NKT Cells Coexpressing a GD2-Specific Chimeric Antigen Receptor and IL15 Show Enhanced In Vivo Persistence and Antitumor Activity against Neuroblastoma

Abstract

Purpose: Vα24-invariant natural killer T cells (NKT) are attractive carriers for chimeric antigen receptors (CAR) due to their inherent antitumor properties and preferential localization to tumor sites. However, limited persistence of CAR-NKTs in tumor-bearing mice is associated with tumor recurrence. Here, we evaluated whether coexpression of the NKT homeostatic cytokine IL15 with a CAR enhances the in vivo persistence and therapeutic efficacy of CAR-NKTs.

Experimental design: Human primary NKTs were ex vivo expanded and transduced with CAR constructs containing an optimized GD2-specific single-chain variable fragment and either the CD28 or 4-1BB costimulatory endodomain, each with or without IL15 (GD2.CAR or GD2.CAR.15). Constructs that mediated robust CAR-NKT cell expansion were selected for further functional evaluation in vitro and in xenogeneic mouse models of neuroblastoma.

Results: Coexpression of IL15 with either costimulatory domain increased CAR-NKT absolute numbers. However, constructs containing 4-1BB induced excessive activation-induced cell death and reduced numeric expansion of NKTs compared with respective CD28-based constructs. Further evaluation of CD28-based GD2.CAR and GD2.CAR.15 showed that coexpression of IL15 led to reduced expression levels of exhaustion markers in NKTs and increased multiround in vitro tumor cell killing. Following transfer into mice bearing neuroblastoma xenografts, GD2.CAR.15 NKTs demonstrated enhanced in vivo persistence, increased localization to tumor sites, and improved tumor control compared with GD2.CAR NKTs. Importantly, GD2.CAR.15 NKTs did not produce significant toxicity as determined by histopathologic analysis.

Conclusions: Our results informed selection of the CD28-based GD2.CAR.15 construct for clinical testing and led to initiation of a first-in-human CAR-NKT cell clinical trial (NCT03294954).

Conflict of interest statement

Conflict of interest disclosure statement: A.H., D.L., A.N.C., G.D., and L.S.M are co-inventors on pending patent applications that relate to the use of NKTs in cancer immunotherapy and have been licensed by BCM to Cell Medica, Ltd. for commercial development. Cell Medica, Ltd. provided research support for this project (to L.S.M.) via a sponsored research agreement with BCM.

X.X., W.H., L.G., M.W., J.J., B.L., E.J.D.P., J.H., G.A.B, H.N., Y.C., and B.S. declare no competing financial interests.

©2019 American Association for Cancer Research.

Figures

Figure 1.. GD2.CAR constructs containing CD28 costimulatory endodomain enable superior ex vivo expansion of CAR-NKTs.

(A) Following a 10-day primary stimulation with α-GalCer-pulsed autologous PBMCs, NKTs were re-stimulated and transduced 2 days later with the indicated CAR constructs or no construct (NT: non-transduced). Representative flow cytometry analysis of indicated GD2.CAR construct expression 3 days after transduction by gating on 1A7+ (14G2a anti-idiotype)/ Vα24-Jα18+ NKT cells (n=13). (B) Quantification of total NKT cell fold expansion, mean ± SD following 10 days of secondary expansion (n=13). NKT cell number was determined using the Cellometer Auto Cell Viability Counter with AOPI staining. *p

Figure 2.. Co-expression of IL-15 enhances survival and reduces exhaustion-associated phenotype in GD2.CAR NKTs.

(A) Expression of phosphorylated STAT5 (pSTAT5) was evaluated in NKTs expressing the GD2.28z or GD2.28z.15 CAR 7 days post-transduction following 12 hours of culture in serum- and cytokine-starved conditions. Shown is a representative histogram from one of four donors (left) and mean ± SD of MFI for all donors (n=4). **p in vitro expansion, NKTs from 14 individual donors transduced with the indicated constructs or non-transduced (NT) NKTs were evaluated for CD4 expression and (E) 11 of them were also examined for CD62L expression by flow cytometry. NS: not significant, *p < 0.05, **p < 0.01, ***p < 0.001, Student’s paired t-test. (F) NKTs transduced with the GD2.28z or GD2.28z.15 CAR or NT were evaluated for expression of the indicated exhaustion markers 7 days after transduction (n=8). *p < 0.05, **p < 0.01, ****p < 0.0001, NS: not significant, one-way ANOVA. (G) SPICE analysis of exhaustion marker expression from three of the donors shown in F. Pie charts reflect proportions of indicated NKT groups expressing indicated numbers (0-3) of exhaustion markers. Colored arcs indicate the specific combinations of exhaustion markers expressed.

Figure 3.. Co-expression of IL-15 enhances the in vitro functional fitness of GD2.CAR NKTs.

(A) Flow cytometry analysis of GD2 expression in the indicated NB cell lines. (B) The indicated luciferase-transduced NB cell lines were co-cultured with GD2.CAR NKTs for 4 or (C) 24 hours. Cytotoxicity was calculated by measuring luminescence intensity of lysed target cells with a plate reader. Shown are results from a representative of four experiments (each with a separate NKT donor) with three technical replicates. (D) GD2.CAR NKTs were co-cultured with calcein-AM-labeled M2-polarized macrophages with or without α-GalCer for 4 hours. Cytotoxicity was assessed by flow cytometry. Shown are mean ± SD from four donors. **p

Figure 4.. The GD2.28z.15 construct supports superior NKT cell in vivo persistence.

(A) Overview of in vivo NKT persistence experiments. NSG mice were injected intravenously with CHLA-255 neuroblastoma cells followed 7 days later by luciferase-labeled CAR.GD2 or control NKTs. NKTs were tracked by bioluminescence imaging every two-to-three days. (B) Bioluminescent monitoring of NT, IL-15, GD2.28z, and GD2.28z.15 NKTs injected into groups of mice (n=8/group) carrying CHLA-255 xenografts. Shown are representative results from one of three experiments (each with a separate NKT donor). (C) Quantification of bioluminescence images in B. ***p < 0.001, GD2.28z versus GD2.28z.15, Student’s unpaired t-test. (D) Quantification of NKT cells (human CD45+/NKT+) in total bone marrow cells at day 20 as determined by flow cytometry. *p < 0.05, **p < 0.01, one-way ANOVA. (E) Quantification of neuroblastoma tumor cells (human CD56+/GD2+) in total bone marrow cells at day 20 as determined by flow cytometry. Five mice in both NT and IL-15 groups were granted exceptions to delay euthanasia until day 20, but could not be imaged on day 18. *p < 0.05, **p < 0.01, NS: not significant, Student’s unpaired t-test.

Figure 5.. GD2.28z.15 mediates effective infiltration of NKTs into solid tumor tissues and expansion of CD4-negative NKTs.

(A) CHLA-255-bearing NSG mice were injected with luciferase-labeled GD2.CAR or control NKTs in a similar experimental setup as shown in Figure 4A. Thirteen days after injection of NKTs, resected livers were photographed. Shown are representative images from 5 mice per group, one of three experiments. (B) Liver metastases from “A” were fixed and processed for cryosectioning. Sections were probed with antibodies against the Vα24-Jα18 NKT TCR, neuroblastoma tumor marker tyrosine hydroxylase, and DAPI. Images were visualized and quantified by confocal microscopy. (C) Absolute numbers of NKT cells and tumor cells were counted in 10 fields per mouse using images in “B”, 5 mice per group. **p

Figure 6.. GD2.28z.15 enables superior in vivo therapeutic activity of GD2.CAR NKTs without causing significant toxicity.

(A) Schematic representation of in vivo tumor challenge experiments. NSG mice were injected intravenously with luciferase-labeled NB cell lines followed 7 days later by CAR.GD2 NKTs. Tumor growth was assessed weekly by bioluminescence imaging. (B) Weekly bioluminescent monitoring of CHLA-255 tumor growth in mice (n=7/group). (C) Kaplan-Meier curve generated from survival of animals in B. ***p < 0.001 Log-rank (Mantel-Cox). (D) Quantification of bioluminescence in B. (E) Weekly bioluminescent monitoring of CHLA-136 tumor growth. (F) Kaplan-Meier curve generated from survival of animals in E. ***p < 0.001 Log-rank (Mantel-Cox). (G) Quantification of bioluminescence in E. (H) Weekly bioluminescent monitoring of LA-N-1 tumor growth. (I) Kaplan-Meier curve generated from survival of animals in H. **p < 0.01 Log-rank (Mantel-Cox). (J) Quantification of bioluminescence in H. Shown are representative results from one of four independent experiments with the CHLA-255 model and one experiment each for the CHLA-136 and LA-N-1 models.