Rapidly Switchable Universal CAR-T Cells for Treatment of CD123-Positive Leukemia

Abstract

Chimeric antigen receptor T cells (CAR-T) targeting CD19 or B cell maturation antigen (BCMA) are highly effective against B cell malignancies. However, application of CAR-T to less differentially expressed targets remains a challenge due to lack of tumor-specific antigens and CAR-T controllability. CD123, a highly promising leukemia target, is expressed not only by leukemic and leukemia-initiating cells, but also by myeloid, hematopoietic progenitor, and certain endothelial cells. Thus, CAR-T lacking fine-tuned control mechanisms pose a high toxicity risk. To extend the CAR-T target landscape and widen the therapeutic window, we adapted our rapidly switchable universal CAR-T platform (UniCAR) to target CD123. UniCAR-T efficiently eradicated CD123+ leukemia in vitro and in vivo. Activation, cytolytic response, and cytokine release were strictly dependent on the presence of the CD123-specific targeting module (TM123) with comparable efficacy to CD123-specific CAR-T in vitro. We further demonstrated a pre-clinical proof of concept for the safety-switch mechanism using a hematotoxicity mouse model wherein TM123-redirected UniCAR-T showed reversible toxicity toward hematopoietic cells compared to CD123 CAR-T. In conclusion, UniCAR-T maintain full anti-leukemic efficacy, while ensuring rapid controllability to improve safety and versatility of CD123-directed immunotherapy. The safety and efficacy of UniCAR-T in combination with TM123 will now be assessed in a phase I clinical trial (ClinicalTrials.gov: NCT04230265).

Keywords: ALL; AML; CAR-T; CD123; UniCAR; adoptive cell therapy; immunotherapy.

© 2020 The Author(s).

Figures

Graphical abstract

Figure 1

Switchable UniCAR-T Triggered by TM123 Are Highly Active Against AML (A) Schematic presentation of the inducible modular UniCAR-T platform. Human T cells were genetically engineered to express either functional UniCARs, signaling-deficient UniCARs (UniCARstop), or enhanced green fluorescent protein only (vector control). (B) Upregulation of activation marker CD25 on CD4+ and CD4− UniCAR-T (2 × 104) cultivated with OCI-AML3 cells at an effector to target (e:t) ratio of 1:1 was determined by flow cytometry after 48 h (mean ± SD). (C) Lysis of OCI-AML3 cells after 48 h in the presence of TM123 normalized to controls lacking UniCAR-T (mean ± SD). (D) Detection of IFN-γ in cell culture supernatants after 48 h of cultivation with OCI-AML3 cells (mean ± SEM). (E) TM123 dose-response-dependent cytotoxic efficacy and IFN-γ release of UniCAR-T after 48 h of cultivation with OCI-AML3 cells (mean ± SD). Culture conditions in (C), (D), and (E) are as described for (B). (F) Quantification of phosphorylation of STAT5 upon signaling induction via the GM-CSF/IL-3/IL-5 receptor complex on U937-CD123 cells through binding of TM123 (lanes 3 and 4, respectively), recombinant IL-3 (lane 5), GM-CSF (lane 6), or without exogenous additives (lanes 1 and 2). Statistical significance for (B), (C), and (D) was assessed by nonparametric one-way analysis of variance (ANOVA; Kruskal-Wallis test) with a post hoc Dunn’s multiple comparison test (∗p ≤ 0.05, ∗∗p ≤ 0.01).

Figure 2

CD19- and CD123-Redirected UniCAR-T Are Highly Efficient at Targeting Primary Acute Leukemia (A) CD123 expression on patient-derived AML and B-ALL blasts. (B) Lysis of patient-derived primary AML blasts by TM123-redirected allogeneic UniCAR-T at an e:t of 1:5 after 48 h. Results were normalized to allogeneic control samples lacking any TM123 (mean ± SD). (C) TM123-redirected allogeneic UniCAR-T (1 × 104) were cultured with patient-derived B-ALL samples at an e:t of 1:5 and indicated concentrations of TM123. Living leukemic cells were determined after 24 h via flow cytometry. Cytotoxic activity was benchmarked against the tumor cells-only control. (D) TM123-redirected autologous UniCAR-T (2 × 104) were co-cultured with primary B-ALL blasts. EC50 concentrations were determined after 24 and 48 h at an e:t of 1:1. Target cell lysis was normalized to control samples without TM (mean ± SD). Statistical significance was assessed by a non-parametric Wilcoxon-Mann-Whitney test (A) or by parametric analysis of variance (ANOVA) with Dunnett’s multiple comparison test (C) (∗p ≤ 0.05, ∗∗p ≤ 0.01).

Figure 3

Pharmacokinetics of TM123 Enables Rapid Switching of UniCAR-T In order to determine half-life of TM123, NSG mice were injected intravenously (i.v.) with indicated amounts of TM123 normalized to the body weight of mice. Samples were taken at the indicated time points. The concentration of TM123 was determined via in-house ELISA. (A) Peripheral blood (PB) pharmacokinetics of 1,000 or 2,500 ng/g TM123 following i.v. bolus. (B) Internalization of TM123 into CD123-positive MOLM-13 cells was measured in vitro via flow cytometry at 37°C, while 4°C samples served as a negative control. (C) Bone marrow (BM) infiltration of 2,500 ng/g i.v. injected TM123. The BM infiltration ratio was calculated relative to corresponding plasma samples. (D) Binding of TM123 in PB or corresponding BM samples after i.v. bolus against MOLM-13 cells was determined via flow cytometry. Data represent means of individual experiments or analyzed mice (mean ± SEM).

Figure 4

Leukemia Eradication by CD123-Redirected UniCAR-T in NSG CDX and PDX Models (A) NSG mice were engrafted with 1 × 105 MOLM-13 cells 72 h prior to intravenous (i.v.) transplantation of 5 × 106 UniCAR-T. Mice were injected intraperitoneally (i.p.) with 1 μg/g body weight TM123 twice a day for 10 consecutive days. Treatment was repeated after an application-free period of 5 days. (B) Overall survival of tumor-transplanted mice. (C) Analysis of remaining leukemic cells in the peripheral blood (PB) and bone marrow (BM) of euthanized animals. (D) T cell engraftment of UniCAR-T transplanted mice in PB and BM. A common color code was used for (B), (C), and (D). (E) Patient-derived CD123-positive B-ALL blasts (1 × 106) were transplanted s.c. into the left flank of NSG mice. UniCAR-T were injected i.v. with doses of 2 × 106 and 5 × 106 at 3 (19 days) and 5 (30 days) weeks after tumor application, respectively. Tumor volumes prior to TM123 treatment showed no significant discrepancies among analyzed cohorts. Mice were i.p. injected with 2.5 μg/g body weight TM123 twice a day for 5 consecutive days. Treatment was repeated for a total of six cycles with application-free periods of 2 days in between. The highlighted region indicates the TM123-free period after tumor therapy. (F) Progression of extramedullary bulky disease was monitored during an observation period of 12 weeks utilizing a digital caliper. Data represent longitudinal median tumor growth of treated cohorts (median ± range). For analysis of tumor progression of all individual mice, please refer to Figure S3D. (G) Kaplan-Meier survival analysis in the experimental groups. Statistical significance for (B) and (G) was assessed by the Kaplan-Meier method with log rank (Mantel-Cox), non-parametric one-way analysis of variance (Kruskal-Wallis test) with Dunn’s multiple comparison test (C), a non-parametric Wilcoxon-Mann-Whitney test (D), or a two-way analysis of variance (ANOVA) (F) (∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001; ns, not significant).

Figure 5

Anti-Leukemic Efficacy of TM123-Redirected UniCAR-T Is Comparable to CD123 CAR-T (A) A CD123-specific CAR construct including a fixed binding moiety was constructed by replacing the UniCAR binding domain with the single-chain fragment variable (scFv) of the CD123-specific TM. (B) CD123 CAR-T as well as UniCAR-T (2 × 104) in the presence or absence of 5 nM TM123 were cultivated with OCI-AML3 cells at an e:t ratio of 1:1. Degranulation of CAR-T and UniCAR-T was determined by CD107a expression after 6 h of culture. (C) Cytotoxic efficacy of CD123 CAR-T (n = 3) and UniCAR-T (n = 5) in combination with 5 nM TM123 against OCI-AML3 cells (5 × 104) was determined after 24 and 48 h for the indicated range of e:t ratios and compared to control samples containing tumor cells only (gray bars). Statistical significance was assessed for indicated numbers of donors by non-parametric one-way analysis of variance (Kruskal-Wallis test) with Dunn’s multiple comparison test (B) or a non-parametric Wilcoxon-Mann-Whitney test (C) (∗p ≤ 0.05; ns, not significant).

Figure 6

Hematotoxic Effects of Constitutively Active CD123 CAR-T in Humanized Mice Are Prevented By the Switch-Off Mechanism of UniCAR-T (A) UniCAR-T or CD123 CAR-T (2 × 103) were incubated with 1:1 pre-mixed CD123high- and CD123low-expressing cells (OCI-AML3, NALM-6) at an e:t ratio of 1:10. TM123 was added once at the start of the experiment in the respective groups. Long-term cytotoxic responses against individual target cell populations were monitored for 216 h (mean ± SEM). (B) CD123 was quantified via an antigen standard curve on cell lines, patient-derived leukemia samples, and CD34-purified hematopoietic stem and progenitor cells (HSPCs) from granulocyte colony-stimulating factor (G-CSF)-mobilized healthy donors. Rapid (rr), delayed (dr), and slow response (sr) ranges of UniCAR-T with TM123 have been adapted to pre-determined anti-tumor efficacy rates against several CD123-expressing cell lines. (C) CD34+ HSPCs purified from G-CSF-mobilized healthy donors were co-cultured with allogeneic CD123 CAR-T or UniCAR-T with or without 5 nM TM123 at an e:t ratio of 1:2. Cytotoxic response against hematopoietic progenitors was determined after 48 h of incubation of CD34+ HSPCs (1 × 105) with CAR-T (5 × 104). Results were normalized to control samples containing CD34+ progenitors only. (D) Mixtures of CD34+ HSPCs (1 × 106) and CD123 CAR-T or UniCAR-T (5 × 105) combined with or without 5 nM TM123 cultured for 48 h were subsequently injected into sub-lethally irradiated NSG mice. Peripheral blood (PB) samples were taken up to 20 weeks after transplantation. (E) Engraftment analysis of CD45+ human leukocytes within PB. Significance levels were calculated relative to the control group injected with CD34+ cells only. (F) Bone marrow (BM) samples were analyzed for the presence of the lineage marker negative compartment (Lin−) and in particular for the hematopoietic stem cell-enriched cell pool via flow cytometry 23 weeks after transplantation (geometric mean ± 95% confidence interval [CI]). For analysis of human cell engraftment for all individual mice, see also Figure S5D. Statistical significance was assessed by a parametric one-way analysis of variance (ANOVA) with Tukey’s (A and B) or Dunnett’s multiple comparison test (C and F) and nonparametric ANOVA (Kruskal-Wallis test) with a post hoc Dunn’s multiple comparison test (E) (∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001).

Figure 7

Clinical Scale-Manufactured UniCAR-TCS Demonstrate High Efficacy Targeting CD123+ AML Good manufacturing practice (GMP)-equivalent manufactured clinical-scale UniCAR-TCS were generated utilizing the CliniMACS Prodigy® automated system. (A) UniCAR surface expression of clinical-scale runs was determined via flow cytometry staining with a monoclonal antibody specific for the extracellular linker domain of the UniCAR. (B) UniCAR-TCS phenotype in the final cell product from six representative batches was estimated via staining for the surface markers CD3, CD4, CD28, CD45RO, CD95, and CD197. UniCAR-T were classified in late effector (TLE), stem cell memory (TSCM), central memory/transitional memory (TCM/TTM), and effector memory (TEM) T cells. Non-defined cells (nd) fall outside of these definitions. (C) TM123 dose-response-dependent cytotoxic efficacy and cytokine release of UniCAR-TCS after 48 h of cultivation with OCI-AML3 cells. Half-maximal response (EC50) for cytokine secretion and specific lysis were calculated from response curves (mean ± SD). Data were further normalized against the maximum value observed in each dataset. (D) Donor-specific data for plots shown in (C). Statistical significance was assessed by non-parametric one-way analysis of variance (Kruskal-Wallis test) with a post hoc Dunn’s multiple comparison test (∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001; ns, not significant).