CAR-CIK vs. CAR-T: benchmarking novel cytokine-induced killer cells as solid tumor immunotherapy in ErbB2+ rhabdomyosarcoma

Abstract

Introduction: CAR-T cell therapy, though successful in hematologic malignancies, faces challenges in solid tumors due to limitations of autologous T cells. Cytokine-induced killer (CIK) cells can be given safely across allogeneic barriers and constitute alternative effector cells generated from healthy donors. CIK cells are a heterogenous population of predominantly T cells with a mixed natural killer (NK) phenotype and combine non-MHC-restricted cytotoxicity with potent anti-tumor capacity of the adaptive immune system. Here, we characterize and compare efficacy, phenotypic subpopulations and modes of action of CAR-CIK cells and conventional CAR-T cells from same-donor samples in ErbB2+ rhabdomyosarcoma (RMS).

Methods: To benchmark CAR-CIK against conventional CAR-T cells, effector cells were generated from same-donor samples and lentivirally transduced with a second generation CD28-CD3ζ CAR. Effector subpopulations and their dynamics upon target cell exposure were phenotypically characterized by flow cytometry. Efficacy was assessed in human ErbB2+ RMS cancer cell lines and primary patient samples in vitro and ex vivo using cytotoxicity and spheroid co-incubation assays. Modes of action were assessed by comparing cytokine secretion profiles using bead-based multiplexed flow cytometry and by liquid chromatography mass spectrometry whole cell proteomics. Finally, we used an in vivo model of RMS mimicking minimal metastatic residual disease to compare anti-tumor potency of CAR-CIK vs. CAR-T cells and to assess their target organ infiltration.

Results: In vitro assays demonstrated superior cytotoxicity of CAR-CIK cells against RMS cell lines and primary tumor samples. Long-term co-incubation with tumor spheroids led to expansion of CAR-CIK cells and enrichment of CD3+CD56+ TNK cells. CAR-CIK cell cytokine signature showed significantly increased secretion of effector molecules like interferon-γ, perforin and granulysin, and lower secretion of Th2 cytokines IL-2, IL-4 and IL-10. Whole cell proteomics showed corresponding upregulation of chemokine signaling and NK-cytotoxicity pathways in CAR-CIK cells. In NSG mice xenografted with ErbB2+ RMS, a single injection of either CAR-effector cells strongly impeded metastatic tumor development and significantly improved survival.

Conclusion: Our results demonstrate that CAR-CIK cells are at least equipotent to CAR-T cells. Combined with their favorable safety profile and allogeneic applicability, these findings position CAR-CIK cells as promising immune effectors for solid tumors.

Keywords: CAR-T; ERBB2; cytokine-induced killer cells (CIK); rhabdomyosarcoma; solid tumors.

Figure 1

(A) Cultivation and transduction of CAR-CIK and CAR-T cells with a second-generation ErbB2-specific CD28-CD3ζ CAR. All endpoint phenotypic characterizations were performed by flow cytometry on day +10 of in vitro expansion. (B) Transduction rates of CAR-CIK and CAR-T cells, measured by GFP expression. Mean ± SEM of n=11. (C) Representative flow cytometry plots showing double expression of GFP and ErbB2-CAR in CAR-CIK and CAR-T cells. (D) CD3 and CD56 expression by CAR-CIK and CAR-T cells, identifying NK (CD3-CD56+), TNK (CD3+CD56+) and T cell subpopulations (CD3+CD56-). Mean ± SEM of n=9–10. (E) Transduction rates in the NK, TNK and T cell compartments of CAR-CIK cells. Mean ± SEM of n=7. (F) CD8+ phenotype of TNK cells and T cells in CAR-CIK and CAR-T cells. Mean ± SEM of n=5–6. (G) CD4+ phenotype of TNK cells and T cells in CAR-CIK and CAR-T cells. Mean ± SEM of n=4. (H) Subpopulations of naive (CD62L+CD45RO-), central memory (CD62L+CD45RO+), effector memory (CD62L-CD45RO+) and late effector (CD62L-CD45RO-) cells in CAR-CIK and CAR-T cells. Mean ± SEM of n=6–8 experiments. P values calculated by paired t-tests. Group differences with p < 0.05 (*), p < 0.01 (**) or p < 0.001 (***) were considered statistically significant.

Figure 2

(A) Experimental scheme. Co-incubation of parental and ErbB2-CAR-CIK and CAR-T cells and T cells with luciferase-expressing target cells for 24 h at different effector to target ratios. Results are given as percentage of luciferase signal of untreated target cells. TritonX-100-treated cells served as maximum lysis control. (B–E) Cytotoxicity against B) Rh30 (n=8–10), (C) Rh41 (n=3–4), and against primary patient-derived tumor cells (D) H1 (n=6–9) and (E) H9 (n=6–7). Mean ± SEM are shown. P values calculated by Wilcoxon tests comparing CAR-CIK vs. CAR-T cells. (F) A linear regression model was used to compare the effect of CAR-CIK vs. CAR-T cells across all four target cells and all effector to target ratios with these parameters as independent predictors. Individual values and regression line with 95% confidence interval (shaded grey area) are shown. (G) Experimental scheme with representative microscopy images. (H, I) Parental and CAR effector cells were co-incubated for 6 d with established tumor spheroids generated from Rh30mCherry (H, n=5) or Rh41mCherry cells (I, n=3) and monitored via fluorescence microscopy. Difference to baseline (1 h after effector cell addition) of red fluorescence of the tumor spheroids is given in percent. Mean ± SEM are shown. P values by unpaired t-tests, comparing areas under the curves (AUC). Group differences with p < 0.05 (*), p < 0.001 (***) were considered statistically significant.

Figure 3

(A) Experimental scheme. Flow cytometric analysis of effector cells after 6 d co-incubation of parental and ErbB2-CAR-CIK and ErbB2-CAR-T cells with Rh30 spheroids. (B) Effector cell counts (log2-transformed) after 6 d ± RMS spheroids. Mean ± SEM of n=3–4; comparison of ± RMS by Poisson regression on untransformed count data; between-group effect of ± RMS was compared between CAR-CIK and CAR-T by testing for differences in slope of Poisson regression lines (likelihood ratio test). (C) Percentage of GFP+ cells after 6 d ± RMS. Mean ± SEM of n=4; analysis by Poisson regression as done in (B, D) CD3 and CD56 expression of CAR-CIK and CIK cells after 6 d ± RMS given as percentages of NK cells (CD3-CD56+), TNK cells (CD3+CD56+) and T cells (CD3+CD56-). Mean ± SEM of n=3–5, comparing CAR-CIK + RMS vs. CIK + RMS, CAR-CIK + vs. – RMS and CIK + vs. – RMS. P values by paired t-tests. (E) Relative CD8 expression of CAR-CIK and CAR-T cells after 6 d ± RMS. (F) Relative sizes of naive (CD62L+CD45RO-), central memory (CD62L+CD45RO+), effector memory (CD62L-CD45RO+) and terminal effector (CD62L-CD45RO-) subpopulations after 6 d ± RMS of CAR-CIK and CAR-T cells. Mean ± SEM of n=3–4 (E) and n=3–5 (F). P values by paired t-tests, comparing CAR-CIK + RMS vs. CAR-T + RMS, CAR-CIK + vs. – RMS and CAR-T + vs. – RMS, respectively. Group differences with p < 0.05 (*), p < 0.01 (**) or p < 0.001 (***) were considered statistically significant.

Figure 4

(A–L) After 24 h co-incubation with Rh30 cells (effector to target ratio of 10:1), supernatants of CAR-CIK or CAR-T cells were harvested and quantitatively analyzed for their cytokine signature using a bead-based immuno-assay. Mean ± SEM of n=3–4. P values were obtained by testing for differences in log2-transformed concentrations (pg/mL) by paired t-tests. sFasL, soluble Fas-Ligand. (M) Experimental scheme. Effector cells were FACS-purified and whole-cell proteomics (WCP) was performed using liquid chromatography mass spectrometry (LC-MS). TMT, tandem mass tag. (N) Principal component analysis of all conditions and replicates. Dim, dimensions. (O) Heat map showing row-scaled Z scores for all measured proteins across all conditions. (P) Volcano plot depicting log2 fold changes (FC) in measured proteins and false discovery rate (FDR)-corrected P values (q values; DEqMS/limma-moderated two-sided t-test). Significantly changing (q < 0.05) proteins with an absolute log2 FC ≥ 1 are highlighted in blue, proteins with an absolute log2 FC ≥ 2 are labeled. (Q) KEGG gene set enrichment analysis on full ranked data from (P); the top 15 significantly enriched pathways are shown. Group differences with p < 0.05 (*), p < 0.01 (**) were considered statistically significant.

Figure 5

(A) Schematic overview of the in vivo RMS xenograft model. After sublethal irradiation on day -1, luciferase-expressing Rh30 cells (Rh30/GFPfLuc) were transplanted into NSG mice by intravenous injection. After 24 h, mice were injected with CAR-effector cells all generated from the same-donor sample (CAR-CIK, n=10; CAR-T, n=12) or the parental, untransduced populations of CIK cells (n=6) or T cells (n=5). Control mice (n=7) received vehicle control. (B) Quantification of serial ventral luminescence measurements after treatment. Log2-transformed individual measurements and group-wise linear regressions are shown. P values by linear regression, comparing slope coefficients between groups. (C) Serial bioluminescence (BLI) images of two representative animals of each group on the indicated days after transplantation; luminescence displayed as log2-transformed photons/s/cm²/steradian. (D) Overall survival of xenografted NSG mice treated with vehicle, CAR-engineered or parental effector cells. P values for the differences in survival were obtained by log-rank tests.

Figure 6

Flow cytometric analysis of cells from blood, bone marrow (BM), spleen, liver and lungs of CAR-CIK- and CAR-T-treated animals at the end of the experiment. (A) Median and individual values of human CD45+ (hCD45+) cells as percentage of total analyzed cells for n=8–12. (B) Median and individual values of GFP+ cells as percentage of hCD45+ cells for n=5–7. (C) Phenotype of cytotoxic (CD8+) and (D) T-helper cells (CD8-) of CAR-CIK and CAR-T cells retrieved from indicated organs. Mean ± SEM of n=3–6. P values by unpaired t-tests. (E) TNK cell phenotype according to CD3 and CD56 expression of CAR-CIK and CIK cells prior to injection and of cells retrieved from indicated organs. Mean ± SEM of n=4–6. (F) Representative flow cytometry plot of CD56 and CD3 expression on CAR-CIK and CIK cells from the spleen. (G) Analysis of naive (CD62L+CD45RO–), central memory (CD62L+CD45RO+), effector memory (CD62L-CD45RO+) and terminal effector (CD62L-CD45RO-) cells of CAR-CIK and CAR-T treated animals retrieved from the indicated organs. Mean ± SEM of n=3–6. (H) Representative immunohistochemistry images from liver, lungs and spleen of animals treated with CAR-CIK or CAR-T cells or controls, showing hCD45+ effector cells (pink-red) among tumor cells (green) and parenchyma (all cell nuclei in blue). 20-fold magnification, scale bar indicates 100 µM. Group differences with p < 0.05 (*), p < 0.01 (**) were considered statistically significant.