Nonviral mcDNA-mediated bispecific CAR T cells kill tumor cells in an experimental mouse model of hepatocellular carcinoma

Abstract

Background: Hepatocellular carcinoma (HCC) is one of the most common cancers worldwide and the adoptive immunotherapy of which is worth studying. CD133, a kind of cancer stem cell (CSC) antigen, together with glypican-3 (GPC3) has been proved to be highly expressed in HCC cells and both of them are used as targets to generate chimeric antigen receptor (CAR) T cells. But there are limitations like "off-target" toxicity, low transfection efficacy and weak antitumor ability in CAR T cells treatment.

Methods: The peripheral blood was acquired from healthy donors and T cells were separated by density-gradient centrifugation. We used an electroporation system to deliver anti-CD133 and anti-GPC3 single chain Fragment variable (scFv) structures as target genes into the T cells. The cell membrane was opened by the momentary electric current effect, and the target gene was delivered into the cell by non-viral minicircle DNA (mcDNA) vector. The flow cytometry and western blot assays were used to detect whether the two scFv were simultaneously transfected and the transfection efficacy of this bispecific CAR T cell generation method. We respectively detected the in vitro and in vivo tumor-suppression efficacy of CAR T cells through the CCK-8 assays and the HCC xenograft mice models. The CoG133-CAR T cells containing both CD133 and GPC3 antigen recognition sites were the effector cells. CD133-CAR T cells and GPC3-CAR T cells were defined as single-targeted control groups, normal T and mock T cells were defined as blank control groups.

Results: The mcDNA vector accommodated two target gene structures successfully transfected to generate bispecific CAR T cells. The detection methods on gene level and protein level confirmed that CoG133-CAR T cells had considerable transfection efficiency and exhibited both antigen-binding capacity of CD133 and GPC3. Compared to single-targeted CAR T cells or control T cells, CoG133-CAR T cells performed enhanced eliminated efficacy against CD133 and GPC3 double-positive HCC cell line in vitro and HCC xenograft mice in vivo. Hematoxylin and eosin (H&E) staining indicated no fatal "off-target" combination existed on CoG133-CAR T cells and major organs.

Conclusion: Our study suggests that it is with higher efficiency and more safety to prepare bispecific CAR T cells through non-viral mcDNA vectors. CoG133-CAR T cells have enhanced tumor-suppression capacity through dual antigen recognition and internal activation. It provides an innovative strategy for CAR T therapy of HCC, even solid tumors.

Keywords: Bispecific CAR T cells; Cancer immunotherapy; Cancer stem cells; Hepatocellular carcinoma; Non-viral mcDNA vector.

Fig. 1

Construction and characterization of two mcDNA vectors and the structure of CoG133-CAR T cells. A Schematic representation of CD133-CAR and GPC3-CAR construction. B The double restriction enzyme digestion for selecting CAR structures. Lanes 1–3 contain three GPC3-CAR-positive bacterial clones, and lanes 4–6 contain three CD133-CAR-positive bacterial clones. M = molecular weight marker. C Schematic diagram showing the generation of the CD133-CAR mcDNA and GPC3-CAR mcDNA. D Electrophoretic analysis to detect mcDNA. After L-arabinose induction, CD133-CAR mcDNA and GPC3-CAR mcDNA were digested into fragments of 5179 bp and 4608 bp, respectively. E Schematic illustration of the CoG133-CAR T cell structure. CD133-CAR T or GPC3-CAR T cells recognized only one tumor cell surface antigen, but CoG133-CAR T cells exerted a destructive effect on tumors containing CD133 and GPC3 antigens by recognizing one of those antigens

Fig. 2

Transfection efficiency and phenotype evaluation of CAR T cells, along with protein expression analysis. A Fluorescence microscopy images of human T lymphocytes transfected with the mcDNA or plasmid encoding GFP. The expression of GFP was gradually increased in mock T cells, CD133-CAR T cells and CoG133-CAR T cells 6, 24 and 48 h after electroporation at 400 × magnification. B Flow cytometric analysis of CD133-CAR and GPC3-CAR expression in CAR T cells. Seven days after electroporation, the transfection efficiency of mock T, CD133-CAR T, GPC3-CAR T, and CoG133-CAR T cells was 69.3%, 64.8%, 65.9%, and 59.1%, respectively. C Flow cytometric analysis showed similar expression levels of CD3, CD4 and CD8 on normal T and CoG133-CAR T cells seven days after electroporation. D Graph of the normal T and CoG133-CAR T cell phenotype analysis. Statistics are presented as the means ± SDs. n = 3 per group, n.s. not significant. E Determination of CAR protein expression after transfection by Western blot analysis. The exogenous CD3ζ protein was detected by chemiluminescence reagents to assess CAR protein expression. The molecular weights of the CD133-CAR and GPC3-CAR proteins were 53 kDa and 58 kDa, and the lysates of CoG133-CAR T cells contained both proteins

Fig. 3

Analysis of CD133 and GPC3 expression in HCC. A Flow cytometric analysis of CD133 and GPC3 antigen expression on the surface of cells from 4 human HCC cell lines. Tumor cells were stained with the PE-conjugated anti-IgG1 antibody (isotype control), PE-conjugated anti-CD133 mAb and PE-conjugated anti-GPC3 antibody. B Representative immunohistochemical staining images showing CD133 and GPC3 antigen expression in human HCC tissues from NOD/SCID xenograft mice. Scale bar = 50 µm. C Semiquantitative IOD analysis of CD133⁺ and GPC3⁺ staining in human HCC cells. Statistics are presented as the means ± SDs. n = 3 per group. # P < 0.001 vs. the SK-HEP-1 and HepG2 groups. * P < 0.001 vs. the SK-HEP-1 and PLC8024 groups. D Western blot analysis showing the expression of CD133 and GPC3 proteins in human HCC tissues extracted from NOD/SCID xenograft mice inoculated with SK-HEP-1, HepG2, PLC8024 and Huh7 cells

Fig. 4

Cytotoxicity activity, cytokine secretion and proliferation of CAR-engineered T cells in vitro. A Normal T, mock T and CAR-engineered T cells were co-incubated with human hepatocellular carcinoma cell lines for 18 h at different Effector: Target ratios. B ELISA analysis showed the secretion of IL-2, IFN-γ and TNF-α by Normal T, mock T and CAR-engineered T cells which were co-incubated with tumor cells at a 1:1 Effector: Target ratio for 24 h. C We prepared 5 × 10⁶ normal T, mock T and CAR-engineered T cells to co-cultivate with human hepatocellular carcinoma cell lines for 28 days. We measured viable T cell numbers every other day to reflect the proliferation of effector cells. All statistics were presented as mean ± SD. n = 6 per group, * P < 0.001 vs. normal T, mock T and CD133-CAR T groups; # P < 0.001 vs. normal T, mock T and GPC3-CAR T groups; § P < 0.001 vs. normal T, mock T, CD133 CAR-T and GPC3-CAR T groups

Fig. 5

In vivo inhibitory effect of CAR-engineered T cells on tumors. A Representative immunofluorescence images of four tumor tissue sections from NOD/SCID xenograft mice injected with CoG133-CAR T cells (sacrificed on day 29 after tumor formation). CD133 is labeled in green, GPC3 is labeled in red and exogenous CD3 is labeled in white; scale bar = 50 μm. B We injected effector cells into NOD/SCID xenograft mice on day 0 and day 7 (arrows marked) after the tumor volume was approximately 100 mm³ and recorded the tumor volume data. C We measured the weight of tumor tissues isolated from the sacrificed mice. D ELISA showing cytokine secretion in mouse blood serum seven days after treatment. All statistics are presented as the means ± SDs. n = 6 per group. * P < 0.001 vs. the normal T, mock T and CD133-CAR T groups; # P < 0.001 vs. the normal T, mock T and GPC3-CAR T groups; § P < 0.001 vs. the normal T, mock T, CD133-CAR T and GPC3-CAR T groups

Fig. 6

Antitumor efficacy of CoG133-CAR T cells in xenograft mice bearing double-positive tumors. A Schematic representation of the in vivo mouse bioluminescence study. B The tumor-derived bioluminescence images of mice inoculated with Huh7 cells and treated with normal T and CoG133-CAR T cells. C Statistical analysis of the bioluminescence images in the ROI at each time point. Statistics are presented as the means ± SDs. n = 5 per group. § P < 0.001 vs. the normal T, mock T, CD133-CAR T and GPC3-CAR T groups. D Survival curve showed the survival time of Huh7 tumor bearing mice treated with different effector cells. The results were evaluated by the log-rank (Mantel-Cox) test. n = 5 per group. § P < 0.001 vs. the normal T, mock T, CD133-CAR T and GPC3-CAR T groups

Fig. 7

Phenotype, persistence and safety of CoG133-CAR T cells. A Flow cytometric analysis showed the phenotype of CAR T and normal T cells from the peripheral blood of Huh7 xenograft mice on day 7. B Statistical analysis showed the proportion of CD3⁺, CD4⁺ and CD8⁺ normal T and CoG133-CAR T cells. Statistics are presented as the means ± SDs. n = 3 per group, n.s. not significant. P < 0.001 vs. the normal T group on CD4/CD8 expression. C CD133 and GPC3 coexpression was detected by flow cytometry. D Statistical analysis showed the CoG133-CAR T cells expression. Statistics are presented as the means ± SDs. n = 3 per group. E We harvested murine organ tissues from four groups of tumor xenograft mice sacrificed 28 days after the second injection of effector cells. The tissues were stained with H&E. The histopathological images of the organ tissues from all groups of tumor-bearing mice showed no differences, and images from the Huh7 group are shown as a representative. Scale bar = 100 µm