Background
Acute myeloid leukemia (AML) is the most common acute leukemia in adults. It is characterized by the accumulation of immature myeloid cells in the bone marrow that results in dysfunction of hematopoiesis [
1]. Chemotherapy and hematopoietic stem cell transplantation (HSCT) are the two standard treatments of AML. While improvements have been made in AML treatment in recent decades, the 5-year survival rate remains below 50% due to chemo-resistance or toxicity to these treatments [
2,
3]. Most patients eventually succumb to relapsed and/or progressive disease [
4], suggesting novel therapeutic strategies are urgently needed for these patients.
Recent advances in immunotherapy have generated substantial excitement for cancer patients. CAR (chimeric antigen receptors)-transduced T cell therapy is one of the new approaches with superior efficacy for the treatment of AML. It combines the specificity of antibody target recognition with the potent effector mechanisms of T cells. CARs are composed of an extracellular antigen-binding domain derived from the single-chain variable fragment (scFv) of the targeting antibody, a transmembrane domain, an intracellular signaling domain of the CD3-ζ, and one or more costimulation domains such as 4-1BB (CD137), CD28, or ICOS (CD278) [
5‐
8]. While the scFv domain mediates target recognition in a major histocompatibility complex (MHC)-independent manner [
9‐
11]. Immune escape through reduction of antigen processing and presentation is not required for CAR-T cells to kill cancer cells. Recently, CAR-T cell therapy has become the most promising approach for leukemia treatment by using lineage specific surface makers of blood cells as targets. For example, CD19-redirected CAR-T cells have been demonstrated with great success in treating CD19
+ B cell malignancies [
12‐
16]. Its response rate for B cell acute lymphoblastic leukemia (B-ALL) has reached approximately 90%.
In the past decades, a number of tumor antigens such as CD33, CD123, CD44, TIM-3, CD47, and CD32 have been explored as target antigens for AML treatment [
17‐
24]. Although mono-antibody (mAb) therapy targeting these antigens have been shown anti-tumor activity in animal models and clinical trials, the overall therapeutic efficacies remain low. With the success of CD19 CAR-T cell therapy on B-ALL, CAR-T cells targeting these AML associated antigens have been developed and showed higher anti-tumor efficacy compared to mAb therapy. Unfortunately, hematological toxicity was observed due to the expression of these antigens in normal HSCs [
25]. Notably, the off-target toxicity of CAR-T therapy may be more severe than traditional antibody-based drugs because of high sensitivity of this therapy and replicative capacity of T cells. Thus, an ideal target for CAR-T therapy against AML should be overexpressed in tumor cells, with minimal or no expression in the HSC compartment and other normal cells.
Human C-type lectin-like molecule-1 (CLL-1, CLEC12A, MICL, KLRL1, or DCAL-2) has been identified as a type II transmembrane glycoprotein that functions as an inhibitory receptor [
26‐
31]. The ligand of this receptor remains to be defined. The expression of CLL-1 is restricted in myeloid lineage cells, as well as in the majority of AML blasts. Leukemic stem cells (LSCs) are regarded as the primary cause of treatment failure and relapse of AML [
32]. CLL-1 is selectively present on LSCs in AML but absent in normal HSCs [
26,
33,
34], suggesting that CLL-1 is an excellent therapeutic target for AML. Indeed, mAb therapy targeting CLL-1 has been revealed its potential efficacy against AML cells and shown to be effective in reducing AML burden in xenograft model [
35,
36].
In this study, we first demonstrated that the CLL-1 antigen is an ideal target of AML for CAR-T therapy displaying restricted expression in myeloid cells and no expression in HSC. We then generated CAR-T cells by using the scFv region of the mAb against CLL-1 coupled to the costimulatory domains of CD28, 4-1BB, and the CD3-ζ chain. The CLL-1 CAR-T cells showed strong therapeutic potential against CLL-1+ AML cells in vitro and in vivo. We further observed that HSCs were not targeted because of the lack of CLL-1 expression, thus avoiding toxicity to HSCs.
Discussion
CAR-T therapy has achieved great success in a variety of malignant diseases, particularly for hematological cancer in which lineage markers of restricted expression are used as target antigens. For example, CAR-T therapy targeting the B cell lineage marker CD19 has shown striking efficacy in treating CD19
+ B cell leukemia. As a myeloid malignant disease, AML express multiple myeloid-specific lineage markers such as CD13, CD14, CD117, CD33, and CD123. Subsequently, antibodies targeting some of these antigens have been developed to treat AML [
17,
37]. However, limited therapeutic efficacy was observed as well as hematological toxicity, owing to target expression by normal HSCs. CAR-T cells targeting these same antigens provided a more effective approach to eliminate cancer cells expressing these proteins; however, they do not get around the toxicity issue for targets expressed on normal cells. Indeed, CAR-T therapy targeting tumor associated antigens like HER2 showed severe side effects on normal cells [
38,
39]. As a result, selecting tumor antigens for CAR-T therapy requires even greater cancer cell specificity that given the T cells greater capacity to recognize cells expressing low amounts of target antigen. In the current study, we confirmed that CLL-1 is selectively expressed on AML blasts but not on normal HSCs and lymphoid lineages, which ensured its safety as an antigen for CAR-T therapy. AML is generally recognized as a stem cell disease. LSCs in AML displayed intrinsic resistance to chemotherapy which results in refractoriness and relapse ultimately [
40]. Interestingly, we found CLL-1 is expressed in LSCs as well, further supporting its high potential for AML treatment. On the whole, targeting CLL-1 will not only eradicate the AML cells but also the LSCs, while sparing the normal HSCs. To this end, we generated a CLL-1 antibody and subsequently developed it into a CAR that redirected T cell specificity to CLL-1. The capabilities of CAR-T cells with respect to antigen-specific killing, proliferation, and cytokine secretion were comprehensively evaluated using in vitro killing assays. As expected, the engineered CLL-1 CAR-T cells displayed strict CLL-1 specificity both on AML cell lines and on primary patient samples. Strong anti-leukemic activity was shown in an in vivo xenograft model as well. More importantly, HSC toxicity was not observed due to its lack of CLL-1 expression. We thus demonstrated that the CLL-1 CAR-T cells are a safe therapy with high potential for AML treatment.
In a cohort of 40 AML patients, we found that CLL-1 expression was detected in most AML specimens (77.5%) with different FAB subtypes, supporting its wide application for AML treatment. Nevertheless, variable expression levels of CLL-1 were observed in different patients. In particular, we found no expression of CLL-1 on erythrocyte, indicating that CLL-1 CAR-T cells may not be appropriate for patient with M6 developed from erythrocyte, unless accompanied by abnormal myeloid hematopoietic cell proliferation. Due to the low incidence and difficulty to diagnose morphologically, specimens of AML-M7 were not included in this study. Future study of the CLL-1 expression on AML-M7 is necessary to clarify if CLL-1 CAR-T cells are appropriate for this subtype of AML. In addition, slightly reduced cytotoxicity on the primary AML specimen with low CLL-1 expression was observed in the in vitro killing assay. A possible explanation for this is that CLL-1 expression levels might affect the targeting efficacy of CAR-T cells, though we cannot rule out the variant T cell functions from different patients. Development of CARs with high affinity might be an alternative approach to resolve this potential drawback.
Given the expression of CLL-1 in the normal myeloid lineage, it is very likely that CLL-1 CAR-T cells will eliminate mature myeloid cells such as granulocyte, macrophage, and monocytes, thus interfering with the functions undertaken by myeloid lineage. We found that normal HSCs and lymphoid lineages do not express CLL-1 and verified that CLL-1 CAR-T cells did not eliminate these cells. Therefore, the adaptive immunity mainly carried out by lymphoid lineages will preserve during the treatment. In order to recover hematopoiesis, however, it is necessary to remove CLL-1 CAR-T cells after treatment. Alternative approaches such as incorporation of suicide genes or molecular switches into CAR vectors are able to resolve this disadvantage [
41,
42].
Consistent with our study, a recent publication from Tashiro et al. also reported that CLL-1 is a promising target for CAR-T cells to treat AML [
43]. We measured CLL-1 expression with other two AML classic markers CD33 and CD34 and found that CLL-1 is more frequently expressed than CD34. In our study, the expression of CLL-1 was detected in 77.5% of the AML specimens with different intensities, while Tashiro et al. observed that 85–92% of their samples were CLL-1-positive. This might be due to the difference of ethnic groups and/or the sizes of samples. We analyzed 40 Chinese patients while they only used 19 French-American-British patients. We both observed that CLL-1 was expressed in most LSCs at high levels but absent in HSCs. Interestingly, the CAR-T cells generated in our study seems be more effective. Tashiro’s report showed an incremental improvement in survival analyze using 0.6 × 10
6 CLL-1 CAR-T cells to treat irradiated mice injected with 5 × 10
4 HL-60 AML cells. In our study, 1–1.5 × 10
6 CLL-1 CAR-T cells were able to treat the disease in mice injected with 1 × 10
6 U937 AML cells. It might be resulted from the differences in the affinity of scFvs, vector designs, or distinct T cell transduction and expansion strategies.
Along with the rapid progress of CAR-T therapy, resistance to treatment has become a major barrier for its clinical application. For example, the Novartis CTL109 trial has shown a remarkable response rate of 93%. However, its 1 year complete response rate decreased to around 55%, suggesting that almost half of the patients developed resistance in a year. Target antigen loss accounted for most of the resistance in CAR-T therapy. In order to overcome that, using CAR targeting multiple antigens is a well-accepted strategy in the field [
44]. In consideration of the variant expression level of CLL-1 in AML patients, a combination with CAR targeting other antigens in AML will definitely enhance the therapeutic effect. Future testing of the co-expression of CLL-1 with other AML antigens such as CD123 and CD117 could provide more options of combinational therapy.
The combination of CLL-1 CAR-T cells with standard chemotherapy is another potential approach to enhance its therapeutic effect [
45]. High rate of tumor burden not only increases the difficulty of treatment but also prone to induce side effects like cytokine release syndrome and tumor lysis syndrome due to over-reacted immune attack. Moreover, pretreatment with chemotherapy is able to increase the graft rate of CAR-T cells into host [
46]. Further introduction of chemo-resistant genes into CLL-1 CAR vector would enable the CAR-T cells to work with chemotherapy side by side.
Overall, we demonstrated that CLL-1 is an ideal antigen of CAR-T therapy for AML. Our CLL-1 CAR-T cells displayed strong anti-leukemic effects both in vitro and in vivo, while having safe profile on normal HSCs. The CLL-1 CAR-T therapy may be a promising approach to treat AML.
Methods
Antibody generation
To prepare the CLL-1-Fc fusion protein, the extracellular domain of human CLL-1 fused with mouse Fc was synthesized (Synbio, Suzhou, China) and cloned into the lenti-vector pELNS with the EF1α promoter. The CLL-1-Fc fusion protein was purified from the culture supernatant of the CLL-1-Fc-transduced CHO cells using protein G-Sepharose column (GE Healthcare Bio-Sciences, Pittsburgh, Pennsylvania, USA) and dialyzed in PBS.
A mouse mAb against CLL-1 was generated by immunizing a BALB/c mouse (Vital River, Beijing, China) with the CLL-1-Fc fusion protein. The supernatants of hybridomas were screened using FACS to test the binding to U937 cells. The hybridoma secreting CLL-1-binding mAb was selected for further experiments.
T cell transduction
The DNA sequence of the scFv derived from the CLL-1 antibody was synthesized and cloned into the lenti-vector pELNS. CD28, 4-1BB, and CD3-ζ signaling domains were then constructed into to generate the CLL-1 CAR. Thy1.1 was constructed into as a reporting marker. The lentiviral supernatants were produced by transfecting 293 T cells and concentrated by ultra-centrifuging. The concentrated CLL-1 CAR lenti-virus was immediately stored at − 80 °C for further use.
Healthy donor-derived peripheral blood mononuclear cells (PBMCs) or CD3+ enriched T cells (Miltenyi) were expanded in vitro using anti-CD3/CD28 mAbs (Ebioscience) and recombinant human interleukin-2 (IL-2) at 20 ng/ml (Peprotech) for 48 h. The activated T cells were then transduced with lentiviral supernatants on day 3 in plates pre-coated with retronectin (Takara). After transduction, T cells were cultured with IL-2, IL-7, and IL-15. CAR expression on T cells was measured 72 h later, and the cells were isolated by immunomagnetic selection using a biotinylated-Thy1.1 antibody followed by a secondary stain with anti-biotin magnetic beads (Biolegend).
Cell lines and primary AML samples
The following AML cell lines were purchased from the ATCC: U937, HL-60, NB4, THP-1, and Molm13. K562 cell line (ATCC), negative for CLL-1, was used as a negative control. The Raji cell line (ATCC) was transduced with lentiviral vectors encoding human CLL-1 DNA to generate Raji-CLL1 cells. All the cells were cultured with RPMI 1640 media (Gibco) containing 10% fetal calf serum, penicillin, and streptomycin. The 293 T (ATCC) cells for lentiviral packaging was cultured in DMEM media (Gibco). In the bioluminescent xenograft models, U937 cells were transduced with a lentiviral firefly luciferase construct. Primary human AML specimens were acquired from the SunYat-sen University Cancer Center. This study design was approved by the SunYat-sen University Cancer Center Research Ethics Board. Written informed consent for publication of their clinical details was obtained from the patient/relative of the patient.
Flow cytometry
All the antibodies were purchased from Biolegend. Cells were washed once in 100 μL of PBS containing 2% fetal bovine serum and labeled on ice after blockade of Fc receptors. Normal bone marrows were treated with erythrocyte lysate. Primary AML cells were isolated through density gradient centrifugal assay and then remaining red blood cells were lysed. A sample was considered positive for CLL-1, CD33, or CD34 if > 20% of the sample cells expressed the antigen (compared to the control sample). For the expression analysis of CLL-1 on CD34+ progenitor/stem cell subsets, CD45-PerCP, CD34-FITC, CD38-PE/cy7, CD33-PE, and CLL1-APC mAbs were stained to identify different subpopulations. The transduction rate of CLL-1 CAR into T cells was detected by staining with a mouse Thy1.1 antibody. Flow cytometry were performed on a BD Fortesa flow cytometer, and results were analyzed using the software FlowJo7.6.5.
In vitro cytotoxicity assay
CFSE-labeled target cells were incubated at the indicated E:T ratios with CLL-1 CAR-T cells plated in triplicate wells for 24 h in RPMI 1640 media. Total cells were harvested and then labeled with 7-AAD for flow cytometry analysis.
T cell proliferation
T cells were washed and resuspended in PBS at a concentration of 1 × 106/ml and then labeled with 1 μM CFSE (Life Technologies) in PBS for 15 min at 37 °C. The CFSE-labeled T cells were washed and incubated with target cells in the absence of exogenous IL-2 for 96 h. CFSE dilution was quantified by flow cytometry.
Cytokine secretion
Effector cells and target cells were cultured at an E:T ratio of 1:1 in RPMI 1640 media for 24 h. Supernatant of culture was analyzed by 30-plex Luminex array according to the manufacturer’s instructions (Milliplex).
Xenograft animal model
All the animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of Guangdong Laboratory Animal Monitoring Institute, and the animal facility were accredited by the American Association for Accreditation of Laboratory Animal Care (AAALAC). Schematic of the used xenograft models are delineated in the relevant figures in “Results.” Leukemia engraftment was defined as > 1% human CD45+ cells in the peripheral blood by flow cytometry. Mice were sacrificed when moribund or upon the development of hind-limb paralysis. For in vivo imaging of ffLuc cells, mice were injected intraperitoneally with D-luciferin (150 mg/kg) and imaged under isoflurane anesthesia. Mice were analyzed using the Xenogen-IVIS imaging system and quantified with the Living Image software (PerkinElmer).
CD34+ cells from CB mononuclear cells were isolated from healthy donors using immunomagnetic column separation (Miltenyi). A total of 1 × 103 CD34+ CB cells were co-cultured with non-transduced T or CLL-1 CAR-T cells or media alone for 4 h at an E:T of 10:1, respectively. At the end of the 4-h co-culture, the entire cell mixture was transferred to a semisolid methylcellulose-based growth medium and plated in duplicate. After 14 days, BFU-E, CFU-GM, and CFU-GEMM colonies were enumerated.
Statistical analysis
The data are presented as mean ± standard deviation (SD). The Student t test was used to determine the statistical significance of differences between samples. Analyses were performed using SPSS version 19 statistical software, p < 0.05 was considered significant.