Ligand-based targeting of c-kit using engineered γδ T cells as a strategy for treating acute myeloid leukemia

Abstract

The application of immunotherapies such as chimeric antigen receptor (CAR) T therapy or bi-specific T cell engager (BiTE) therapy to manage myeloid malignancies has proven more challenging than for B-cell malignancies. This is attributed to a shortage of leukemia-specific cell-surface antigens that distinguish healthy from malignant myeloid populations, and the inability to manage myeloid depletion unlike B-cell aplasia. Therefore, the development of targeted therapeutics for myeloid malignancies, such as acute myeloid leukemia (AML), requires new approaches. Herein, we developed a ligand-based CAR and secreted bi-specific T cell engager (sBite) to target c-kit using its cognate ligand, stem cell factor (SCF). c-kit is highly expressed on AML blasts and correlates with resistance to chemotherapy and poor prognosis, making it an ideal candidate for which to develop targeted therapeutics. We utilize γδ T cells as a cytotoxic alternative to αβ T cells and a transient transfection system as both a safety precaution and switch to remove alloreactive modified cells that may hinder successful transplant. Additionally, the use of γδ T cells permits its use as an allogeneic, off-the-shelf therapeutic. To this end, we show mSCF CAR- and hSCF sBite-modified γδ T cells are proficient in killing c-kit⁺ AML cell lines and sca-1⁺ murine bone marrow cells in vitro. In vivo, hSCF sBite-modified γδ T cells moderately extend survival of NSG mice engrafted with disseminated AML, but therapeutic efficacy is limited by lack of γδ T-cell homing to murine bone marrow. Together, these data demonstrate preclinical efficacy and support further investigation of SCF-based γδ T-cell therapeutics for the treatment of myeloid malignancies.

Keywords: acute myeloid leukemia (AML); c-kit (CD117); chimeric antigen receptor (CAR); gamma delta (γδ) T cells; ligand-based therapeutics; secreted bispecific T cell engager; stem cell factor (SCF).

Figure 1

c-kit expression on AML. (A) c-kit expression as fragments per kilobase per million (FPKM) on pediatric cancers from the St. Jude Cloud (pecan.stjude.cloud). Error bars represent SD. (B) c-kit mRNA expression across one normal PBMC dataset, one normal HSC dataset, and two AML datasets from R2: Genomics Analysis and Visualization Platform (https://r2.amc.nl). Error bars represent SD. Statistical analysis represents One-Way ANOVA (****p< 0.0001; ns, p > 0.05). (C) c-kit expression as nTPM on select leukemia cell lines from the Human Protein Atlas (v22.proteinatlas.org). Arrow denotes AML cell line. (D) Histograms depicting c-kit expression in healthy donor PBMCs (n = 3) and select leukemia cell lines in (C).

Figure 2

Design of novel ligand-based mSCF CAR and modification of αβ T cells. (A) Schematic of lentiviral GFP mSCF CAR DNA construct. (B) Transduction efficiency as depicted by %GFP⁺ live Jurkat T cells. Error bars represent SD. n = 3 experimental replicates. (C) Activation of live mSCF CAR- or CD19 CAR-modified Jurkat T cells as depicted by %CD69⁺ when stained with 20 ng murine c-kit-Fc chimera. Error bars represent SD. n = 1-3 experimental replicates. (D) %CD69 activation of live mSCF CAR- or CD19 CAR-modified GFP⁺ Jurkat T cells when co-cultured with 4 different leukemia cell lines at 4 different effector-to-target (E:T) ratios. Error bars represent SD. n = 2 experimental replicates. (E) %CD69 activation of live un-modified GFP- Jurkat T cells under the same conditions as (D). Error bars represent SD. n = 2 experimental replicates. (F) c-kit MFI of 4 leukemia cell lines when co-cultured with mSCF CAR- or CD19 CAR-modified Jurkat T cells under the same conditions as (D). Error bars represent SD. n = 2 experimental replicates. (G) Schematic of lentiviral mSCF CAR DNA construct without GFP marker. (H) Representative flow plots depicting mSCF CAR expression on primary T cells transduced at MOI 20 when stained with 20 ng of murine or human c-kit-Fc chimera. (I) Western blot depicting CAR expression from whole cell lysates of primary T cells transduced with mSCF CAR at MOI 20. Western blot antibody against human CD3ζ. (J) Four- and 24-hour flow cytometry cytotoxicity assays of mSCF CAR-modified primary T cells against c-kit expressing AML cell line CMK compared to mock T cell controls. Percent cytotoxicity is the sum of 7AAD⁺, Annexin V⁺, and 7AAD⁺ Annexin V⁺ cells when gated on VPD450-stained target cells only. Error bars represent SD. n = 2 experimental replicates with 1 donor.

Figure 3

mSCF CAR αβ T cells expand in vivo in the absence of AML. (A) Representative flow plots depict mSCF CAR expression of injected αβ T cells in the naive context and in the context of AML. (B) Briefly, NSG mice were injected with 5 x 10⁶ mSCF CAR-modified αβ T cells intraveneously in the absence of AML. %CAR+ αβ T cells (gated on live human CD45+ cells) within the peripheral blood (PB), spleen, and bone marrow (BM) compartments. Error bars represent SD. Statistical analysis represents Student's t test (***p<0.001; ns, p > 0.05). (C) Briefly, NSG mice recieved 100 rads of x-ray irradiation in the morning followed by IV injection of 5 x 10⁶ CMK cells in the afternoon on day -1. The following day, mice were treated with 5 x 10⁶ mock αβ T cells or mSCF CAR-modified αβ T cells (<8% CAR+) intraveneously. Peripheral blood leukocytes were collected 3 weeks after treatment to assess for leukemia engraftment. n = 3 untreated, n = 6 mock T, n = 6 mSCF CAR T. %CD3 live αβ T cells circulating within the periphery 3 weeks after treatment. Error bars represent SD. Statistical analysis represents Student's t test (*p<0.05; ns, p > 0.05). (D) %CAR+ αβ T cells (gated on live human CD3+ cells) circulating within the periphery 3 weeks after treatment. Error bars represent SD. Statistical analysis represents Student’s t test (***p<0.001; ns, p > 0.05). (E) %CD33+ live AML cells circulating within the periphery 3 weeks after treatment. Error bars represent SD. Statistical analysis represents Student's t test (****p < 0.0001; ns, p > 0.05).

Figure 4

AML and leukemia cell lines express sress antigens MICA/MICB, ULBP1, and ULBP2/5/6. mRNA expression of stress antigens MICA/MICB (A), ULBP1 (B), and ULBP2 (C) on normal PBMC and two AML datasets queried using R2: Genomics Analysis and Visualization Platform (https://r2.amc.nl). Error bars represent SD. Statistical analysis represents One-Way ANOVA (****p < 0.0001; *p < 0.05). Histograms depict stress antigen expression of MICA/MICB (D), ULBP1 (E), and ULBP2/5/6 (F) in healthy donor PBMC (n = 3) and leukemia cell lines by flow cytometry.

Figure 5

Design of novel ligand-based therapeutics and transient modification of γδ T cells. (A) Schematic of mSCF CAR construct for mRNA generation. (B) Diagram of hSCF sBite construct for mRNA generation. (C) Representative flow plots depicting mSCF CAR expression on primary γδ T cells transfected with 15 μg mRNA encoding denoted construct when stained with 20 ng of murine or human c-kit-Fc chimera. (D) Pooled transfection efficiency of primary γδ T cells modified with the mSCF CAR. n = 8 donors with n = 1-7 biological replicates. (E) Four-hour flow cytometry cytotoxicity assays of mSCF CAR- and hSCF sBite-modified primary γδ T cells against c-kit expressing AML cell lines compared to mock γδ T cell controls. Percent cytotoxicity is the sum of 7AAD⁺, Annexin V⁺, and 7AAD⁺ Annexin V⁺ cells when gated on VPD450-stained target cells only. Error bars represent SD. n = 3 donors with n = 2-7 biological replicates each. (F) Four-hour flow cytometry cytotoxicity assay of hSCF sBite γδ T cells against c-kit⁺ CMK cells compared to mock γδ T cell control. Mock γδ T cells were co-cultured with media from hSCF sBite-modified cells and c-kit⁺ CMK cells to measure sBite secretion. Error bars represent SD. n = 1 donor with n = 4 biological replicates.

Figure 6

mSCF CAR-modified γδ T cells, but not hSCF sBite-modified γδ T cells, are cytotoxic against murine bone marrow ex vivo. (A-C) Briefly, murine sca-1⁺ cells were harvested from bone marrow of C57BL/6 mice, rested for 1 day in media supplemented with mIL-3 (20 ng/mL), hIL-11 (100 ng/mL), and hFlt3 (100 ng/mL) and with or without mSCF (100 ng/mL), then co-cultured with γδ T cells for 24-hours at a 1:2 E:T ratio. Co-cultures were subject to flow cytometry analysis to assess LSK and c-kit⁺ compartments. (A) Representative flow plots depict LSK and c-kit⁺ compartments of murine sca-1⁺ cells in a 24-hour ex vivo co-culture of mock γδ T cells, mSCF CAR γδ T cells, or hSCF sBite γδ T cells supplemented with all cytokine excluding mSCF. (B) %LSK (gated on live hCD3^- hγδTCR^- cells). Error bars represent SD. n = 2-3 biological replicates. (C) Number of colonies counted from colony forming unit (CFU) assay after 24-hour co-culture. Error bars represent SD. n = 3 technical replicates. n = 2 biological replicates.

Figure 7

Persistence and toxicity of mSCF CAR γδ T cells in immunocompromised mice. (A) Schematic. NSG mice were injected with 1 × 10⁷ γδ T cells IV alone on day 0 (n = 6), 1 × 10⁷ γδ T cells IV on day 0 and two doses of 13,000 IU IL-2 IP on day 0 and day 2 (n = 6), or 1 × 10⁷ γδ T cells IV on day 0, two doses of 13,000 IU IL-2 IP on day 0 and day 2, and one dose of 70 μg/mg zoledronic acid SC on day 0 (n = 6). Peripheral blood leukocytes were collected daily beginning 24-hours after the start of treatment and assessed for the presence of human γδ T cells. Four days later, mice were sacrificed to assess for human γδ T cells within the spleen and bone marrow. (B) %hCD45+ (gated on live cells) within the peripheral blood at each timepoint. Error bars represent SD. Statistical analysis represents Student’s t test (ns, p > 0.05). (C–D) %hCD45+ (gated on live cells) within the spleen (C) and bone marrow (D) at end point. Statistical analysis represents Student’s t test (ns, p > 0.05). (E–G) %hCD3+ hγδTCR+ (gated on live cells) within the peripheral blood (E), spleen (F), and the left and right femurs (G). Error bars represent SD. Statistical analysis represents Student’s t test (ns, p > 0.05). (H–J) %c-kit+ (gated on hCD3- hγδTCR- live cells) within the peripheral blood (H), spleen (I), and left and right femurs (J). Error bars represent SD. Statistical analysis represents unpaired (H, I) or paired (J) Student’s t test (*p < 0.05; ns, p > 0.05). (K) Number of colonies counted from a CFU assay on the left and right femurs. Statistical analysis represents paired Student’s t test (ns, p > 0.05). Data point graphed is averaged from n = 3 technical replicates per mouse.

Figure 8

Treatment of hSCF sBite-modified γδ T cells only moderately prolongs survival in vivo, despite aggressive treatment regimen. (A) Experimental design. Briefly, NSG mice were pre-conditioned with 20 mg/kg busulfan IP on day -1, then injected with 5 × 10⁶ CMK cells via tail-vein injection in the morning on day 0. Beginning in the afternoon on day 0, and then once daily for the next 3 days for a total of 4 doses, 1 × 10⁷ γδ T cells were injected via tail-vein injection. Mice were subjected to bioluminescence imaging for the following 3 weeks, then followed for survival until they met endpoint. n = 8 untreated, n = 6 mock T treated, n = 6 hSCF sBite treated, n = 3 mSCF CAR treated. (B) Bioluminescence images. (C) Peripheral blood leukocytes were collected 3 weeks after the start of treatment and assessed for presence of CD33⁺ CMK cells. Error bars represent SD. Statistical analysis represents Student’s t test (ns, p > 0.05). (D) MFI of c-kit on CMK cells within the periphery. Statistical analysis represents Student’s t test (ns, p > 0.05). (E) Kaplan-Meier survival analysis. Untreated and mock γδ T treated groups were combined as a control group. Statistical analysis represents log rank (Mantel-Cox) test (ns > 0.05). P-value is shown. (F) Representative flow plots of hCD33⁺ hCD45⁺ CMK cells in the bone marrow of an untreated mouse sacrificed near end-point.