Committee of Tianjin Medical University (No. 2025-SYDWLL-000096). Detailed information on methods and materials is presented in Supplementary Material, https://links.lww.com/CM9/C750. This study involved the development and evaluation of novel CD312 and TIM-3 dual CAR-T cells using an “AND” gating strategy. The efficacy of these split CD312-TIM3 CAR-T cells was assessed against AML cells both in vitro (using cell lines, primary patient cells, and isolated LSCs) and in vivo (using xenograft mouse models). The study also included analysis of antigen expression in patient samples, cytotoxicity assays, cytokine release measurements, and safety evaluations. The specific information of 50 patients is presented in Supplementary Table 1, https://links.lww.com/CM9/C750. CD312 analysis of bone marrow mononuclear cells from 50 AML patients revealed that 88% (44/50) of AML patients exhibited positive CD312 expression (defined as abnormal antigen expression of ≥20% in AML blasts), with a median expression level of 69.87%. Additionally, 70% (35/50) of AML patients showed positive TIM-3 expression, with a median expression level of 63.62%. Notably, 70% (35/50) of these patients expressed both CD312 and TIM-3. Additionally, the AML cell lines SKM-1 and Kasumi-3 were predominantly positive for CD312+ and TIM-3. Lentivirus system was used to generate CD312 single and CD312-TIM-3 split CAR-T cells; CD312 CAR and CD312-TIM-3 split CAR were successfully transfected into T cells, with transfection efficiencies of 44.05% and 53.17%, respectively. SKM-1 or Kasumi-3 AML cell lines were co-cultured with CD312 CAR-T cells, CD312-TIM-3 split CAR-T cells, or uninfected T cells to evaluate the precise cytotoxic effects. The result showed that CD312-TIM-3 split CAR-T cells exhibited superior cytotoxicity against CD312+ AML cells than CD312 CAR-T cells after co-culturing for eight hours. CD312-TIM-3 split CAR-T cells demonstrated an enhanced cytotoxic effect against CD312 knockout K562 cells compared to conventional CD312 CAR-T cells. Additionally, CD312-TIM-3 split CAR-T cells slightly upregulated interferon-γ (IFN-γ) and interleukin 6 (IL-6) cytokine levels compared the CD312 CAR-T cells Supplementary Figures 1 and 2, https://links.lww.com/CM9/C750. Thus, the preclinical data, including specific cytotoxicity and cytokine release, provide initial evidence for the bioactivity of CD312-TIM-3 split CAR-T cells. Future studies are required to translate these findings into clinical applications. Besides, we also assessed the cytotoxicity of CD312-TIM-3 split CAR-T cells compared to CD312 CAR-T cells in primary human AML cells isolated from relapsed/refractory AML patients, whose clinical characteristics are listed in Supplementary Figure 3A, https://links.lww.com/CM9/C750. Notably, enhanced CD312 and TIM-3 co-expression was observed in these patients. The cell-killing efficiency was evaluated after incubating AML primary cells with CD312-TIM-3 split CAR-T or CD312 CAR-T cells for 8 h and 24 h. CD312-TIM-3 split CAR-T cells’ cytotoxicity was significantly higher than CD312 CAR-T cells against primary AML cells. Moreover, co-culturing CD312-TIM-3 split CAR-T cells with primary AML cells slightly decreased the supernatant’s IL-6 and IFN-γ levels Supplementary Figures 3 and 4, https://links.lww.com/CM9/C750. In conclusion, CD312-TIM-3 split CAR-T cells demonstrated a substantial anti-AML activity against primary AML cells. In order to evaluate the efficiency of CD312-TIM-3 split CAR-T cells against LSCs in comparison to CD312 CAR-T cells, we performed an in vitro cytotoxicity assay on CD34+CD38–CD123+TIM3+ LSCs isolated from AML patients, whose clinical characteristics are listed in Supplementary Figure 5A, https://links.lww.com/CM9/C750. To assess the LSC cytotoxicity, CD312-TIM-3 CAR-T and CD312 CAR-T cells were co-cultured with LSCs for 24 h at an E:T ratio of 3:1. The CD312 CAR-T cells did not demonstrate specific targeting activity against LSCs, while the CD312-TIM-3 split CAR-T cells exhibited significant cell-killing efficacy against LSCs. Besides, CD312-TIM-3 split CAR-T cells co-cultured with LSCs significantly augmented the secretions of IFN-γ cytokine in the supernatant. Furthermore, we also isolated CD64-positive monocytes from healthy donors and co-cultured them with LSCs and CAR-T cells. Similar to previous results, CD312-TIM-3 split CAR-T cells showed significant cell-killing efficacy against LSCs when CD64-positive monocytes, LSCs, and CAR-T cells were co-cultured at a 1.5:1.5:1.0 ratio Supplementary Figures 5–8, https://links.lww.com/CM9/C750. In conclusion, CD312-TIM-3 split CAR-T cells are effective as optimal target antigens compared to single CD312 CAR T cells. Xenograft models were established in NSG (NOD.Cg-PrkdcscidIL2rgtm1Wjl/SzJ) mice to assess and compare the efficacy of CD312-TIM-3 split CAR-T cells and CD312 CAR-T cells against AML blasts in an in vivo setting. SKM-1-LGP cells (2 × 106) were injected to induce relapse in the models for assessing the efficacy of CD312-TIM-3 split CAR-T cells against AML relapse models. The results showed a significant reduction in leukemic cell burden after 28 days, following administration of either CD312 CAR-T cells or CD312-TIM-3 split CAR-T cells, highlighting their efficacy. However, there was a notable disparity in the prognosis between these two groups after the second injection of 2×106 SKM-1-LGP cells to induce relapse models. In the CD312 CAR-T cell-treated group, tumor cells initially exhibited a transient inhibition and then displayed a subsequent rapid invasion, resulting in the death of all mice before day 52. Conversely, the mice in the split CD312-TIM-3 CAR-T cell-treated group exhibited sustained tumor growth suppression over a period of at least 60 days. We also quantified the presence of CD312 CAR-T or CD312-TIM-3 split CAR-T cells in the bone marrow, and the results demonstrated a sustained presence of CD312-TIM-3 split CAR-T cells from days 7 to 47 in the relapsed AML xenograft model, indicating a sustained proliferative capacity in vivo. The biochemical and histochemical analyses of the mice liver, spleen, and kidney demonstrated the safety of both CD312 CAR-T and split CD312-TIM-3 split CAR-T therapies. Moreover, CD312-TIM-3 split CAR-T cells also exhibited killing effect in CD312 KO K562 models in vivo Supplementary Figure 10, https://links.lww.com/CM9/C750. Our results showed the potent in vivo anti-AML activity of CD312-TIM-3 split CAR-T cells in a relapsed AML xenograft model with safety, highlighting their efficacy in improving survival rates Supplementary Figures 9 and 10, https://links.lww.com/CM9/C750. CD312-TIM-3 split CAR-T cells effectively and safely inhibit AML growth, providing novel treatment options for patients with AML. Notably, in a relapsed AML animal transplantation model, split CD312-TIM-3 CAR-T cells exhibited a significant anti-AML activity, leading to a prolonged survival. The split CAR design enables the selective activation of T cells in the presence of specific ligands, thereby mitigating attacks on normal tissues and minimizing side effects. Moreover, split CARs exhibit an enhanced efficacy in targeting tumor cells while preserving the normal cells’ protective function during cancer treatment.8 Additionally, comprehensive research might help in developing novel split CARs that can simultaneously target multiple antigens, thereby enhancing therapeutic efficacy and reducing relapses.9 However, selecting appropriate targets is crucial to reducing toxicities and improving efficiency. In our study, CD312-TIM-3 split CAR-T cells exhibited a significantly enhanced cytotoxicity against AML cell lines, primary blasts, and CD34+CD38– LSCs than CD312 CAR-T cells. We found that CD312-TIM-3 split CAR-T cells slightly deregulated the release of IFN-γ and IL-6 compared to CD312 CAR-T cells in co-culture with primary AML cells at 24h. This discrepancy may be reconciled by considering several factors. First, cytokine release is a highly dynamic and context-dependent process. The timing of measurement is critical; an early, transient peak in cytokine production (e.g., at 8 hours) might not be captured at a later time point (24 hours) because feedback mechanisms or T cell exhaustion could lead to decreased levels. Second, the target cell population plays a crucial role. The observed decrease was noted specifically in co-cultures with primary AML cells, which possess a more complex and heterogeneous tumor microenvironment compared to cell lines. Interactions with other immune cells or soluble factors present in the primary sample could differently modulate the cytokine secretion profile of CAR-T cells. Third, patient-specific biological heterogeneity, particularly in relapsed/refractory AML, is a significant factor. The individual patient's disease biology, baseline immune status, and the unique composition of their leukemic cells (including the enhanced co-expression of CD312 and TIM-3 noted in some samples) could profoundly influence the CAR-T cell response and subsequent cytokine milieu.From a clinical perspective, this variability underscores the importance of personalized monitoring. An early, robust cytokine release (like the initially observed upregulation) could be indicative of potent CAR-T cell activation but may also correlate with a higher risk of cytokine release syndrome (CRS). Conversely, a dampened cytokine response in a specific context might reflect different tumor-immune interactions but does not necessarily preclude efficacy, as evidenced by the potent cytotoxicity observed. This highlights the need to decouple efficacy biomarkers from toxicity biomarkers. Moreover, CD312-TIM-3 split CAR-T cells demonstrated potent cytotoxicity that specifically targeted CD34+CD38– LSCs and unaffected the monocytes derived from healthy donor. Furthermore, CD312-TIM-3 split CAR-T cells displayed an exceptional anti-AML activity resulting in prolonged survival of the treated animals in an animal xenograft model simulating relapsed AML conditions. Additionally, a natural safety switch mechanism allows the split CAR-T cells to be safely eliminated in emergencies, thereby preserving hematopoietic stem cells. The split CAR-T cells targeting CD312 and TIM-3 cells can become a novel therapeutic approach for refractory and relapsed AML patients, due to their capacity to eradicate LSCs and mitigate relapses post-CAR-T cell infusion. The results indicate that split CARs can effectively recognize and eliminate tumor cells as well as protect normal hematopoietic cells, showing promising application prospects. More detailed discussion is presented in Supplementary Material, https://links.lww.com/CM9/C750. In conclusion, targeting CD312 and TIM-3 with split CAR-T cells is a promising yet preliminary approach for AML. The clinical relevance and therapeutic efficacy of CD312-TIM-3 split CAR-T cells warrant further investigation in well-designed clinical trials. Funding This work is sponsored by Tianjin Health Science and Technology Projects (No. TJWJ2025MS011); National Natural Science Foundation of China (No. 82570277); the General Program of Tianjin Natural Science Foundation Joint Fund (No. 25JCLMJC00580) and Tianjin Key Medical Discipline (Specialty) Construction Project (No. TJYXZDXK-3-001A-004). Conflicts of interest None.